Normative Data for Myocardial Native T1 and Extracellular Volume Fraction in Children

Abstract

Purpose

To establish normative data for myocardial T1, including extracellular volume (ECV) fraction, in healthy children.

Materials and Methods

In this retrospective, single-center study, T1 mapping data were collected from 48 healthy pediatric patients (14 years ± 3 [standard deviation]; range, 9–18 years; 27 of 48 [56%] male) referred for cardiac screening 1.5-T MRI between 2014 and 2017. T1 relaxometry was performed using a 5(number of heartbeats [nHB])3 modified Look-Locker inversion recovery (MOLLI) sequence, where nHB was three to five heartbeats depending on the heart rate, and was repeated 15 minutes following the administration of 0.2 mmol per kilogram of body weight of gadobenate dimeglumine, with 19 patients receiving contrast material. T1 values were calculated using a curve-fitting algorithm on average region-of-interest signal and corrected for imperfect inversion pulse efficiency. Comparisons within patients were performed with paired Student t test, between groups with unpaired Student t test or Mann-Whitney U test, and linear regression was performed to examine for associations with other variables.

Results

Average native T1 was 1008 msec ± 31, with a nonsignificant increase in females (1017 msec ± 27 vs 1001 msec ± 33, P = .066). Average ECV was 20.8% ± 2.4, with a nonsignificant increase in values in females (21.7% ± 1.9 vs 20.0% ± 2.6, P = .123). T1 and ECV values were increased in the septum versus the free wall.

Conclusion

Normative data are presented for myocardial native T1 and ECV using the MOLLI T1 mapping sequence at 1.5 T.

Supplemental material is available for this article.

© RSNA, 2020

Summary

Normative myocardial native T1 and extracellular volume fraction (ECV) data are reported for healthy children and adolescents at 1.5 T, with average native T1 time of 1008 msec ± 31 (standard deviation) and ECV of 20.8% ± 2.4, with nonsignificantly increased values in girls.

Key Points

• ■ Normative T1 mapping data at 1.5 T using a contemporary modified Look-Locker inversion recovery sequence in healthy children and adolescents are reported, and an understanding of the specifics of the T1 mapping pulse sequence is vital to compare with reported literature values.

• ■ Increased myocardial native T1 is present in girls compared with boys for the interventricular septum (1035 msec ± 26 vs 1012 msec ± 34, P = .014), with nonsignificant increased extracellular volume (ECV) fraction (22.7% ± 2.1 vs 20.9% ± 2.9, P = .133).

• ■ Increased myocardial native T1 and ECV are found in the interventricular septum compared with the free wall (1022 msec ± 32 vs 1001 msec ± 42, P < .001).

Introduction

Measurement of myocardial T1 relaxation times and extracellular volume (ECV) fraction by using cardiac MRI provides noninvasive biomarkers for diffuse myocardial fibrosis, demonstrating excellent correlation with invasive histologic methods (1,2). Myocardial fibrotic remodeling occurs in a variety of pediatric and congenital heart diseases and is associated with ventricular dysfunction and arrhythmias (3,4). Cardiac MRI metrics of diffuse myocardial fibrosis have been shown to be increased in children following repair of tetralogy of Fallot (5), Fontan physiology (6), heart transplantation (2), and muscular dystrophy (7).

The clinical and scientific utility of measures of diffuse fibrosis in children is hampered by a lack of reference values. While it has been suggested that each center establish their own local reference values (8), this is challenging in the pediatric population due to limitations to research in children, especially when it involves intravenous access and administration of gadolinium-based agents.

The primary aim of this study was to establish pediatric normative myocardial native T1 times and ECV fractions using the modified Look-Locker inversion recovery (MOLLI) approach.

Materials and Methods

Study Population

This study was approved by the research ethics board at The Hospital for Sick Children (study number 1000053256). Due to the retrospective nature of the study, the need for informed consent was waived.

Between April 2014 and March 2017, a convenience sample of healthy pediatric patients, between 9 and 18 years of age, who underwent a clinical cardiac MRI examination including T1 mapping, were identified via the institutional cardiac MRI database. Patients were included if there was no known history of recent viral illness and the cardiac MRI was indicated for: (a) screening of an asymptomatic individual based on a family history of cardiomyopathy or sudden cardiac death, in whom all other tests, including the cardiac MRI study and genetic workup, were normal; (b) anatomic clarification based on findings from echocardiography, such as difficulty visualizing portions of the aorta, pulmonary veins, or coronary arteries, which were found to be normal at cardiac MRI; (c) a workup of syncope or chest pain if the clinical suspicion of a cardiac cause was low and if the cardiac MRI study revealed normal proximal coronary arteries and origins and was otherwise normal; and (d) a workup for inverted T waves in leads V1-V3 (which are a normal finding in children and most adolescents) or frequent monomorphic premature ventricular complexes without coupled complexes if the cardiac MRI study was normal.

Cardiac MRI

Studies were performed with a single 1.5-T system (Avanto, software release VB17 and VE11B; Siemens Medical Solutions, Erlangen, Germany), with a phased-array flexible surface coil for signal receiving and the inherent system body coil for radiofrequency transmission. Assessment of ventricular volumes, function, and myocardial mass was performed using balanced steady-state free precession imaging, acquired as a short-axis stack. Typical parameters were as follows: minimal repetition and echo times; flip angle, 70°; in-plane spatial resolution between 1.5 and 2.0 mm²; slice thickness, 5 mm; gap adjustment to cover both ventricles with at least nine slices; and temporal resolution to provide 20 true reconstructed frames per cardiac cycle.

T1 quantification was performed at a midventricular short-axis level during diastasis, using a 5(number of heart beats [nHB])3 MOLLI sequence, including inline motion correction, where nHB was three to five heartbeats, depending on the heart rate, to allow for T1 recovery between the two inversion experiments. Typical scan parameters are provided in Table E1 (supplement) and include the following: slice thickness, 8 mm; flip angle, 35°; echo time, 1.13–1.26 msec; repetition time, 2.68–2.95 msec; bandwidth, 1085 Hz/pixel; minimum inversion time, 100–120 msec with 80-msec increments; and generalized autocalibrating partially parallel acquisitions with an acceleration factor of two. When the gadolinium-based contrast agent was injected, T1 quantification was repeated 15 minutes after administration of 0.2 mmol of gadobenate dimeglumine (MultiHance; Bracco Diagnostics, Montreal, Canada) per kilogram of body weight.

Image Analysis

Ventricular volume analysis was performed in Qmass (Medis Suite, version 2.1; Medis Medical Imaging Systems, Leiden, the Netherlands) following a standardized approach, with manual tracing of the endocardial borders of both ventricles on the short-axis images for quantification of end-diastolic and end-systolic volumes. Papillary muscles were included in the blood pool. Ejection fraction was calculated as the stroke volume (end-diastolic − end-systolic volumes), normalized to the end-diastolic volume. Left ventricular (LV) epicardial tracing was also performed to derive the myocardial mass, calculated from the myocardial muscle volume multiplied by the specific gravity of the myocardial tissue (1.05 g/mL). All volumes and masses were indexed to the individual’s body surface area, with z scores derived based on data from Buechel et al (9). Volumetry analysis was performed by S.J.Y., M.S., and L.G.W., each with greater than 10 years of cardiac MRI experience.

T1 quantification was carried out by one observer (J.J.P.), with 7 years of cardiac MRI experience, in Qmap (Medis Suite, version 2.1; Medis Medical Imaging Systems). For assessment of interobserver and intraobserver reliability, analysis was repeated by a second, independent observer (C.Z.L., 2 years of cardiac MRI experience) blinded to any previous results. Studies with marked artifacts were excluded from analysis. LV endo- and epicardial borders were contoured on the inline motion–corrected T1-weighted images, including only the inner 50% of myocardium to avoid partial volume errors (Fig 1). Contours were manually adjusted on individual T1-weighted images to compensate for incomplete motion correction. A single region of interest was placed in the blood pool of the LV cavity. The LV myocardium was divided into six segments according to the American Heart Association model, which allowed for removal of segments containing visible artifact and for quantification of septal (average of T1 in the antero- and inferoseptal segments) and LV free wall (average of T1 in the antero- and inferolateral segments).

Figure 1:

Sample native T1 image from a healthy 13-year-old male pediatric proband, including endocardial (red), epicardial (green), and blood pool contours. The slice is divided into six equidistant segments (S1-S6) as indicated by the yellow lines, anchored to the anterior right ventricular insertion point (blue circled cross).

T1 values were calculated using average region of interest signal and a curve-fitting algorithm. Resultant T1 times were corrected for incomplete inversion using correction factors depending on the inversion pulse utilized: 1.0811 for a traditional hyperbolic secant pulse and 1.0365 for a tangent/hyperbolic tangent adiabatic pulse (10). In those patients receiving intravenous gadolinium-based contrast agent, ECV was calculated as previously described (8) using pre- and postcontrast T1 times and the patient’s hematocrit level, with blood drawn immediately prior to the cardiac MRI and analyzed in the hospital central laboratory.

Statistical Analysis

Values are reported as mean ± standard deviation, or count (as percentages), where applicable. Comparisons within the same patients were performed using paired Student t tests. Comparisons between groups of patients were undertaken using the unpaired Student t test or Mann-Whitney U test, depending on the normality of the data. Associations between myocardial native T1 or ECV and other parameters were assessed using univariate linear regression, with variables included in a multivariate regression model if the P value was less than .1 on univariate testing. Reliability was assessed using intraclass correlation coefficient for interobserver and intraobserver reliability, with additional assessment using Bland-Altman plots and associated limits of agreement. A P value of .05 was regarded as significant. Statistical analysis was performed using STATA software (version 11.2; Stata, College Station, Tex).

Results

The cardiac MRI examinations of 50 patients met the inclusion criteria and were included in the study. Two patients were excluded due to significant image artifacts precluding analysis. Patient characteristics for the 48 analyzed patients are outlined in Table 1. The most common indication for cardiac MRI was for screening of an asymptomatic individual based on a positive family history (19 of 48, 40%), followed by anatomic clarification (12 of 48, 25%), syncope or chest pain (12 of 48, 25%), and finally for abnormal T waves/premature ventricular contractions (five of 48, 10%). The youngest patient was 8.6 years old, while the oldest was nearly 18 (Fig 2), and there was no significant difference in ages between boys and girls (P = .057). There is a significant correlation with body surface area (BSA) and age (R² = 0.513, P < .001).

Table 1:

Demographics, Mass, and Volumes

Figure 2:

Age distribution of study cohort.

Cardiac MRI results from volumetry are listed in Table 1. While ventricular mass and volumes were different between sexes (LV mass, P = .003; indexed left ventricular end-diastolic volume [LVEDVi], P = .015; indexed left ventricular end-systolic volume [LVESVi], P = .001), there was no difference between z score values for these metrics between sexes (LV mass z score, P = .873; LVEDVi z score, P = .078; LVESVi z score, P = .889). While there was no significant sex difference between LV mass-to-volume ratio (P = .082), there was a slightly higher ejection fraction found in females (P = .047). T1 and ECV results are presented in Tables 2 and 3. Myocardial native T1 was longer in girls when considering only the septal (P = .014) or free wall segments (P < .001), with nonsignificantly longer T1 for the entire LV in females (P = .066). Overall, the average native T1 in the interventricular septum was significantly higher than in the free wall (1022 msec vs 1001 msec, P < .001); however, this difference remained significant only in boys (P < .001). The average native blood T1 was longer in girls versus boys (P = .013), likely related to lower average hematocrit level in girls (P = .0334).

Table 2:

Native T1 Data for the Healthy Pediatric Cohort

Table 3:

Extracellular Volume Fraction Data for the Healthy Pediatric Cohort

Contrast material was administered in 19 of 48 (40%) cases, and therefore permitted the calculation of ECV. ECV, although higher on average in girls, was not statistically different between sexes (P = .123). Average ECV in the interventricular septum was significantly higher than in the free wall (P = .018), though this difference did not remain statistically significant in only boys or girls (P = .052 and P = .209, respectively).

Univariate regression demonstrated significant associations between myocardial native T1 and nearly all parameters for the entire LV, interventricular septum, and free wall (Table E2 [supplement]). However, only LV mass z score remained significant for each location after multivariate analysis, while sex associations remained for the septum and free wall. Univariate regression only showed a relationship for ECV with age and BSA; however, neither remained significant on multivariate analysis. Septal native T1 and ECV are shown as a function of heart rate and age in Figure E1.

As assessed by intraclass correlation coefficients, interobserver and intraobserver reliability is overall excellent, with all values greater than or equal to 0.75 and most greater than or equal to 0.9 (Table 4). Bland-Altman plots are presented in Figures E2 and E3 (supplement).

Table 4:

Interobserver and Intraobserver Variability

Discussion

This study provides normative pediatric data for native T1, along with myocardial ECV, from a single institution, using the 5(nHB)3 MOLLI acquisition with a 1.5-T Siemens system. Myocardial T1 and ECV are altered in a variety of conditions in adults, including myocarditis (11) and cardiomyopathies (12). They are linked to outcomes and aid in the prediction of cardiac events (13,14). Reports in pediatrics are scarcer; however, elevated myocardial T1 and ECV have been demonstrated in acquired and congenital heart disease (4–6,15), following heart transplantation (2), and in cardiomyopathies (7,16,17). The use of T1 and ECV has not yet reached a stage where it independently impacts clinical management, and it is typically used to shape an overall diagnostic impression for individuals undergoing a comprehensive cardiac MRI examination. A major barrier toward a more widespread clinical application of this technique in the care of children with heart disease is the lack of normative data. In the recently published consensus statement on parametric mapping (8), it is suggested that while local reference ranges for T1 and ECV should be derived and obtained, they require comparisons with published reported ranges. Whereas it may be possible for centers to establish their respective norms for myocardial native T1 in healthy children, limitations in the administration of contrast agents will hamper the derivation of local reference values for ECV. The consensus statement (8) recommends utilization of literature values in this situation, particularly given the reduced dependence of ECV on various system and sequence parameters. Thus, this study, which to our knowledge presents one of the largest known pediatric normative values for T1 and ECV to date, using a widely applied T1 relaxometry technique (2,7,16–18), provides important benchmarks for these metrics for those performing cardiac MRI at 1.5 T, using the 5(nHB)3 MOLLI pulse sequence as described above.

The average T1 values reported in this study are higher than most published adult values using MOLLI, which are between 950 and 982 msec (19–21). This may be due to the use of the 5(nHB)3 acquisition scheme in the current study, as most adult reports employ the “traditional” 3(3)3(3)5 acquisition scheme. However, important differences between the traditional 3(3)3(3)5 and 5(nHB)3 MOLLI variants include the number of inversion pulse sets (three inversion sets for traditional MOLLI, two sets for the variant in this study), the number of images in the first inversion set (three images for the traditional MOLLI, five for the variant in this study), and the total number of images obtained (3 + 3 + 5 = 11 for traditional MOLLI, 5 + 3 = 8 for the variant in this study). By acquiring five images in the first inversion set, the 5(nHB)3 scheme increases the number of heartbeats between inversion pulses, which increases the time between inversion sets to allow for fuller T1 recovery. This decreases the heart rate sensitivity (22) and results in increased T1 times versus the original MOLLI scheme (23). Therefore, usage of the 3(3)3(3)5 acquisition scheme is becoming less common. A recent study of healthy children, using a modified version of the traditional 3(3)3(3)5 MOLLI sequence with fixed time intervals between inversion sets (3[3 sec]3[3 sec]5), which reduces heart rate dependence, reports native T1 values for the midventricular slice of 1010 msec for the whole LV and 1017 msec for the interventricular septum, which appear similar to those of the current study (24). However, ECV was not reported. In a study of a large healthy adult Chinese cohort, average myocardial T1 using the 5(3)3 MOLLI sequence was 1013 msec ± 27 and yielded higher values in female versus male patients (1025 msec ± 26 vs 1001 msec ± 23) (18), similar to our findings. This study did not report ECV values. In contrast to native T1 times, values for ECV in the current study are lower than those reported using the 3(3)3(3)5 acquisition scheme in adults which range from 25% to 27% (20,21). Whether this difference is reflective of a smaller relative extracellular matrix size or a result of differences in pulse sequences is unclear.

While a relationship between either T1 or ECV and age has been shown (25,26), most published studies did not confirm an age-related variation in healthy adults (18,20,21,27,28) or children (24,29). Myocardial native T1 values from a small control cohort of healthy children and young adults (n = 21, 71.4% males; 15.7 years ± 1.5, with a range of 12–18 years) that were published in 2017 report an average T1 value of 965.6 msec ± 30.2 using a similar 5(nHB)3 MOLLI acquisition (17). Similar to the current study, they did note lower T1 values in the lateral wall compared with the septum, but no difference in whole LV native T1 between male and female participants (964.7 msec ± 16.4 vs 967.2 msec ± 19.4, respectively; P = .40), though the control group was smaller and predominantly boys. The exact reason for the slightly lower reported T1 values as compared with those found in the present study is not readily clear. It is possible that it relates to the higher prevalence of boys, as well as a slightly older cohort, compared with our study. As only four control patients received contrast material, they did not report ECV. A slightly larger healthy cohort from the same institution is reported in a recent publication, with an average native T1 of 990 msec and ECV of 23.3% (30). Unfortunately, regional information and sex differences were not reported. Olivieri and colleagues reported myocardial native T1 and ECV for small cohorts of exclusively male patients aged 16.1 years ± 2.2 (7), using a similar acquisition strategy as in the current study. Entire LV values were not provided, only reporting septal and free wall values of 990 msec ± 34 and 978 msec ± 36, respectively (7). These values are comparable to those for boys in the current study. Olivieri et al found septal and free wall ECV values of 26.0% ± 3.3 and 24.4% ± 3.5, respectively, again all in male participants. These ECV values are higher than those found here. Of note, these studies used a different postcontrast sampling scheme than the current study, used registration algorithms to create ECV maps, and averaged values from four short-axis slices and a four-chamber slice, all of which may contribute to the differences, and thus limit the comparability.

Native T1 and ECV were increased in the septum compared with the free wall. This finding is in keeping with previous reports (17,21) and likely relates to factors affecting the accuracy of T1 in the free wall region, most prominently field inhomogeneity at the air-tissue interface (22). For this reason, current guidelines suggest the use of the interventricular septum for T1 and ECV quantification, unless the disease process is regional (8).

The increased myocardial T1 times in females in this study are in line with previously published results in adults (18,26,27), although not consistently (21,25). Additionally, there is inconsistency in published pediatric literature, with studies showing no sex differences in native T1 (24,29). In adults, hearts of men and women demonstrate different myocardial remodeling (31). This is noted on imaging, including sex-related differences in age dependencies on LV mass, mass-to-volume ratio, and stroke volume (28), and with histologic findings, with preservation of both the number and size of myocytes in women over time, versus a loss of number and increased myocyte size in men (32). This is thought to be related to different hormone environments, such as a proposed protective effect of estrogen and possible more detrimental effect of testosterone (31). Much of these data are derived from studies involving differences in adaptation before and after menopause, and thus may not be relevant in children or adolescents. In the pediatric group, the reasons for possible differences are not established. Sex differences in blood T1 were found in this study, likely related to differences in hematocrit, which may affect myocardial T1 via signal from the capillary compartment, or via partial volume effects with the intraventricular blood pool. Conservative region of interest placement was used in this study in attempts to mitigate the latter effects; however, given the differences in indexed myocardial mass between the sexes, it is possible that an effect persists. Irrespective of the cause, consideration should be given to the use of different normative values for myocardial native T1 and ECV for boys and girls.

Due to the changes in cardiac mass and chamber size during childhood, we investigated relationships with native T1 and ECV using cardiac MRI variables less associated with age and body surface area, such as ejection fraction or the z score for LV mass. Univariate regression analysis revealed significant associations with myocardial native T1 and age, sex, heart rate, and LV mass z score, but less consistently with BSA. The negative association with age and native T1 or ECV may relate to differences in the extracellular or cellular components, such as a relative increase in myocyte size, with age leading to a compensatory reduction in the extracellular component. Alternatively, this association may simply reflect difficulties in avoiding partial volume errors in those with smaller hearts. It should be noted that, as all our patients weighed more than 20 kg, a slice thickness of 8 mm was used; however, in smaller children a thinner slice appears preferable. Following multivariate analysis, only sex and LV mass z score typically remained independently associated with native T1. The relationship with LV mass z score is supportive of the hypothesis of the difficulty in eliminating partial volume errors. Interestingly, only age and BSA were significantly associated with ECV on univariate analysis; however, neither remained significant in multivariate analysis. It is possible that the smaller number of cases with ECV obscured further associations of physiologic parameters with ECV.

Several important limitations existed. The modest sample size of our study population may have obscured differences, such as between boys and girls for ECV, as well as associations between fibrosis markers and other imaging biomarkers. There is a possibility that the association of sex with T1 is confounded by gender disparities other than the differences in myocardial architecture. The cohort size may limit overall generalizability to other groups, particularly as the number of patients with ECV measurements is not large; however, the sample size is comparable to widely used reference studies for normative pediatric ventricular volumes (9,33). Second, the cohort is a retrospective sample of convenience, derived from the approximately 700 clinical cardiac MRI studies per year in our institution, in which T1 mapping is routinely performed. While all known testing, including the cardiac MRI, was normal, we cannot rule out subclinical pathology. While no age dependence was noted with T1 measures, our study population was skewed toward older children, and no information is available on children younger than 8 years of age, as most children below that age often require a form of sedation, which raises the clinical threshold to proceed with cardiac MRI and limits the prevalence of healthy, younger children. It is important to note that native T1 values can vary considerably with differences in field strength, sequence choice, and imaging parameters (22), and it is important to ensure that these parameters are comparable between clinical practice and the reference values used. In comparison, ECV is more robust toward variations in hardware, software, and settings (8), though it can show small variations between contrast agents (34). Thus, due to a variety of factors, T1 and ECV may differ between sites even with what would otherwise be matching sequence parameters. Scanner operating performance can vary or drift over time and, although we suspect that any impact would be small, we cannot rule out an influence on the values obtained. Although the discrepancies are likely small, literature references must be considered with care. This may become a less significant concern over time thanks to efforts to develop corrective algorithms and/or phantom calibration methods, such as the T1MES program (35). However, whenever possible, site-specific normative data are established in accordance with current recommendations (8).

In conclusion, normative data are presented for myocardial native T1 and ECV using the MOLLI T1 mapping sequence at 1.5 T. Consistent with findings in adults, there are suggestions of increased values in females, supporting the consideration for different normal ranges for boys and girls.