Longitudinal changes in extent of late gadolinium enhancement in repaired Tetralogy of Fallot: a retrospective analysis of serial CMRs

A retrospective review of all patients with repaired TOF who had undergone 2 or more CMR examinations with LGE imaging between January 2005 and November 2019 was performed. Patients with pulmonary atresia and major aortopulmonary collateral arteries and those with significant imaging artifacts were excluded. Demographic and clinical data were abstracted from electronic medical records. The hospital’s Committee on Clinical Investigation waived the requirement for informed consent.

All CMR examinations were performed using a whole-body 1.5 T CMR scanner (Achieva, Philips Healthcare, Best, The Netherlands) using standard imaging techniques recommended by the Society for Cardiovascular Magnetic Resonance [19]. Ventricular volumes and ejection fraction, and pulmonary regurgitation fraction were calculated using standard techniques as previously described by our laboratory [20, 21]. LGE imaging was performed 10–15 min after intravenous administration of a gadolinium-based contrast agent (0.15–0.2 mmol/kg of gadopentate dimeglumine or gadobutrol) using a 2-D phase-sensitive inversion-recovery prepared, fast-gradient echo sequence in ventricular long- and short-axis planes using the following imaging parameters: slice thickness 6–8 mm, slice gap 0–2 mm, reconstructed in-plane spatial resolution between 1.4 × 1.4 mm and 1.7 × 1.7 mm. The short-axis imaging planes were prescribed to cover both ventricles from base to apex. Image analyses were performed using commercially available software (cvi42, v 5.10.1, Circle Cardiovascular Imaging, Calgary, Alberta, Canada).

The extent of RV LGE was determined using 2 techniques (Fig. 1): a semiquantitative RV LGE score, and quantitative RV LGE extent expressed as % of RV mass as follows:

Comparisons among independent groups were performed using Student’s t-test, analysis of variance, Wilcoxon rank-sum test or Kruskal–Wallis test, or a Fisher exact test, as appropriate. Spearman correlation was used to estimate the magnitude of association between LGE score and % LGE extent. To evaluate whether change scores in CMR outcomes differed from zero, a one-sample t-test and Wilcoxon signed-rank test were used. Mixed effects linear regression (fixed time, random subject) was used to evaluate the rate of change in RV LGE over time. For this study, assuming a standard deviation of 2.0 for the change in RV LGE score between 2 CMRs and a 2-sided 0.05-level test, with 40 subjects in each of 2 independent groups, there is 80% power to detect a mean group difference of 1.0 between the mean RV LGE change scores. Intra-rater and inter-rater reliability were assessed using the intra-class correlation coefficient (ICC). A p-value of less than 0.05 was considered significant. Analyses were performed using SAS (version 9.4, SAS Institute, Inc., Cary, North Carolina, USA) and R (version 3.5.1, R Foundation for Statistical Computing, Vienna, Austria).

Patient characteristics are summarized in Table 1. A total of 269 CMR examinations were analyzed on 127 subjects with a median follow-up duration of 4.0 years (IQR 2.1,5.9) between initial and latest CMR examinations. Among the study cohort, 112 subjects (88%) had 2 CMR examinations, and 15 (12%) had 3 CMR examinations available for analysis. The study sample consisted of children and adults, a majority of whom had a transannular patch during their initial surgical repair, which was performed most commonly during the first year of life. 84/127 subjects had no interventions between serial CMRs (Group 1), while 43/127 subjects had transcatheter or surgical intervention between CMRs (Group 2). The types of interventions performed in Group 2 are summarized in Table 1.

RV LGE was present in all patients on the first CMR examination in both groups. The location of LGE was a basal superior segment in 127 subjects (100%), at the mid superior segment in 82 (65%), and at the mid anterior segment in 16 subjects (13%) (Fig. 3). There was good correlation between RV LGE score and % RV LGE extent in both groups (r = 0.63 in Group 1, r = 0.80 in Group 2; p < 0.001 for both). Intra-observer reliability in measuring % RV LGE extent was good (ICC 0.83) in a subset of 40 CMR examinations that were reanalyzed at a later date by the same blinded observer (KS). Inter-observer reliability was modest (ICC 0.6), assessed in a subset of 20 CMR examinations by a blinded second observer (AP).

Serial changes in RV LGE score and % RV LGE extent are summarized in Tables 2 and 3 and in Figs. 2 and 3. There was no increase in RV LGE over time, with a small decline in both LGE parameters over time in both patient groups. Mean 5 yeaar changes in both parameters from linear mixed-effects models are summarized in Table 3. As seen in Fig. 2, RV LGE score and % RV LGE extent increased over follow-up in a small minority of patients, and remained stationary or declined in most patients.

RV LGE score and % RV LGE extent showed a similar serial trend of gradual decline over serial CMRs in both Groups 1 and 2, with similar 5 yeaar rates of decline (Table 3). As seen in Fig. 1, in Group 1, no patient showed a serial increase in % RV LGE extent, while a small minority showed a slight increase in the RV LGE score. In Group 2, although the overall trend was similar to that in Group 1, a somewhat larger number of patients showed an increase in % RV LGE extent and RV LGE score. In this group, the only factor associated with an interval increase in both % RV LGE extent and LGE score (n = 6) was older age at repair (p = 0.009, supplemental Table 1).

We explored the hypothesis that a serial increase in RV end-diastolic volume indexed to body surface area (RVEDVI) with a corresponding increase in RV mass may lead to a serial decline in RV LGE score or % RV LGE extent. However, we found no significant association between the rate of change in either parameter of RV LGE, and the rate of change in RVEDVI.

The patterns of RV LGE in our cohort were similar to those previously reported, with the most common site for LGE being the RV outflow tract. [6] Several prior studies have identified the presence and extent of RV LGE as a risk factor for adverse long-term outcomes in TOF patients [2, 12, 13, 22]. However, the serial rate of change in the RV LGE extent has not been previously studied. Ylitalo et al. [23]. found in a cross-sectional study that the extent of RV LGE is higher in patients with a longer duration of follow-up and they hypothesized that LGE extent increased over time. However, the results of our longitudinal study suggest a lack of significant progress in RV LGE extent during a median follow-up period of 4 years. Further, we found no correlation between the extent of RV LGE and changes in RV volumes during follow-up. It is possible that the finding of a higher LGE extent in patients with a longer duration of follow-up in prior studies may reflect a higher extent of RV surgical scarring, possibly related to an older surgical era. The strengths of the current analysis are its longitudinal design and the quantitative assessment of RV LGE using 2 methods by a single observer.

In the present analysis, we evaluated 2 groups of patients—those free of intervention between serial CMRs and those who had surgical or transcatheter intervention between serial CMRs. We analyzed these groups separately, since surgical manipulation of the outflow tract, as well as artifact caused by implanted valves could potentially confound the serial assessment of LGE extent. The findings of a similar lack of progression in RV LGE in both groups suggests that RV LGE in TOF patient possibly represents fixed post-surgical scarring that is not significantly modulated by long term volume loading due to pulmonary regurgitation, or by intervening procedures, at least in the intermediate term (median follow-up duration was 4 years).

There are inherent difficulties in quantifying RV LGE due to the thin free wall. We used a previously described method for calculation a semiquantitative RV LGE score, in addition to a new method to quantify RV LGE extent [8]. In using the second method for quantifying RV LGE extent, we avoided techniques that automatically designate areas of LGE using a pre-defined standard deviation cutoff relative to normal myocardium, as this can be fraught with error in the thin-walled RV. Instead, we manually identified areas of RV LGE and then quantified the extent of RV LGE in these areas. We found good correlation in the extent of RV LGE calculated using both methods and longitudinal trends in RV LGE were also similar using both techniques. We found good inter-rater and modest inter-observer reliability in quantifying % RV LGE extent. Good inter-rater reliability in assessing the RV LGE score has been previously reported [8].

Our results have important clinical implications. Since we did not observe a significant increase in RV LGE over a median follow-up duration of 4 years, this suggest that RV LGE does not progress rapidly. Therefore, frequent serial assessment of LGE may not be necessary, thereby avoiding the frequent administration of gadolinium-based contrast agents in these patients, in light of several reports of long-term retention of gadolinium in patients who receive multiple doses of contrast [17, 18]. It should be noted that our findings apply to patients in a more recent surgical era.