- Email: [email protected]
No evidence of increased fetal haemoglobin in an adult diabetic population
ELSEVIER
Diabetes Research and Clinical Practice 33 (1996) 37-41
No evidence of increased fetal haemoglobin in an adult diabetic population A. Di Benedetto, G. Romano,
S. Campo, E. Di Cesare, D. Cucinotta*
Department of Internal Medicine, University of Messina Medical School, Messinn, Italy Revised 31 January 1996; accepted 23 April 1996
Abstract Increased fetal haemoglobin levels, which could interfere with glycohaemoglobin assessment, have been observed in some diabetic populations, especially in insulin-treated patients. In this study, we have consecutively examined 1042 adult (aged > 18 y) diabetic subjects (102 IDDM patients, 263 insulin-treated NIDDM patients and 677 non-insulintreated NIDDM patients) and 156 sex- and age-matched control subjects. Fetal haemoglobin was assessed with a fully automated high performance liquid chromatography (HPLC) device. Its average levels were 0.19% f 0.28%1 in the control group, 0.17% f 0.23% in IDDM patients, and 0.19% + 0.25% in insulin-treated NIDDM patients; these differences were not significant. Also, the percentage of patients with fetal haemoglobin exceeding the 95th percentile of normal values (0.75%) was not different in the various subgroups. In conclusion, in our large cohort of adult diabetic patients, fetal haemoglobin levels are within the normal range, including those who have IDDM or insulin-treated NIDDM. Genetic factors could explain this difference with other reports. Keywords: Fetal haemoglobin; Insulin-dependent treatment; Glycated haemoglobin
diabetes mellitus; Non-insulin-dependent
diabetes mellitus; Insulin
1. Introduction
life by mechanisms not yet fully characterized and is almost entirely substituted by adult
(HbF), the most relevant haemoglobin (Hb) component in the fetus and in the newborn, decreases during the first years of
haemoglobin (HbA), so that HbF adult levels are usually below 0.5% of total Hb [l]. Increased HbF levels have been reported in some chronic or malignant haematological diseases [2] and during treatment with cytostatic drugs or substances able to increase HbF production [3,4]. Elevated HbF levels, which can interfere with glycated haemoglobin (HbAl c)
Fetal
haemoglobin
* Corresponding author, Dipartimento di Medicina Interna, Policlinico Universitario, 98100 Messina, Italy. Tel: + 39 90 2212390; fax + 39 90 2935162.
016%8227/96]$15.000 1996 Elsevier Science Ireland Ltd. All rights reserved Pff SOl68-8227(96)01271-5
38
A.D. Benedrtto et al. / Diabetes Research and Clinical Practice 33 (1996) 37-41
Table 1 Clinical data, HbAlc and HbF levels (mean f SD) in the control group and in the diabetic patients, both as a whole and divided into 3 subgroups, according to the type of diabetes and insulin treatment Sex, M/F
Control subjects (n = 156) Diabetic patients (n = 1042) IDDM patients (n = 102) Insulin-treated NIDDM patients (n = 263) Non insulin-treated NIDDM patients (n = 677)
75Bl
Age, years (range)
57 + 17 (20-88)
5021540
58 )
15 (19991)
49153
31.8 i
9.8 (19-65)
Known diabetes duration, years (range)
HbAlc, (range)
-
5.1 + 0.3 (4.2-6.1) 8.3 k 1.8” (4.8-14.8) 8.7 + 2.1 (5514.8) 9.0 + 1.6 (5.2-14.2)
12 + 9 (O-48) 11.3 *
8.6 (O-41)
124/139
63.9 + 9.9 (29-85)
16.7 + 8.9’ (O-44)
3291348
60.8 k
10.3 + 8.1 (O-48)
10.8” (22291)
“P < 0.0001 vs. control subjects; ‘P < 0.0001 vs. IDDM patients;‘P < 0.0001 vs. IDDM and insulin treated NIDDM
assessment, have also been observed in insulin-dependent diabetes mellitus (IDDM), both during childhood [5] and adult life [6], and recently in insulin-dependent mellitus non diabetes (NIDDM), especially in patients treated with insulin [7,8]. The basis for this study was our observation that, in contrast with the above-mentioned reports, in patients attending our clinic and routinely performing HbAlc assessment, HbF levels were not increased. In our laboratory, HbAlc was measured by a high performance liquid chromatography (HPLC) method which also measures HbF and has been validated against more specific HbF assays [6]. To verify this observation, we decided to systematically assess HbF levels in a large cohort of adult diabetic patients.
2. Subjects and methods 1042 consecutive diabetic subjects attending our outpatient clinic for routine examination were enrolled in this study. Exclusion criteria were age < 18 y and evidence of haematological diseases at routine clinical and laboratory (Hb levels, blood cells count) examination. Patients with types of
‘%
7.9 + I.74 (4.8-14.1)
patients; ‘P < 0.001 vs. IDDM patients
HbF, % (range)
0.19 * 0.28 (O-2.2) 0.17 + 0.25 (O-2.3) 0.17 k 0.23 (O-1.9) 0.19 k 0.25 (O-2.1)
0.17 2 0.26 (O-2.3)
and non-insulin-treated
NIDDM
diabetes other than IDDM or NIDDM and those who couldn’t be classified were also excluded. The control group included 156 normal subjects, matched for sex and age, selected among our hospital blood donors or elderly people living in a community. According to World Health Organisation (WHO) criteria [9], 102 patients had IDDM and 940 had NIDDM. In this last group, 263 were treated with insulin (alone or combined with oral hypoglycaemic agents); the remaining 677 were either on diet alone (125 patients) or oral agents (sulfonylureas in 173 patients, biguanides in 81, sulfonylureas + biguanides in the remaining 298). All patients and controls were of Italian descent and were born and lived in Sicily and Calabria regions (South Italy). Their clinical data are reported in Table 1. Each subject gave his informed consent to participate in the study, which was approved by the Ethical Committee of our University Hospital. Venous blood samples for HbAlc and HbF assessment were collected during a morning routine visit in ethylenediamine tetraacetic acid (EDTA)-vacutainer tubes, stored at 4°C and analyzed within the next two days. HbAlc and HbF were simultaneously determined by HPLC with a
A.D. Benedetto et al. 1 Diabetes Research and Clinical Practice 33 (1996) 37-41 Table 2 Clinical data, HbAlc
and HbF levels (mean k SD) in subjects with HbF > mean + 2 SD of the control group values Sex, M/F
Control subjects (n = 7) IDDM patients (n = 3) Insulin-treated NIDDM patients (n = 9) Non insulin-treated NIDDM patients (n = 28)
Age, years (range)
Known diabetes duration, years (range)
HbAlc,
314
48.7 + 13.2 (21-78)
-
5.2 k 0.2 (4.445.7)
2/l
27.6 + 6.5 (21-34)
15.6 & 5.8(9-20)
7.7 f
2.4 (5.4410.2)
8.7 i
1.7 (6.3311.5)
514
13/15
68 f
10.8 (47-80)
15.7 & 8,X(3328)
59.4 k
10.8 (33-82)
8.4 + 5.3(1&21)
fully automated analyser (Diamat Bio-Rad, Milan, Italy) according to previously described methods [6]. Intra- and inter-assay coefficients of variation were, respectively, 3.5% and 5.5% for HbF at 0.5% concentration and 0.8% and 1.8% for HbAlc at 5% concentration. Our laboratory participates in a quality control program organized by the HPLC manufacturer (Bio-Rad Haemoglobin Proficiency Survey). 2.1. Stutistical
39
analysis
Data are presented as means + SD. Statistical analysis for inter-group comparisons was performed by non-parametric analysis of variance (ANOVA). Discrete variables were compared by chi-square test. Bivariate interdependencies were assessed by linear regression analysis. A P-value less than 0.05 was considered significant.
3. Results The average HbF level in the whole adult diabetic population was not different from that observed in the control group. As shown in Table 1, the two study groups were well matched for age and sex, the only obvious difference being the
% (range)
7.2 + 1.1 (5.5510.5)
HbF, ‘%, @we) 1.06 & 0.42 (0.8-2.2) 1.23 k 0.5 (0.991.9) 1.04 k 0.42 (0.8-2.1) 1.08 * 0.34 (0.8-2.3)
greatly higher HbAlc level in the diabetic population. When diabetic patients were divided in three subgroups according to the type of diabetes and insulin treatment (IDDM patients, insulin-treated non-insulin-treated NIDDM patients and NIDDM patients), again no differences were observed among these subgroups concerning average HbF levels. As expected, NIDDM patients were older than IDDM and insulin-treated NIDDM subjects had a longer diabetes duration than the non-insulin-treated ones. This last group also displayed the lowest HbAlc values (Table 1); however, in all groups no correlation was observed between HbF and HbAlc levels. Also, no differences in HbF levels were observed in non-insulintreated NIDDM patients according to the kind of therapy (diet, sulfonylureas, biguanides). A HbF value greater than 2 SD over the mean of the control group, i.e., > 0.75%, was present in 7/l 56 (4.5”/0) normal subjects, in 3/102 (2.9%) IDDM patients, in 9/263 (3.4%) insulin-treated NIDDM and in 28/677 (4.1%) non-insulin-treated NIDDM patients. These differences were not significant. Clinical data and haemoglobin levels of these subjects are depicted in Table 2. No significant differences were observed between these patients and the remaining others of the same group concerning sex, age, diabetes duration and HbAlc levels.
40
A.D. Benedetto et al. /Diabetes
Research and Clinical Practice 33 (1996) 37-41
4. Discussion Increased HbF levels have been recently reported not only in IDDM diabetic children, but also in adult IDDM and in NIDDM patients, especially in those who were treated with insulin [5,6,8]. Consequently, it has been suggested that insulin therapy can delay the switch from fetal to adult haemoglobin synthesis, or reactivate the yglobin gene, which is responsible for HbF production. Other factors, such as those associated with poor metabolic control, have been also suspected as responsible for this increase [6], since high HbF was observed in infants of diabetic mothers [lo] and in patients with abnormalities of fatty acids metabolism [ll]. The phenomenon is of clinical importance, because HbF can interfere with some methods for HbAlc determination [12-141. However, in our large adult diabetic population, we were not able to observe any increase of HbF levels. In the whole diabetic group the mean HbF value was very similar to that recorded in an age- and sex-matched control group and no differences were observed even when IDDM or insulin-treated NIDDM patients were separately examined. Also, the percentage of subjects with HbF exceeding the 95th percentile of normal values was not significantly different in the various subgroups and the lowest number of patients with elevated HbF was surprisingly observed among IDDM. Thus, our data confirm a recent report where no elevation of HbF levels was found in a small group of diabetic patients [15]. It is difficult to explain this discrepancy with previous reports: our reference population was accurately matched with the diabetic one for age and sex and the method we have used for HbF determination has been largely validated and is internally and externally controlled. Different to others, we have chosen the 95th percentile of normal values (HbF > 0.75%) to better identify those subjects with strongly increased HbF. This cut-off cannot have falsely reduced our number of positive patients, since it is lower than in other similar studies [8]. Moreover, even when our data were re-evaluated according to a lower HbF cutoff (0.5%) again no differences were observed among the different groups (data not shown).
One might speculate that genetic factors, linked to the selected ethnic characters of our study subjects (all of south-Italian descent and Mediterranean origin) can explain these results: it has been suggested that HbF values may be genetically determined [8] and no information concerning HbF levels before the onset of diabetes is available in studies where diabetic patients have increased HbF. It is also possible that these genetic factors exert their influence through quantitative modification of other haemoglobins such as HbA2, whose levels are often elevated in countries where thalassemia is widely present, such as our country [16], but are not commonly detected by HPLC assay for HbAlc [17]. This could lead to underestimation of percentage HbF levels in our study groups; however, patients with clinical or laboratory evidence of thalassemia were clearly excluded from our study. In conclusion, in a large cohort of adult diabetic patients with Mediterranean descent, HbF levels are normal, even in those patients who have IDDM or insulin-treated NLDDM; genetic factors are likely to be responsible for the striking difference with the increased prevalence reported in other diabetic populations of different ethnic origin.
References 111Phillips, H.M., Holland, B.M., Jones, S.C., Abdel-Moiz, A.L., Turner, T.L. and Wardrop, C.A.J. (1988) Definitive estimate of rate of haemoglobin switching: measurehaemoglobin F in neonatal ments of percent reticulocytes. Ped. Res. 23, 595-597. 121Wood, W.G. (1993) Increased HbF in adult life. Baillier’s Clin. Hemat. 6, 1777213. [31 Samid, D., Yeh, A. and Prasanna P. (1992) Induction of erytroid differentiation and fetal hemoglobin production in human leukemic cell treated with phenylacetate. Blood 80, 1576- 1581. 141 Dover, G.J., Brusilow, S. and Samid, D. (1992) Increased fetal haemoglobin in patients receving sodium 4-phenylbutyrate. N. Engl. J. Med. 327, 568-570. 151Mullis, P., Schuler, S. and Zuppinger, K. (1989) Increased prevalence of fetal haemoglobin in Type 1 (insulin-dependent) diabetes mellitus. Diabetologia 32, 2277230. 161Diem, P., Mullis, P., Hirt, A., Schuler, J.J., Burgi, W., Zuppinger, K.A. and Teuscher, A. (1993) Fetal
A.D. Benedetto et al. / Diabetes Research and Clinical Practice 33 (1996) 37-41
[7]
[8]
[9]
[IO]
[l I]
[12]
haemoglobin levels in adult Type 1 (insulin-dependent) diabetic patients. Diabetologia 36, 129- 132. Kilpatrick, ES., Small, M., Rumley, A. and Dominiczak, M.H. (1993) Increased fetal hemoglobin in insulin-treated diabetes mellitus contributes to the imprecision of glycohemoglobin measurements. Clin. Chem. 39, 833-835. Koskinen, L.K., Koivula, T.A. and Lahtela, S.T. (1994) Fetal hemoglobin in diabetic patients. Diabetes Care 17, 828-831. World Health Organization Expert Committee (1980) Second report on diabetes mellitus. Technical Report Series, No. 646. Perrine, S.P., Greene, M.F. and Failer, D.V. (1985) Delay in the fetal globin switch in infants of diabetic mothers. N. Engl. J. Med. 312, 334-334. Little, J.A., Dempsey, N.J., Tuchman, M. and Cinder, G.D. (1995) Metabolic persistence of fetal hemoglobin. Blood 85, 1712-1718. Yatscoff, R.W., Tevaarwerk, J.M., Clarson, CL. and Warnoch, L.M. (1983) Interference of fetal hemoglobin
[I31
[14]
[15]
[16]
[17]
41
and labile glycosylated hemoglobin with measurements of glycosylated hemoglobin. Clin. Chem. 29, 543-545. Thivolet, C., Goujon, R., Vassy, V., Revol, A. and Toumiaire, J. (1991) Interference of elevated fetal hemoglobin on HbAlc measurements in adult type I diabetic patients, Diabetes Care 14, 1108. Cox, T., Hess, P.P., Thompson, G.D. and Levinson, S.S. (1993) Interference with glycated hemoglobin by hemoglobin F may be greater than is generally assumed. Am. J. Clin. Pathol. 99, 137-141. Kopp, H.P., Krugluger, W., Festa, A. and Schernthaner, G. (1995) The clinical relevance of fetal hemoglobin in IDDM and NIDDM patients. Diabetes Care 18, 883. Schiliro, G., Comisi, F.F., Testa, R., Di Benedetto, S.P., Samperi, P. and Marino, S. (1990) Haematological findings in 375 Sicilians with HbS trait. Haematologica 75, 113-116. Piras, G., Carluccio, A., Coe, A., Domkel, I., Miravalles, E. and Nanij, A. (1993) Monitoring of HbAlc in patients with thalassemia and sickle cell disease. Klin. Lab. 39, 1033-1037.
Diabetes Research and Clinical Practice 33 (1996) 37-41
No evidence of increased fetal haemoglobin in an adult diabetic population A. Di Benedetto, G. Romano,
S. Campo, E. Di Cesare, D. Cucinotta*
Department of Internal Medicine, University of Messina Medical School, Messinn, Italy Revised 31 January 1996; accepted 23 April 1996
Abstract Increased fetal haemoglobin levels, which could interfere with glycohaemoglobin assessment, have been observed in some diabetic populations, especially in insulin-treated patients. In this study, we have consecutively examined 1042 adult (aged > 18 y) diabetic subjects (102 IDDM patients, 263 insulin-treated NIDDM patients and 677 non-insulintreated NIDDM patients) and 156 sex- and age-matched control subjects. Fetal haemoglobin was assessed with a fully automated high performance liquid chromatography (HPLC) device. Its average levels were 0.19% f 0.28%1 in the control group, 0.17% f 0.23% in IDDM patients, and 0.19% + 0.25% in insulin-treated NIDDM patients; these differences were not significant. Also, the percentage of patients with fetal haemoglobin exceeding the 95th percentile of normal values (0.75%) was not different in the various subgroups. In conclusion, in our large cohort of adult diabetic patients, fetal haemoglobin levels are within the normal range, including those who have IDDM or insulin-treated NIDDM. Genetic factors could explain this difference with other reports. Keywords: Fetal haemoglobin; Insulin-dependent treatment; Glycated haemoglobin
diabetes mellitus; Non-insulin-dependent
diabetes mellitus; Insulin
1. Introduction
life by mechanisms not yet fully characterized and is almost entirely substituted by adult
(HbF), the most relevant haemoglobin (Hb) component in the fetus and in the newborn, decreases during the first years of
haemoglobin (HbA), so that HbF adult levels are usually below 0.5% of total Hb [l]. Increased HbF levels have been reported in some chronic or malignant haematological diseases [2] and during treatment with cytostatic drugs or substances able to increase HbF production [3,4]. Elevated HbF levels, which can interfere with glycated haemoglobin (HbAl c)
Fetal
haemoglobin
* Corresponding author, Dipartimento di Medicina Interna, Policlinico Universitario, 98100 Messina, Italy. Tel: + 39 90 2212390; fax + 39 90 2935162.
016%8227/96]$15.000 1996 Elsevier Science Ireland Ltd. All rights reserved Pff SOl68-8227(96)01271-5
38
A.D. Benedrtto et al. / Diabetes Research and Clinical Practice 33 (1996) 37-41
Table 1 Clinical data, HbAlc and HbF levels (mean f SD) in the control group and in the diabetic patients, both as a whole and divided into 3 subgroups, according to the type of diabetes and insulin treatment Sex, M/F
Control subjects (n = 156) Diabetic patients (n = 1042) IDDM patients (n = 102) Insulin-treated NIDDM patients (n = 263) Non insulin-treated NIDDM patients (n = 677)
75Bl
Age, years (range)
57 + 17 (20-88)
5021540
58 )
15 (19991)
49153
31.8 i
9.8 (19-65)
Known diabetes duration, years (range)
HbAlc, (range)
-
5.1 + 0.3 (4.2-6.1) 8.3 k 1.8” (4.8-14.8) 8.7 + 2.1 (5514.8) 9.0 + 1.6 (5.2-14.2)
12 + 9 (O-48) 11.3 *
8.6 (O-41)
124/139
63.9 + 9.9 (29-85)
16.7 + 8.9’ (O-44)
3291348
60.8 k
10.3 + 8.1 (O-48)
10.8” (22291)
“P < 0.0001 vs. control subjects; ‘P < 0.0001 vs. IDDM patients;‘P < 0.0001 vs. IDDM and insulin treated NIDDM
assessment, have also been observed in insulin-dependent diabetes mellitus (IDDM), both during childhood [5] and adult life [6], and recently in insulin-dependent mellitus non diabetes (NIDDM), especially in patients treated with insulin [7,8]. The basis for this study was our observation that, in contrast with the above-mentioned reports, in patients attending our clinic and routinely performing HbAlc assessment, HbF levels were not increased. In our laboratory, HbAlc was measured by a high performance liquid chromatography (HPLC) method which also measures HbF and has been validated against more specific HbF assays [6]. To verify this observation, we decided to systematically assess HbF levels in a large cohort of adult diabetic patients.
2. Subjects and methods 1042 consecutive diabetic subjects attending our outpatient clinic for routine examination were enrolled in this study. Exclusion criteria were age < 18 y and evidence of haematological diseases at routine clinical and laboratory (Hb levels, blood cells count) examination. Patients with types of
‘%
7.9 + I.74 (4.8-14.1)
patients; ‘P < 0.001 vs. IDDM patients
HbF, % (range)
0.19 * 0.28 (O-2.2) 0.17 + 0.25 (O-2.3) 0.17 k 0.23 (O-1.9) 0.19 k 0.25 (O-2.1)
0.17 2 0.26 (O-2.3)
and non-insulin-treated
NIDDM
diabetes other than IDDM or NIDDM and those who couldn’t be classified were also excluded. The control group included 156 normal subjects, matched for sex and age, selected among our hospital blood donors or elderly people living in a community. According to World Health Organisation (WHO) criteria [9], 102 patients had IDDM and 940 had NIDDM. In this last group, 263 were treated with insulin (alone or combined with oral hypoglycaemic agents); the remaining 677 were either on diet alone (125 patients) or oral agents (sulfonylureas in 173 patients, biguanides in 81, sulfonylureas + biguanides in the remaining 298). All patients and controls were of Italian descent and were born and lived in Sicily and Calabria regions (South Italy). Their clinical data are reported in Table 1. Each subject gave his informed consent to participate in the study, which was approved by the Ethical Committee of our University Hospital. Venous blood samples for HbAlc and HbF assessment were collected during a morning routine visit in ethylenediamine tetraacetic acid (EDTA)-vacutainer tubes, stored at 4°C and analyzed within the next two days. HbAlc and HbF were simultaneously determined by HPLC with a
A.D. Benedetto et al. 1 Diabetes Research and Clinical Practice 33 (1996) 37-41 Table 2 Clinical data, HbAlc
and HbF levels (mean k SD) in subjects with HbF > mean + 2 SD of the control group values Sex, M/F
Control subjects (n = 7) IDDM patients (n = 3) Insulin-treated NIDDM patients (n = 9) Non insulin-treated NIDDM patients (n = 28)
Age, years (range)
Known diabetes duration, years (range)
HbAlc,
314
48.7 + 13.2 (21-78)
-
5.2 k 0.2 (4.445.7)
2/l
27.6 + 6.5 (21-34)
15.6 & 5.8(9-20)
7.7 f
2.4 (5.4410.2)
8.7 i
1.7 (6.3311.5)
514
13/15
68 f
10.8 (47-80)
15.7 & 8,X(3328)
59.4 k
10.8 (33-82)
8.4 + 5.3(1&21)
fully automated analyser (Diamat Bio-Rad, Milan, Italy) according to previously described methods [6]. Intra- and inter-assay coefficients of variation were, respectively, 3.5% and 5.5% for HbF at 0.5% concentration and 0.8% and 1.8% for HbAlc at 5% concentration. Our laboratory participates in a quality control program organized by the HPLC manufacturer (Bio-Rad Haemoglobin Proficiency Survey). 2.1. Stutistical
39
analysis
Data are presented as means + SD. Statistical analysis for inter-group comparisons was performed by non-parametric analysis of variance (ANOVA). Discrete variables were compared by chi-square test. Bivariate interdependencies were assessed by linear regression analysis. A P-value less than 0.05 was considered significant.
3. Results The average HbF level in the whole adult diabetic population was not different from that observed in the control group. As shown in Table 1, the two study groups were well matched for age and sex, the only obvious difference being the
% (range)
7.2 + 1.1 (5.5510.5)
HbF, ‘%, @we) 1.06 & 0.42 (0.8-2.2) 1.23 k 0.5 (0.991.9) 1.04 k 0.42 (0.8-2.1) 1.08 * 0.34 (0.8-2.3)
greatly higher HbAlc level in the diabetic population. When diabetic patients were divided in three subgroups according to the type of diabetes and insulin treatment (IDDM patients, insulin-treated non-insulin-treated NIDDM patients and NIDDM patients), again no differences were observed among these subgroups concerning average HbF levels. As expected, NIDDM patients were older than IDDM and insulin-treated NIDDM subjects had a longer diabetes duration than the non-insulin-treated ones. This last group also displayed the lowest HbAlc values (Table 1); however, in all groups no correlation was observed between HbF and HbAlc levels. Also, no differences in HbF levels were observed in non-insulintreated NIDDM patients according to the kind of therapy (diet, sulfonylureas, biguanides). A HbF value greater than 2 SD over the mean of the control group, i.e., > 0.75%, was present in 7/l 56 (4.5”/0) normal subjects, in 3/102 (2.9%) IDDM patients, in 9/263 (3.4%) insulin-treated NIDDM and in 28/677 (4.1%) non-insulin-treated NIDDM patients. These differences were not significant. Clinical data and haemoglobin levels of these subjects are depicted in Table 2. No significant differences were observed between these patients and the remaining others of the same group concerning sex, age, diabetes duration and HbAlc levels.
40
A.D. Benedetto et al. /Diabetes
Research and Clinical Practice 33 (1996) 37-41
4. Discussion Increased HbF levels have been recently reported not only in IDDM diabetic children, but also in adult IDDM and in NIDDM patients, especially in those who were treated with insulin [5,6,8]. Consequently, it has been suggested that insulin therapy can delay the switch from fetal to adult haemoglobin synthesis, or reactivate the yglobin gene, which is responsible for HbF production. Other factors, such as those associated with poor metabolic control, have been also suspected as responsible for this increase [6], since high HbF was observed in infants of diabetic mothers [lo] and in patients with abnormalities of fatty acids metabolism [ll]. The phenomenon is of clinical importance, because HbF can interfere with some methods for HbAlc determination [12-141. However, in our large adult diabetic population, we were not able to observe any increase of HbF levels. In the whole diabetic group the mean HbF value was very similar to that recorded in an age- and sex-matched control group and no differences were observed even when IDDM or insulin-treated NIDDM patients were separately examined. Also, the percentage of subjects with HbF exceeding the 95th percentile of normal values was not significantly different in the various subgroups and the lowest number of patients with elevated HbF was surprisingly observed among IDDM. Thus, our data confirm a recent report where no elevation of HbF levels was found in a small group of diabetic patients [15]. It is difficult to explain this discrepancy with previous reports: our reference population was accurately matched with the diabetic one for age and sex and the method we have used for HbF determination has been largely validated and is internally and externally controlled. Different to others, we have chosen the 95th percentile of normal values (HbF > 0.75%) to better identify those subjects with strongly increased HbF. This cut-off cannot have falsely reduced our number of positive patients, since it is lower than in other similar studies [8]. Moreover, even when our data were re-evaluated according to a lower HbF cutoff (0.5%) again no differences were observed among the different groups (data not shown).
One might speculate that genetic factors, linked to the selected ethnic characters of our study subjects (all of south-Italian descent and Mediterranean origin) can explain these results: it has been suggested that HbF values may be genetically determined [8] and no information concerning HbF levels before the onset of diabetes is available in studies where diabetic patients have increased HbF. It is also possible that these genetic factors exert their influence through quantitative modification of other haemoglobins such as HbA2, whose levels are often elevated in countries where thalassemia is widely present, such as our country [16], but are not commonly detected by HPLC assay for HbAlc [17]. This could lead to underestimation of percentage HbF levels in our study groups; however, patients with clinical or laboratory evidence of thalassemia were clearly excluded from our study. In conclusion, in a large cohort of adult diabetic patients with Mediterranean descent, HbF levels are normal, even in those patients who have IDDM or insulin-treated NLDDM; genetic factors are likely to be responsible for the striking difference with the increased prevalence reported in other diabetic populations of different ethnic origin.
References 111Phillips, H.M., Holland, B.M., Jones, S.C., Abdel-Moiz, A.L., Turner, T.L. and Wardrop, C.A.J. (1988) Definitive estimate of rate of haemoglobin switching: measurehaemoglobin F in neonatal ments of percent reticulocytes. Ped. Res. 23, 595-597. 121Wood, W.G. (1993) Increased HbF in adult life. Baillier’s Clin. Hemat. 6, 1777213. [31 Samid, D., Yeh, A. and Prasanna P. (1992) Induction of erytroid differentiation and fetal hemoglobin production in human leukemic cell treated with phenylacetate. Blood 80, 1576- 1581. 141 Dover, G.J., Brusilow, S. and Samid, D. (1992) Increased fetal haemoglobin in patients receving sodium 4-phenylbutyrate. N. Engl. J. Med. 327, 568-570. 151Mullis, P., Schuler, S. and Zuppinger, K. (1989) Increased prevalence of fetal haemoglobin in Type 1 (insulin-dependent) diabetes mellitus. Diabetologia 32, 2277230. 161Diem, P., Mullis, P., Hirt, A., Schuler, J.J., Burgi, W., Zuppinger, K.A. and Teuscher, A. (1993) Fetal
A.D. Benedetto et al. / Diabetes Research and Clinical Practice 33 (1996) 37-41
[7]
[8]
[9]
[IO]
[l I]
[12]
haemoglobin levels in adult Type 1 (insulin-dependent) diabetic patients. Diabetologia 36, 129- 132. Kilpatrick, ES., Small, M., Rumley, A. and Dominiczak, M.H. (1993) Increased fetal hemoglobin in insulin-treated diabetes mellitus contributes to the imprecision of glycohemoglobin measurements. Clin. Chem. 39, 833-835. Koskinen, L.K., Koivula, T.A. and Lahtela, S.T. (1994) Fetal hemoglobin in diabetic patients. Diabetes Care 17, 828-831. World Health Organization Expert Committee (1980) Second report on diabetes mellitus. Technical Report Series, No. 646. Perrine, S.P., Greene, M.F. and Failer, D.V. (1985) Delay in the fetal globin switch in infants of diabetic mothers. N. Engl. J. Med. 312, 334-334. Little, J.A., Dempsey, N.J., Tuchman, M. and Cinder, G.D. (1995) Metabolic persistence of fetal hemoglobin. Blood 85, 1712-1718. Yatscoff, R.W., Tevaarwerk, J.M., Clarson, CL. and Warnoch, L.M. (1983) Interference of fetal hemoglobin
[I31
[14]
[15]
[16]
[17]
41
and labile glycosylated hemoglobin with measurements of glycosylated hemoglobin. Clin. Chem. 29, 543-545. Thivolet, C., Goujon, R., Vassy, V., Revol, A. and Toumiaire, J. (1991) Interference of elevated fetal hemoglobin on HbAlc measurements in adult type I diabetic patients, Diabetes Care 14, 1108. Cox, T., Hess, P.P., Thompson, G.D. and Levinson, S.S. (1993) Interference with glycated hemoglobin by hemoglobin F may be greater than is generally assumed. Am. J. Clin. Pathol. 99, 137-141. Kopp, H.P., Krugluger, W., Festa, A. and Schernthaner, G. (1995) The clinical relevance of fetal hemoglobin in IDDM and NIDDM patients. Diabetes Care 18, 883. Schiliro, G., Comisi, F.F., Testa, R., Di Benedetto, S.P., Samperi, P. and Marino, S. (1990) Haematological findings in 375 Sicilians with HbS trait. Haematologica 75, 113-116. Piras, G., Carluccio, A., Coe, A., Domkel, I., Miravalles, E. and Nanij, A. (1993) Monitoring of HbAlc in patients with thalassemia and sickle cell disease. Klin. Lab. 39, 1033-1037.