ORIGINAL ARTICLE
Study of Insulin Resistance in Patients With
β Thalassemia Major and Validity of Triglyceride
Glucose (TYG) Index
Arif M. Ansari, MD,* Kamalakshi G. Bhat, MD,* Smitha S. Dsa, MD,*
Soundarya Mahalingam, MD,* and Nitin Joseph, MD†
Summary: Complications like impaired glucose tolerance and diabetes
mellitus due to iron overload need early identification in thalassemia.
We studied the proportion of insulin resistance in thalassemia major
patients on chronic transfusion, identified insulin resistance using
homeostasis model assessment of insulin resistance (HOMA-IR) and
triglyceride glucose (TYG) index, compared them and validated TYG
index. In total, 73 thalassemia patients on regular transfusion for
3 years with serum ferritin > 1500 ng/mL were studied. Serum ferritin,
fasting blood glucose, triglycerides, and insulin levels were measured,
HOMA-IR, and TYG index calculated and analyzed. Mean fasting
glucose, triglyceride, and serum insulin values were 104 mg/dL,
164.18 mg/dL, and 19.6 m IU/mL, respectively. Mean serum ferritin
was 5156 ng/mL. Insulin resistance was prevalent in one third of
thalassemia patients and showed increase with age and serum ferritin.
Insulin resistance by HOMA-IR was 32% as against 16% by TYG
index with a cut-off value of 4.3. Using receiver operating char-
ecteristic curve analysis, it was found that, by lowering the value of
TYG index to 4.0215, sensitivity improved to 78.3% (from 39.13%)
with specificity of 70%. Hence, we recommend a newer lower cut-off
value of 4.0215 for TYG index for better sensitivity and specificity in
identifying insulin resistance.
Key Words: β thalassemia, insulin resistance, HOMA-IR, TYG
index
(J Pediatr Hematol Oncol 2017;00:000–000)
β thalassemia is an inherited disorder characterized by
deficiency in the production of β globin chains resulting
in ineffective erythropoiesis.1 Iron overload of tissue due to
inadequate or absence of chelation therapy can lead to
damage in the liver, heart, and endocrine glands.2 One of the
important endocrine pathology due to iron overload in β
thalassemia major is diabetes mellitus.3 The prevalence of
impaired glucose tolerance and diabetes in thalassemia
major varies from 8.5% to 12.2% and 5.4% to 19.5%,
respectively.1 It can occur due to impairment of insulin
secretion or due to insulin resistance.4 However, insulin
resistance ocurs before decreased insulin production and
needs to be identified early. One of the methods to measure
Received for publication March 2, 2017; accepted September 13, 2017.
From the Departments of *Pediatrics; and †Community Medicine,
Kasturba Medical College, Mangaluru, Manipal University, Karnataka,
India.
The authors declare no conflict of interest.
Reprints: Kamalakshi G. Bhat, MD, Department of Pediatrics, Kasturba
Medical College, Mangaluru, Manipal University, Karnataka
575001, India (e-mails: kamalakshibhat@gmail.com; kamalakshi.
bhat@manipal.edu).
Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
J Pediatr Hematol Oncol
Volume 00, Number 00, ’’ 2017
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
insulin resistance is homeostasis model assessment of insulin
resistance (HOMA-IR) which is reliable and used in large
epidemiological studies.5,6 However, it requires the meas-
urement of fasting serum insulin which is costly and not
available in most of the laboratories of developing
countries.4 Insulin resistance can be measured by an alter-
native method (triglyceride glucose [TYG] index) which is
the product of fasting glucose and triglycerides and this is
comparable with the HOMA-IR index.6
METHODS
A hospital-based cross-sectional study was conducted
at a medical college hospital in coastal Karnataka, India
from October 2015 to September 2016. β thalassemia
patients aged 5 years and above who were on regular packed
cell transfusions (10 to 15 mL/kg, every 3 to 4 wk) for at
least 3 years with a serum ferritin value > 1500 ng/mL were
included in the study. Subjects with β thalassemia inter-
media, β thalassemia with history of diagnosed diabetes,
hyperlipidemia, thyroid disorders, and receiving insulin or
antidiabetic drugs were excluded from the study. Sample
size was calculated based on the prevalence of diabetes in
chronically transfused patients with β thalassemia major to
be 13% as stated in a previous study and at 95% confidence
interval and 80% power. After adjustment of the calculated
sample size for finite population the final sample size was
calculated as 73 patients.
Patients were recruited into the study after getting the
institutional ethics committee approval and informed con-
sent/assent. A semistructured proforma was prepared to
record the data. Patients were instructed not to consume any
other medications (apart from chelators) before transfusion.
All blood samples were obtained after an overnight 8 to
10 hours fasting. The fasting plasma glucose and trigly-
cerides were analyzed immediately within 1 hour using
COBAS 6000(E501) fully automatic clinical chemistry
analyzer. Serum ferritin was assessed using COBAS 6000
(E601) using principle of enzyme chemiluminiscence
immuno assay (ECLIA). The sample obtained for serum
insulin was stored immediately at −20°C in the central
laboratory. The serum insulin samples were analyzed in 2
batches, within 1 month after collection using human insulin
enzyme-linked immunosorbent assay kit (DRG-Insulin
ELISA KIT) as per the standard procedure by a trained
biochemist.
The insulin resistance was calculated using HOMA-IR
index and TYG index from the obtained samples. The
mathematical formula which was used for the calculation of
HOMA-IR index and TYG index are as follows.7
|1 www.jpho-online.com
 Ansari et al J Pediatr Hematol Oncol
Volume 00, Number 00, ’’ 2017

Homeostasis Model Assessment of insulin Resistance ðHOMA‐IRÞ index ¼

Fasting Plasma Insulin mIU=L  Fasting Plasma Glucose mg=dL
405



Fasting Triglycerides mg=dL  Fasting Glucose mg=dL
TYG INDEX ¼ Log :
2

On the basis of previous studies, the HOMA-IR cut-off
values for insulin resistance were obtained as 2.67 in boys
and 2.22 in girls in the prepubertal period, and in pubertal
period the cut-off values were 5.22 in boys and 3.82 in girls,
respectively.8 From earlier study the minimum value of the
TYG Index for the diagnosis of insulin resistance corre-
sponds to 4.3.7 Data entry and analysis was carried out
using SPSS (Statistical Package for Social Sciences) Version
11.5. Basic sociodemographic data of the study participants
were expressed in terms of frequency and percentage for
categorical variables and means with SD for continuous
variables. Serum levels of fasting glucose, triglycerides,
serum insulin, serum ferritin were expressed in means with
SDs. Association between insulin resistance using HOMA-
IR index and TYG index and important variables (age, sex,
fasting glucose, triglycerides, serum insulin, and serum fer-
ritin) were tested using the χ2 test, the Fishers exact test,
Karl pearsons coefficient of correlation. P < 0.05 was con-
sidered as significant and <0.01 as highly significant.
receiver operating charecteristic curve (ROC) curve was
generated for TYG index and sensitivity and specificity was
calculated.
RESULTS
Our study population was 73 which consisted of 37 males
(51%) and 36 females (49%), 28 (38.4%) belonged to 1 to
5 years, 35 (47.9%) aged 10 to 15 years, and 10 (13.7%) 15 years
and above of age. All our patients were 5 years and above of
age, receiving packed cell transfusion for > 3 years and having a
serum ferritin value > 1500 ng/mL. Our study population had
equal sex distribution with majority of them (48%) aged between
10 to 15 years and 14% of them 15 years and above. Two thirds
of the children were diagnosed before the age of 1 year. The
mean age of our study population was 12.27 ± 4.18 years
Anthropometric assessment conducted showed 42.46% to be
underweight and 46.57% to be stunted for age.
Insulin resistance was found in 32% using HOMA-IR
index and 16% using TYG index. The mean fasting glucose
was 104.04 ± 22.615 mg/dL (with 95% confidence interval,
98.852-109.228), mean triglycerides 164.18 ± 59 mg/dL (with
95% confidence interval, 150.417-177.943) and serum insulin
19.63 ± 13.75 mIU/L (with 95% confidence interval, 16.476-
22.784). The mean serum ferritin was 5156 ± 62 ng/mL (with
95% confidence interval, 5141.777-5170.222).
In the present study 32% of patients had HOMA-IR
index values above age-specific and sex-specific cut-off levels
which showed that these children had insulin resistance
(Tables 1, 2). The mean HOMA-IR value was found to be
5.19 ( ± 4.28) with 95% confidence intervals of 4.208-6.172.
In our study, we found that there was statistical significance
between the age of the subjects and the HOMA-IR index
values, indicating that as age advances insulin resistance
increases (Table 1). The association between serum ferritin
values and HOMA-IR index value was highly statistically
significant (P < 0.001) (Table 3). This indicates that at higher
serum ferritin levels, HOMA-IR values were above the cut-
offs, suggesting the presence of insulin resistance.
Mean TYG index value was 3.99 ± 0.28 (with 95%
confidence intervals of 3.926-4.054). In total, 16% of the
patients had TYG index values above the cut-off levels
indicating insulin resistance. It did not have association with
age and serum ferritin levels, but had statistically significant
correlation with serum insulin levels (Table 4). Among the
subjects with HOMA-IR index values above cut-off levels,
31.5% had TYG values above cut-off and 68.5% had TYG
values below cut-off (Table 5). The sensitivity and specificity
of TYG index was 39.13% and 94%, respectively (Table 6).
A ROC analysis was obtained to determine the mean cut-off
for TYG index (Fig. 1). It was found that at a value of
4.0215 the TYG index has a sensitivity of 78.3% and a
specificity of 70%. The area under the curve was found to be
0.787. By lowering the cut-off of TYG index to 4.0215, the
sensitivity increased significantly to 78.3% while retaining a
specificity of 70%.

TABLE 1. Association of HOMA-IR Index and TYG Index With the

Age of Patients
HOMA-IR TYG TABLE 2. Association of HOMA-IR Index and TYG Index With Sex
Age of Patients
Group (y) < Cut-off < Cut-off < Cut-off < Cut-off Total
HOMA-IR TYG
5-10 26 2 24 4 28
Age Group < Cut-off > Cut-off < Cut-off > Cut-off Total
92.9% 7.1% 85.7% 14.3% 100%
10-15 18 17 29 6 35 Male 25 12 28 9 37
51.4% 48.6% 82.9% 17.1.% 100% 67.6% 32.4% 75.7% 24.3% 100%
> 15 6 4 8 2 10 Female 25 11 33 3 36
60.0% 40.0% 80.0% 20.0% 100% 69.4% 30.6% 91.7% 8.3% 100%
Total 50 23 61 12 73 Total 50 23 61 12 73
68.5% 31.5% 83.6% 16.4% 100% 68.5% 31.5% 83.6% 16.4% 100%
HOMA-IR indicates homeostasis model assessment of insulin resistance; HOMA-IR indicates homeostasis model assessment of insulin resistance;
TYG, triglyceride glucose. TYG, triglyceride glucose.

2 | www.jpho-online.com Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
 J Pediatr Hematol Oncol
Volume 00, Number 00, ’’ 2017 Study of Insulin Resistance in Thalassemia Major by TYG Index

TABLE 3. Association Between Serum Ferritin Values, HOMA-IR Index, and TYG Index
HOMA-IR TYG Index
Ferritin (ng/mL) < Cut-off > Cut-off Total < Cut-off > Cut-off Total
1500-2500 21 21 21 20 1 21
100% 100% 100% 95.2% 4.8% 100%
2501-4000 11 16 16 14 2 16
68.8% 100% 100% 87.5% 12.5% 100%
4001-5500 7 13 13 10 3 13
53.8% 100% 100% 76.9% 23.1% 100%
5501-7000 6 9 9 6 3 9
66.7% 100% 100% 66.7% 33.3% 100%
> 7000 5 14 14 11 3 14
35.7% 100% 100% 78.6% 21.4% 100%
Total 50 73 73 61 12 73
68.5% 31.5% — 83.6% 16.4% —
HOMA-IR indicates homeostasis model assessment of insulin resistance; TYG, triglyceride glucose.

DISCUSSION 164.18 ± 59 as compared with 141.712 of Sedigeh et al13 with

Iron overload resulting in diabetes mellitus is one of the
major complications9 of chronic transfusion-dependent β
thalassemia major and it may be missed unless looked for
and can even present as an acute emergency like diabetic
ketoacidosis. It can occur due to impairment of insulin
secretion or due to insulin resistance.10 However, insulin
resistance ocurs before decreased insulin production and
needs to be identified early.11
Mean age of our study population was similar to the
earlier studies from India.1,12 In the present study the mean
fasting glucose was 104.04 ± 22.615 mg/dL. In total, 22% of our
study population had impaired fasting glucose and 12% had
fasting glucose levels in the range of provisional diagnosis of
diabetes which was similar to the earlier study.13 In total, 22%
of the patients had fasting serum insulin levels > 25 mIU/L,
indicating possible insulin resistance. The higher insulin levels
might possibly be due to a compensation for the insulin
resistance in an attempt to maintain euglycemia. The rise in
serum insulin is postulated to be due to decreased hepatic
insulin extraction rather than an increase in secretion.14 The
mean fasting serum triglycerides in our study population was
26% having borderline high levels.The serum ferritin value
was > 7000 ng/mL in 12.3% of the patients which could be due
to inadequate dosage of chelation therapy with increasing age
or related to compliance issues.
In the present study 31.5% of patients had HOMA-IR
index values above age-specific and sex-specific cut-off
levels, indicating insulin resistance. The mean HOMA-IR
value was found to be 5.19 ± 4.28. This was much higher
than the mean HOMA-IR index values of Bhat and
Periasamy,1 (3.12 ± 1.669), Suvarna et al12 (2.18 ± 0.64), and
Sedigeh et al13 (4 ± 4.7). In our study, we found that there
was statistical significance between the age of the subjects
and the HOMA-IR index values, indicating that as age
advances insulin resistance increases.
Mean TYG index in the study population was
3.99 ± 0.28. 16% of the patients had TYG index values
above the cut of levels indicating insulin resistance. It did
not have association with age and serum ferritin levels, but
had statistically significant correlation with serum insulin
levels. Among the subjects with HOMA-IR index values
above cut-off levels, 31.5% had TYG values above cut-off

TABLE 5. Association Between HOMA-IR Index and TYG Index

TABLE 4. Association Between Fasting Serum Insulin Values and
TYG Index HOMA-IR

TYG Parameter > Cut-off < Cut-off Total

Fasting serum insulin (mIU/L) < Cut-off > Cut-off Total TYG
> Cut-off
< 25 Count 9 3 12
Count 52 5 57 % within fasting insulin 75.0 25.0 100
% within fasting insulin 91.2 8.8 100 % within TYG 39.1 6.0 16.4
% within TYG 85.2 41.7 78.1 < Cut-off
≥ 25 Count 14 47 61
Count 9 7 16 % within fasting insulin 23.0 77.0 100
% within fasting insulin 56.3 43.8 100 % within TYG 60.9 94.0 83.6
% within TYG 14.8 58.3 21.9 Total
Total Count 23 50 73
Count 61 12 73 % within fasting insulin 31.5 68.5 100
% within fasting insulin 83.6 16.4 100 % within TYG 100 100 100
% within TYG 100.0 100 100
HOMA-IR indicates homeostasis model assessment of insulin resistance;
TYG indicates triglyceride glucose. TYG, triglyceride glucose.

Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved. |3 www.jpho-online.com
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
 Ansari et al J Pediatr Hematol Oncol
Volume 00, Number 00, ’’ 2017

CONCLUSIONS
TABLE 6. Validity of TYG Index
Prevalence of Insulin resistance in our study by using
Confidence Interval HOMA-IR was 32% and by TYG index was 16% in
Parameter Estimate Lower Upper children with β thalassemia major on regular blood trans-
fusion. In ROC analysis by reducing the TYG index cut-off
Sensitivity 39.13 19.18 59.08
Specificity 94.00 87.42 100.58
to 4.0215 from 4.3, sensitivity can be increased to 78.3%
Positive predictive value 75.00 50.50 99.50 with a specificity of 70%. Hence we recommend TYG index
Negative predictive value 77.05 66.50 87.60 as a screening test for insulin resistance in patients with
Overall accuracy 76.71 67.02 86.41 β thalassemia major at the cut-off level of 4.0215.
κ = 0.38 P = 0.0001 —
REFERENCES
TYG indicates triglyceride glucose.
1. Bhat KG, Periasamy PK. Effect of long-term transfusion
therapy on the glycometabolic status and pancreatic Beta cell
function in patients with Beta thalassemia major. J Fam Med
Prim Care. 2014;3:119–123.
2. Bas M, Gumruk F, Gonc N, et al. Biochemical markers of
glucose metabolism to estimate the degree and progression of
iron overload in patients with beta-thalessemia major. Ann
Haematol. 2015;94:1099–1104.
3. Metwalley KA, Saied AR. Glucose homeostasis in Egyptian
children and adolescent with beta thalassemia major. Indian
J Endocrinol Metab. 2014;18:333–339.
4. Guerrero-Romero F, Simental-Mendía LE, Gonza lez-Ortiz M,
et al. The product of triglycerides and glucose, a simple measure
of insulin sensitivity. Comparison with the euglycemic-hyper-
insulinemic clamp. J Clin Endocrinol Metab. 2010;95:3347–3351.
5. Soliman AT, Yasin M, Awwa EA, et al. Detection of glycemic
abnormalities in adolescent with beta thalassemia using continuous
glucose monitoring. Indian J Endocrinol Metab. 2013;17:490–495.
6. Simental-Mendía LE, Rodríguez-Morán M, Guerrero-Romero
F. The product of fasting glucose and triglycerides as surrogate
for identifying insulin resistance in apparently healthy subjects.
Metab Syndr Relat Disord. 2008;6:299–304.
7. Jamshir M, Fayaz M, Mirbehbahani N, et al. TyG index and
insulin resistance in beta-thalassemia. Int J Diabetes Dev Ctries.
2015;35(S3):s529–s534.
8. Kurtoglu S, Hatipoglu N, Mazcoglu M, et al. Insulin resistance
in obese children and adolescents: HOMA-IR cut-off levels in
the prepubertal and pubertal periods. J Clin Res Pediatr Endocrinol.
2010;2:100–106.
FIGURE 1. ROC curve of TYG Index. ROC indicates receiver 9. Kushner JP, Porter JP, Olivieri NF. Secondary iron overload.
operating charecteristic curve; TYG, triglyceride glucose. Hematology Am Soc Haematol Educ Program. 2001:47–61.
10. Hafez M, Youssry I, El-Hamed FA, et al. Abnormal glucose
and 68.5% had TYG values below cut-off. We also calcu- tolerance in beta-thalassemia: assessment of risk factors.
Hemoglobin. 2009;33:101–108.
lated the sensitivity and specificity of TYG index which
11. Singh H, Garg MK, Tandon N. A study of insulin resistence by
came to 39.13% and 94%, respectively. HOMA-IR and its cut off value to identify metabolic syndrome
The κ value was calculated to be 0.38 (0.2-0.4) which in urban Indian adolescent. J Clin Res Paediatr Endocrinol. 2013;
shows fair agreement. Pearsons correlation coefficient was 5:245–251.
0.572 which showed good correlation between HOMA-IR 12. Suvarna J, Ingle H, Deshmukh CT. Insulin resistence and beta
and TYG. The P-value was found to be 0.0001 which is cell function in chronically transfused patients of Thalassemia
highly statistically significant. A ROC analysis was obtained Major. Indian Pediatr. 2006;43:393–399.
to determine the mean cut-off for TYG index (Fig. 1). It was 13. Sedigheh S, Taghi M, Ashtiani H, et al. Evaluation of serum
found that at a value of 4.0215 the TYG index has a insulin, glucose, lipid profile, and liver function in β -Thalassemia
major patients and their correlation with iron overload. Lab
sensitivity of 78.3% and a specificity of 70%. The area under Med. 2010;41:486–489.
the curve was found to be 0.787. By lowering the cut-off of 14. Gherghechchi R, Habibzadeb A. Insulin resistence and beta cell
TYG index to 4.0215, the sensitivity increased significantly function in patients with beta thalassemia major. Hemoglobin.
to 78.3% while retaining a specificity of 70%. 2015;39:69–73.

4 | www.jpho-online.com Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
Copyright r 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.