Uncovering Cardiac Involvement in Childhood Diabetes: Is it Time to Move Toward Speckle Tracking Echocardiography in Childhood Diabetes Management?

CONTEXT: One of the most common endocrine disorders in children is diabetes which is the leading cause of premature cardiovascular disease in adulthood. AIMS: This study is aimed to investigate the extend of cardiac involvement in diabetic children by speckle tracking echocardiography (STE) in comparison to two-dimensional (2D) echocardiography and routine laboratory data. SETTINGS AND
DESIGN: A cross-sectional study conducted on patients under 18 years of age who deal with type one diabetes mellitus for more than 5 years. SUBJECTS AND METHODS: To compare the STE results, we included the STE data of 25 normal age-matched children. All patients underwent laboratory analysis for lipid profile, blood sugar, and 2D echocardiography plus STE. STATISTICAL ANALYSIS USED: Two-sample independent t-test, Chi-square test, logistic regression test, Spearman's rank correlation coefficient, and the Pearson correlation coefficient.
RESULTS: From March 2018 to 2019, we included 53 patients, mean age 15.8 ± 0.39 years and 52.8% female, and 25 nondiabetic control in this study. STE revealed global longitudinal strain (LS) -18.4 versus - 24.2 for patient (44 valid cases) versus control group, respectively, with significant statistical difference. Diabetic patients had lower LS in all segments compared to the control group.
CONCLUSIONS: STE has very high sensitivity to detect cardiac involvement far earlier than 2D echocardiography. None of the routine biomarkers or demographic features can predict cardiac involvement based on segmental abnormalities of STE. Active investigation to clear the remote impact of STE abnormalities and its practical role in childhood diabetes management is highly recommended. Copyright:

INTRODUCTION

Type one diabetes mellitus (T1DM) is defined as the relative or absolute deficiency of insulin secondary to the autoimmune destruction of pancreatic beta cells. Although studies suggest that T2DM is the predominant form of DM and accounts for 85%–90% of diabetic patients, T1DM is the most prevalent type of the disease among children and more than 500,000 children are living with this condition worldwide.[1] Long-term hyperglycemia in both types of diabetes can lead to microvascular and macrovascular complications.[2] Epidemiologic studies have elucidated the association between diabetes and cardiovascular diseases (CVDs).[345] Despite the fact that CVD in children is not usually expected, even when the child has T1DM, atherosclerosis process initiates in childhood[6] and CVD risk factors in children with T1DM are more prevalent than in normal children[7] and some studies have shown that children with DM had higher low-density lipoprotein cholesterol (LDL-C), blood pressure, blood glucose, and body mass index (BMI) in comparison to the control group.[8] Unfortunately, no widely utilized prognostic algorithm is available for projecting the CVD risk in T1DM patients. In the absence of contradictory data, one of the methods used for diagnosing CVD among T1DM patients is identical to that implemented for evaluating the risk of coronary artery disease in the general population.[9] Some conventional methods exist for diagnosing CVD, though each have certain drawbacks.[10] Speckle-tracking echocardiography (STE) is a more specific method to detect segmental abnormalities of myocardium in contrast to global view of conventional two-dimensional (2D) echocardiography. STE has overcome the limitations of conventional methods (such as angle dependence and sonic interference) using a series of 2D or 3D images and tracking the distance between the pixels during the cardiac cycle.[11] Based on previous studies, 2D-STE shows global longitudinal strain (GLS) impairment in 24% of adult diabetic patients, fur more than routine 2D echocardiography, and lesser extend reported in childhood age, also. Importantly, a decrease in GLS detected by STE is associated with cardiovascular disorders and depicts an increased risk of developing CVD within the next 10 years. In children with T1DM, impairments of the longitudinal and circumferential strain in 2D-STE are known signs of hyperdynamic left ventricular (LV) contractility in the initial stages of the disease.[12] Considering the increasing prevalence of diabetes and its association with various complications, especially CVDs, population-based studies are of great importance. Hence, we need gather more evidence of cardiac involvement, mechanism, and related complications in diabetic patients. At the present time, we have paucity of evidences and studies on cardiac mechanics in T1DM patients.[1314] This study aimed to investigate the prevalence of segmental abnormalities in children with T1DM using 2D-STE and determination of the role of this modality in the evaluation and follow-up of diabetic children.

SUBJECTS AND METHODS

Study methodology

This study was an observational, cross-sectional study conducted on patients under 18 years of age with a minimum T1DM disease duration more than 5 years referred between March 2018, and 2019 to the Diabetes Center of the Imam Reza (PBUH) Clinic affiliated to the Shiraz University of Medical Sciences (Shiraz, Iran).

Target population

This study investigated the prevalence of cardiac involvement based on STE in children with T1DM using 2D-STE. The sample size was determined after a literature review of similar studies and consultation with a statistician; the calculation was performed using the MedCalc Statistical Software version 19.2.6 (MedCalc Software bv, Ostend, Belgium) with an error of 5% and a power of 80%.

Sampling and methods

Considering the inclusion and exclusion criteria of this study and the necessary sample size kept in mind, after verification of T1DM in patients, patient data such as age, sex, duration of the disease, height, weight, BMI and blood pressure were recorded by a pediatrician in a special form along with laboratory data including the levels of triglycerides (TGs), LDL-C, glycated hemoglobin (HbA1c), and fasting blood sugar; the presence or absence of microalbuminuria was also noted. A written consent form was signed by each patient's guardian. After submission of their information, the patients were referred to the cardiology ward for electrocardiography (conventional 2D echocardiography and 2D-STE). We used Vivid S6 GE Echocardiography machine. The recorded 2D echocardiography parameters were ejection fraction (EF), tissue Doppler imaging of mitral valve (MV), E/A ratio of MV inflow and E/e' ratio(septal e'), Regurgitation severity of heart valves and peak continuous wave Doppler for tricuspid regurgitation, pulmonary insufficiency, and mitral regurgitation (MR) wave. Longitudinal STE was done in 3 views of 4 chamber, Longitudinal 5 chamber and LV 2 chamber by automated function imaging semi-automated method to report 18 segments longitudinal strain (LS) and GLS.[15] In this method after 3 points selection, automated process will be started and then we readjusted the border of endocardium and epicardium to proceed automatic speckle tracking. The obtained data were registered and statistically analyzed. To minimize the STE vendor difference, we used the recorded results of 25 normal children (age 10–18 years; male/female ratio: 14/11) STE with the same echocardiography instrument and performed comparison analysis with this group for segmental analysis of STE. In total, 53 patients were investigated. The assessment of ventricular function was done by LV speckle tracking and conventional 2D-transthoracic echocardiography with the Vivid S6 GE ultrasound machine.

Statistical analysis methods

In order to assess the prevalence of cardiac involvement in children with T1DM and assess its association between other measured index and laboratory data, the analysis was done by SPSS software version 16 (IBM Corp, Armonk, New York). The association of patients' information and clinical features was assessed, where appropriate, by the two-sample independent t-test, Chi-square test, logistic regression test, Spearman's rank correlation coefficient, and the Pearson correlation coefficient. P < 0.05 was considered statistically significant and alpha, p, and d were assumed as 0.05%, 62.5%, and 7.5%, respectively. To evaluate normal distribution of data in speckle tracking results, we used the Kolmogorov–Smirnov test and considered a variable not normally distributed if Sig. <0.05.

RESULTS

Fifty-three patients under the age of 18 referred to the diabetes center (mean age = 15.8 ± 0.39 years) were enrolled in the study. The frequency distribution analysis of the subjects in this study revealed that 28 (52.8%) were female and 25 (47.2%) were male. The demographic information including age, sex, duration of the disease, and BMI are provided in Table 1 together with laboratory data including the levels of TG, cholesterol (LDL-C), and HbA1c. Among our patients, 56% had no experience of diabetic ketoacidosis (DKA) and 3.8% had 3 episodes of DKA [Figure 1]. The results of STE were valid for 44 patients and the others were excluded from the study due to poor technique. All segments LS in bull's eye model is presented in Figure 2.

Table 1

Distribution of the demographic and laboratory data of the 53 pediatric type 1 diabetes mellitus patients included in the study

Parameter	Minimum	Maximum	Mean±SD
Age (years)	8	21	15.8±0.39
BMI (kg/m2)	17.1	28.2	21.8±0.48
Disease duration (years)	5.2	14.1	8.9±0.34
HbA1ca (%)	5	11.6	9.1±0.17
TGb (mg/dL)	51	280	121.4±6.75
LDL-Cc (mg/dL)	40	140	73.3±2.90

aReference range of HbA1c=Normal <5.7%, Prediabetes=5.7%-6.4%, bReference range of TG=<100 mg/dL or 1.13 mmol/L, cReference range of LDL-C=<110 mg/dL. SD=Standard deviation, BMI=Body mass index, HbA1c=Glycated hemoglobin, TG=Triglycerides, LDL-C=Low density lipoprotein cholesterol

Figure 1

Frequency of diabetic ketoacidosis episode in the 53 pediatric patients included in the study (No: 0; 1: 1 episode; 2: 2 episode; 3: 3 or more episodes)

Figure 2

Speckle tracking echocardiography data analysis results in the diabetes patients with map of coronary arteries territory. LAD = Left anterior descending artery, LCX: Left circumflex artery, RCA: Right circumflex artery

Normality test for STE segmental analysis indicated that some segments did not follow a normal distribution and we used nonparametric tests for their LS analysis. These segments were basal anterior, basal anterolateral, basal inferolateral, basal inferior, basal inferoseptal, MAL (middle inferolateral), middle inferior, and middle inferoseptal. Other segments had normal distribution and analyzed by parametric tests.

The assessment of GLS by STE revealed −18.4 versus −24.2 in patients group versus control group with significant difference (P < 0.001). Detail comparison of all segments is presented in Table 3 (parametric data) and Figure 3 (nonparametric data). In segmental analysis, all of the segments LS had significant impairment in diabetic patients.

Table 3

The strain of 10 segments with normal distribution in patient and control groups

	Group	Mean±SD	P
BAS	Patients	−19.07±3.385	<0.001
	Control	−23.76±2.437
MAS	Patients	−21.45±3.151	0.005
	Control	−23.48±2.064
MA	Patients	−20.07±4.134	<0.001
	Control	−25.12±3.551
MIL	Patients	−17.89±6.499	<0.001
	Control	−23.84±2.656
AAS	Patients	−20.84±6.254	0.009
	Control	−24.56±3.820
AA	Patients	−19.93±6.425	0.001
	Control	−24.80±3.342
AAL	Patients	−16.66±6.527	<0.001
	Control	−24.36±2.447
AIL	Patients	−19.75±7.104	0.001
	Control	−24.80±2.901
AI	Patients	−19.68±6.512	<0.001
	Control	−25.04±4.228
AIS	Patients	−18.70±5.994	<0.001
	Control	−24.84±2.688
GLS	Patients	−18.4±2.8	<0.001
	Control	−24.2±1.4

Independent t-test was used for comparison. Two of the most impaired segments were bolded. BAS=Basal antero-septal, MAS=Middle antero-septal, MA=Middle anterior, MIL=Middle inferolateral, AAS=Apical antero-septal, AA=Apical anterior, AAL=apical antrolateral, AIL=Apical inferolateral, AI=apical inferior, AIS=Apical inferoseptal, GLS=Global longitudinal strain, SD=Standard deviation

Figure 3

comparison of patient versus controlfor nonnormally distributed segmental analysis of speckle tracking echocardiography by Mann– Whitney test and independent-samples median test that all had significant P < 0.001. The Box-plot chart showed grand median (the bold horizontal line of chart) and median of each group (bold line of each box)

On the other hand, 2D echocardiography was normal (EF mean: 60.5 ± 3.2; range 57%–78%) in most cases except in two cases who had diastolic dysfunction (reverse E/A ratio and Impaired Tissue Doppler in septal and lateral MV) [Table 2]. Both cases also had significant abnormalities in STE. Eight patients had trivial MR and 3 patients had mild MV prolapse. None of these findings were associated with symptoms. None of the 2D echocardiography parameters had not significant difference in compare to normal group.

Table 2

Two-dimensional echocardiography value for patients and control group

	Diabetes patients	Control group	P
E/A ratio	1.81±0.4	1.94±0.5	0.812
EF	60.5±3.2	62.5±2.9	0.325
E/septal e’ ratio	6.1±0.82	6.4±0.93	0.634
Septal e’	12.1±3.20	12.8±2.8	0.113
Lateral e’	19.2±2.86	18.4±3.04	0.097

EF=Ejection fraction

Descriptive analysis of the obtained data showed that 43 children (81.1%) had normal blood pressure ranging from 110/70 to 130/90 mmHg, while 10 (18.9%) had abnormal blood pressure values (hypertension or hypotension). Furthermore, investigating the distribution of microalbuminuria in our subjects showed that 20 children (37.7%) did not have microalbuminuria, whereas 33 (62.3%) were diagnosed with this condition. GLS was compared across these factors. We could not find any significant difference between GLS across these factors and their categories [Table 4]. Even DKA episodes had no significant impact on GLS.

Table 4

Comparison of mean global longitudinal strain across sex, microalbuminuria, blood pressure, and frequency of diabetes ketoacidosis episodes

	Category (n)	Mean GLS	P
Frequency of DKA episodes	0 (22)	−18.3±2.1	0.345
	1 (11)	−17.8±2.6
	2 (9)	−18.6±3.6
	3 (2)	−21.8±3.8
Sex	Male (21)	−18.3±3.1	0.891
	Female (23)	−18.4±2.6
Microalbuminuria	No (18)	−18.4±2.6	0.990
	Yes (26)	−18.4±3
Blood pressure	Normal (37)	−18.1±2.9	0.207
	Abnormal (7)	−19.6±2.3

GLS=Global longitudinal strain, DKA=Diabetes ketoacidosis, n=Number

Pearson correlation also did not show any linear correlation between GLS, BMI, TG, Cholesterol, diabetes duration, or HbA1c [Table 5].

Table 5

Pearson correlation analysis of global longitudinal strain and different parameters

Pearson correlation	Duration of diabetes	BMI	HBA1C	TG	Cholesterol
GLS mean	−0.034	−0.112	0.165	−0.030	0.104
P	0.826	0.469	0.285	0.845	0.501

None of them had a significant correlation. GLS=Global longitudinal strain; BMI=Body mass index; HbA1c=Glycated hemoglobin hemoglobin; TG=Triglycerides

DISCUSSION

The present study investigated the role of STE in the evaluation of diabetic children. This new technology facilitates more accurate observation of cardiac mechanics and function. STE highlights the segmental motion of myocardium and new concepts of cardiac function analysis. The GLS is used to evaluate the function of the left ventricle and predicts the prognosis of patients with CVD in various populations. This parameter can be used as a more sensitive diagnostic modality for early heart dysfunction, providing higher efficiency in diagnosis than conventional measures.[16] The most valuable role of STE is detection of segmental ischemic change in the heart. Hence, considering screening tools in patients with DM, one of the leading causes of CAD, it may be a good choice to move toward STE.

The descriptive data demonstrated that patients had a mean age of 15.8 years and were homogeneously distributed in terms of sex (52.8% female vs. 47.2% male).

In total, 81% of the children had normal blood pressure values; the mean level of HbA1c was calculated at 9.1%. Microalbuminuria was evident in 62.3% of the cases and 43.4% had experienced ketoacidosis. The evaluation of mean BMI (21.80 ± 0.4 kg/m2) revealed that patients had normal weight range (50th to 75th percentile). The mean TG level of the patients (121.4 ± 6.7 mg/dL) was higher than the normal limit for children which may contribute to premature CVD. The normal BMI and blood pressure values among both female and male pediatric T1DM patients in this study may be due to the short duration of the affliction or the environmental and nutritional factors associated with the geographic situation of Iran. The published reports suggest that obesity and hypertension are more prevalent among male diabetic patients, though these reports contain some paradoxes, especially regarding the sex factor.[17181920]

The presented study showed the clear difference between STE finding of diabetic child (44 valid cases) and normal child (GLS: −18.4 vs. −24.2) unlike the minimum finding in 2D echocardiography (only 2 patients with diastolic dysfunction). It clearly showed the lack of accuracy of 2D echocardiography to detect subtle changes in cardiac function that were uncovered by STE. The result of segmental analysis of longitudinal strain showed that all of cardiac segments were involved. The STE abnormalities not restricted to coronary involvement and in many other cases such as hypertrophic cardiomyopathy, chemotherapy-induced cardiotoxicities, and many other cardiomyopathies showed valuable data to elucidate cardiomyopathy in very early stages.[1621222324] In one of our patients due to significant STE abnormalities and diastolic dysfunction, we recommended to perform computed tomography angiography that was normal. To reach a scientific result, we need more cases and studies, but at least, it can imply that the significant coronary arteries involvement is not the only explanation for STE abnormalities in a diabetic patient. Different studies present several mechanisms for cardiomyopathies in diabetes that Hensel et al. named some of these mechanisms. “Diabetes-induced loss of t-tubule structure” or abnormal cross linking of collagen and change in mitochondrial energy production are some of important pathological explanation for these changes.[25] To find the vivid mechanism, we need more studies including animal studies.

The most involved segments, apical anterolateral and middle inferolateral, also should be interpret with considering of these explanations and should be rechecked by other investigators. These segments mainly supplied by left anterior descending artery and left circumflex artery and somewhat right circumflex Artery; but confirmation of direct correlation of this finding with coronary artery involvement is one of our study limitations. These segments may be more vulnerable in confronting with cardiotoxic drugs, if these data will represent by other investigators.

We tried to predict patients who are at more risk for abnormal GLS, but we could not find any correlation between duration of diabetes, BMI, lipid profile, and even HbA1c and the result of STE. It seems that the segmental changes of STE are independent of routine biomarker in diabetes and are not predictable base on these laboratory data. Some other investigators have found some correlation with HbA1c when the duration of diabetes extend to more than 10 years. In these patients, high HbA1c is a predictor of worsening of GLS (No circumferential strain).[2526] Data regarding GLS impairment in diabetes are more consistent but regarding duration there is some controversies.[2527] It is worth to mention, most children with more than 8 years' history of diabetes enter adolescence or adulthood, and may have some other confounding factors such as cigarette smoking and carry other risk factors of CAD that may interfere in comparison to these studies. Hence, most of similar studies agree with low sensitivity of 2D echocardiography and laboratory data to detect childhood diabetic cardiomyopathy,[2627] and on the other hand, improvement of technique and feasibility of STE highlight the role of this modality to evaluate these patients.[28]

The unclear side of STE in children is the prognostic impact of STE findings. The questions are: What the importance of these changes are? what the cutoff point is for therapeutic intervention in diabetic child when we find these abnormalities? Answering to these questions need longitudinal studies in diabetes patients and cohort studies to follow cardiovascular events in these patients based on STE finding. There is a large gap in this area and we highly recommend more investigations in this field to make clear the practical importance of STE in childhood diabetes.

Study limitation

The segmental approach has some technical limitations such as poor view in apical segments that we tried to capture multiple images from each patient to record best possible 2D image before progressing to STE.

Finding the exact mechanism of poor STE including coronary artery supply need more investigation by contrast study of this vessel. Hence, we cannot present the final mechanism of involvements.

CONCLUSIONS

2D echocardiography has poor sensitivity for the detection of cardiac function abnormalities in context of childhood DM. On the other hand, most children had segmental abnormalities in STE and it can detect subclinical involvement of cardiac disease in diabetic children. STE has very good potentiality to detect cardiac involvement far earlier than 2D echocardiography and we recommend it for use in practical approach. None of the routine biomarkers or demographic features can predict cardiac involvement based on segmental abnormalities of STE. We recommend that the remote impact of STE abnormalities in diabetes patients should be actively investigated to make clear the practical role of STE in childhood diabetes management.

Ethical clearance

This study was approved by the Ethics Committee of the Shiraz University of Medical Sciences (IR.SUMS.MED.REC.1398.096) and all the study subjects were informed about the study protocol and a written consent was obtained before baseline examinations.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.