Incremental prognostic value of global longitudinal strain to the coronary microvascular resistances in Takotsubo patients

Background: Global longitudinal strain (GLS) allows an accurate assessment of left ventricular function with prognostic value. We aimed to evaluate whether the assessment of GLS in the acute phase of Takotsubo syndrome (TTS) provides incremental prognostic value to the degree of impaired microvascular resistance (MR) in TTS patients at 1-year follow-up. Methods: We recruited patients admitted for TTS who underwent cardiac angiography and echocardiography from January 2017 to June 2020. Left anterior descending coronary artery non-hyperaemic angiography-derived index of microcirculatory resistance (LAD NH-IMRangio) was calculated. NT-proBNP, high-sensitive cardiac troponin T (hs-cTnT), left ventricular ejection fraction (LVEF) and GLS were measured at admission. Major adverse cardiac events (MACE) were defined as the composite of cardiovascular death, repeat hospitalizations for heart failure (HF) and acute myocardial infarctions. Results: 67 patients had both GLS and NH-IMRangio available and were included in the study. Median age was 75.2 years and 88% were women. Rate of MACE at 1-year was 13.4%. Kaplan-Meier curves showed higher rates of MACE at 1-year in patients with both higher LAD NH-IMRangio and GLS values compared with those with higher LAD NH-IMRangio and lower GLS values (33.3% vs. 11.1%; p = 0.049). NT-proBNP levels at admission and the recovery of LVEF were correlated with GLS values while MR and hs-cTnT were not. Conclusion: GLS provides incremental prognostic value to the degree of impaired MR in TTS patients. The combination of a poorer GLS with a higher degree of impaired MR was associated with a higher rate of MACE in these patients.


Introduction
Takotsubo syndrome (TTS) is an acute and transient cardiomyopathy that primarily affects postmenopausal women. The presence of left ventricular wall motion abnormalities in absence of obstructive epicardial coronary artery disease has been associated with an impaired microvascular function due to catecholaminergic toxicity [1][2][3]. Recently, we have reported the prognostic value of an impaired microvascular resistance (MR) in TTS patients using non-hyperaemic angiography-derived index of microcirculatory resistance (NH-IMRangio) [4], a novel technique to assess coronary microvascular dysfunction (CMD) [5][6][7][8].
patients. In addition, we also evaluated its correlation with cardiac biomarkers such as NT-proBNP and high sensitivecardiac troponin T (hs-cTnT).

Study population
We conducted a retrospective, observational, single-center study which recruited all consecutive patients admitted for TTS from January 2017 to June 2020 in a tertiary center in Barcelona (Spain). All inclusion criteria had to be met: a) ≥ 18 years of age, b) diagnosis of TTS according to modified Mayo Clinic criteria [19], c) performance of an echocardiogram within the first 6 h after admission to the emergency room and a coronary angiography (CAG) in the first 24 h of the onset of the symptoms, and d) signed informed consent. We excluded those patients who presented: (a) history of coronary artery bypass grafting, (b) newly diagnosed coronary artery disease in the same territory of the regional wall motion abnormality, and (c) patients in atrial fibrillation.
The study was performed in accordance with the standards set by the "Declaration of Helsinki" and was approved by the Clinical Research Ethics Committee.

Study variables
Patient's demographics, cardiovascular risk factors, and electronical clinical history were collected from medical reports at admission and discharge. Analytical parameters were registered from the first blood test or arterial blood gas samples at admission. NT-proBNP and high-sensitivity cardiac troponin-T (hs-cTnT) were measured by electrochemiluminescence immunoassays on a Cobas e601 platform (Roche Diagnostics, Switzerland). NT-proBNP range was 5-35.000 pg/ml with a coefficient of variation (CV) ≤ 3.5%, whereas the hs-cTnT range was and 3-10.000 ng/L with a CV ≤ 4.0%. The Modification of Diet in Renal Disease Study equation (MDRD) [20] was used to calculate the estimated glomerular function rate (eGFR). We calculated the LVEF by echocardiography using the biplane Simpson method. Treatments and/or procedures performed during hospital stay were also registered.

Global longitudinal strain calculation
Echocardiographic images were obtained using commercially available CX50 or Sparq-DS echocardiographic systems (Koninklijke Philips Electronics N.V. 2019) equipped with a S5-1 (Purewave) 1-5 MHz sectorial cardiac transducer. Echocardiography was performed according to European Association of Cardiovascular Imaging recommendations [21]. Echocardiographic data were digitally recorded in cine loop format and 2D strain analysis were performed offline using 2D strain imaging software (QLAB Advanced Quantification Software 13.0, Koninklijke Philips Electronics N.V. 2019) by one investigator who was blinded to the clinical data. Software automatically traced the endocardial borders at the end of the systole with further manual adjustments if necessary to optimize automated speckle tracking. An 18-segment model was presented and average GLS was calculated as the mean of all segments. Moreover, LV was divided into three slices to calculate basal, mid-ventricular and apical longitudinal strain (LS) as the average value of six segments of each one. GLS and LS presented negative values due to the ventricular shortening during systole.

3D-QCA, QFR and NH-IMRangio calculation
We analyzed the state of the coronary microcirculation by assessing the left anterior descending coronary artery (LAD) NH-IMRangio value. Two certified readers performed the 3D-QCA analysis and the QFR computation in the CoreLab of the MedStar Washington Hospital Center using the QAngio XA 3D software package (Medis Suite 3.2.48.8, Medis, Leiden, Netherland). A proximal and a distal point were established in two angiographic projections > 25º apart. The software automatically reconstructed a 3D model of the LAD without its side branches and performed the 3D-QCA analysis and the QFR computation. Mean aortic pressure during CAG and TIMI frame count were also recorded for the calculation of the LAD NH-IMRangio according to the formula used in the validation studies [5,7].
Nframes (rest) frame adquision rate Impaired MR was defined as a value of LAD NH-IMRangio ≥ 25. Although there is no established specific cut-off point to define an impaired MR in TTS patients, we consider that the widely spread cut-off point of 25 could be suitable for this population [22].

Follow-up and outcomes
The primary endpoint of this study was the rate of major adverse cardiac events (MACE) at 1-year follow-up. MACE included cardiovascular death, repeat hospitalization for HF [23], and acute myocardial infarction (AMI). Cardiovascular death was defined as any death secondary to AMI, HF, sudden cardiac death, stroke, cardiovascular procedure, cardiovascular hemorrhage or other cardiovascular causes. AMI was defined according to the fourth definition of AMI [24]. Repeat hospitalization for HF was defined as ≥ 2 hospitalizations for HF that occurred from the first month after hospital discharge.
A secondary endpoint (composite endpoint) was assessed which included cardiovascular death, HF events, AMI and readmissions for symptomatic arrhytmias. HF events included any emergency department visit or hospitalization for HF as well as any urgent or unscheduled outpatient office visits with a primary diagnosis of HF, where the patient exhibited new or worsening symptoms of HF on presentation and received initiation or intensification of treatment specifically for HF. [25]. Readmissions due to symptomatic arrhythmias included any hospitalization for symptomatic atrial fibrillation, atrial flutter, ventricular tachycardia, or advanced atrioventricular block. Follow-up and the adjudication of MACE and composite endpoint were performed by the study investigators reviewing the patients' medical records through the territorial health network and with phone calls, if necessary. All patients completed the 1-year follow-up period.
A follow-up echocardiogram was performed between the third and sixth month after hospital discharge. To assess the recovery of LVEF the delta LVEF was calculated. Delta LVEF was defined as the difference between LVEF on follow-up echocardiogram and LVEF on admission echocardiogram. LVEF reduced at follow-up was defined as LVEF < 55% on follow-up echocardiogram.
Other secondary endpoints were: (a) to study the relationship between CMD and GLS in TTS patients, (b) to investigate the correlation between GLS and cardiac biomarkers at admission (NT-proBNP and hs-cTnT).

Statistical analysis
Results are presented as the mean (standard deviation) for continuous variables with a normal distribution, median (interquartile range) for continuous variables with a non-Gaussian distribution, and with counts and percentages in case of categorical variables. We divided our cohort in 4 groups according to GLS and NH-IMRangio median values (≥-11.24% and ≥ 40.94, respectively). We used the x 2 test or Fisher exact test for comparing categorical variables. For continuous variables, they were analyzed by t-test or ANOVA in the case of a normal distribution and by Mann-Whitney U-test or Kruskall-Wallis test in the case of a nonnormal distribution.
Correlations were analyzed using Pearson if variables presented a gaussian distribution or Spearman if they presented a non-gaussian distribution.
We performed Kaplan-Meier survival curves to compare 1-year MACE rates in groups divided as a function of the median value of GLS and NH-IMRangio, using the long-rank test to compare the rates of MACE and composite endpoint.
Significance level was set at p < 0.05. All statistical analyses were performed using Stata 13.0 for Windows.

Baseline characteristics
We registered 78 patients with TTS during the study period. We calculated both the GLS and the LAD NH-IMRangio in 67 of them (6 patients were excluded due to impossibility of calculating NH-IMRangio in the LAD for calibration failure and 5 patients due to impossibility of calculating GLS). No significant differences were found between patients without GLS and / or LAD NH-IMRangio available (Table S1 in supplemental material). Almost 90% of the patients were women with a median age of 75.2 years. The classical pattern with apical involvement was the most prevalent wall motion abnormality while secondary forms of TTS were around 35%. Chest pain (65.7%) was the most frequent symptom at admission while dyspnea (52.2%) was more frequent in patients with more impaired GLS values (p: 0.007). More than 45% of the patients were admitted with signs of HF and 20% of our cohort needed some kind of mechanical ventilation. Fifty-eight patients (86.6%) showed a LAD NH-IMRangio value ≥ 25.
Baseline characteristics of the study population according to median values of LAD NH-IMRangio and GLS are detailed in Table 1.

Outcomes at 1-year follow-up based on GLS and NH-IMRangio values
The rate of MACE at 1-year follow-up was 13.4% (9 events), mainly due to repeated hospitalization for HF (6 patients). Cardiovascular mortality was 3% (2 patients, one died of HF and the other had a sudden cardiac death). Only one patient presented an AMI during follow-up (non-ST elevation AMI). All-cause mortality at 1-year follow-up was 4.5% (3 patients). Rate of composite endpoint was 23.9%, mainly due to HF events (12 events Fig. 1. percentage of MACE at 1-year of follow-up, while patients with lower GLS and LAD NH-IMRangio values (both below the median) showed the lowest rate of MACE (p = 0.006). No differences were found in follow-up LVEF, delta LVEF or medication at discharge between groups.

Prognostic value of GLS and LAD NH-IMRangio
Patients who developed MACE during the 1-year followup presented a worse LVEF and higher NT-proBNP values at admission. LAD NH-IMRangio values were higher in patient with MACE but GLS values did not differ between groups. No differences were found in medication at    (2) 0.199 Continuous variables are expressed as median (IQR) and categorical data as %. kg/m2: kilograms per meter squared; TTS: Takotsubo syndrome; CAD: Coronary artery disease; GLS: global longitudinal strain; SBP: systolic blood pressure; bpm: beats per minute; msec: milliseconds; mmol/L: millimoles per liter; g/L: grams per liter; hs-TnT: high-sensitive Troponin T; ng/L: nanograms per liter; eGFR: estimated glomerular filtration rate; ml/min/1.73m2: milliliters per minute per 1.73 m squared; pg/ml: picograms per milliliter; NH-IMRangio: non-hyperaemic angiography-derived index of microcirculatory resistance; LAD: left anterior descending artery; IQR: interquartile range coronary microvascular resistance (MR) together with a worse GLS were associated with a higher rate of MACE and composite endpoint at 1-year of follow-up (Fig. 3), (2) LS was globally impaired in patients with TTS with worse values in apical segments, and (3) NT-proBNP at admission and the recovery of LVEF, but not MR and hs-cTnT, were correlated with GLS values.
First, a reduced LVEF on admission has been shown to have prognostic implications in patients with TTS [9, 26,27]. Assessment of the myocardial function of the left ventricle using GLS is a more accurate evaluation of the recovery of the myocardial abnormalities in these patients. Alashi et al. [18] reported that GLS could have prognostic value incremental to LVEF in the long-term prognosis in TTS patients, while Dias et al. [17] showed the prognostic value of GLS for in-hospital mortality in these patients. In our study, we found that patients who had simultaneously NH-IMRangio and GLS values above the median population values developed a higher rate of MACE during the

Correlations between GLS with biomarkers, delta LVEF, and LAD NH-IMRangio
The GLS values presented a moderate positive correlation with NT-proBNP levels at admission (rho: 0.59; p < 0.001) and we also found a trend to a mild positive correlation between GLS with hs-cTnT (rho: 0.23; p = 0.063). On the other hand, a moderate negative correlation was found between GLS values and delta LVEF (rho: -0.40; p = 0.022) while there was a trend to a mild negative correlation between values of GLS and LAD NH-IMRangio values in TTS patients (rho: -0.21; p = 0.0829) (Fig. 2).

Discussion
As far as we know, this is the first study that investigates the prognostic value and the relationship between GLS and microvascular resistances in TTS patients. The main findings of our study were: (1) the presence of a more impaired  a trend to a mild positive correlation between hs-cTnT levels and GLS values. The release of NT-proBNP is associated with the degree of LV wall stress while hs-cTnT is a marker of myocardial injury. Thus, a worse myocardial contractility during the acute phase of TTS would favor a higher release of both cardiac biomarkers [33][34][35]. On the other hand, although we did not find a significant correlation between NH-IMRangio and GLS values, we observed a trend towards worse GLS values in those patients with less impaired MR, suggesting that ventricular dysfunction in TTS patients may not be only due to myocardial damage. However, as mentioned, these data are of hypothesisgenerating value at this stage.

Limitations
The main limitation of our study is the small number of patients from a single center. However, to our knowledge, this is the first study analyzing the incremental prognostic value of GLS to the degree of impaired MR in TTS patients. Secondly, in patients who were not referred directly to the cath lab (STEMI code protocol), echocardiography was not performed at the same time as the CAG, so this would carry a bias for the correlation of both measurements. In addition, echocardiography was performed using a portable ultrasound machine in the cath lab or in the emergency room, which could lead to a poorer image quality and, in turn, a possible bias in GLS analysis. Nevertheless, we believe 1-year follow-up than those with one or both values below the median value. Thus, in our cohort GLS provides incremental prognostic value to the degree of impaired MR in TTS patients. Some authors have reported that myocardial dysfunction in TTS could be not only secondary to catecholaminergic toxicity and CMD but also a "protective mechanism" to avoid further damage by the catecholaminergic surge. It has been hypothesized that in the myocardium, the switch from β2-adrenergic receptor Gs coupling to β2-adrenergic receptor Gi coupling in response to the catecholamine surge could limit the induction of apoptotic pathways but causing a negative inotropic effect [28,29]. Interestingly, we found a trend towards a mild negative correlation between LAD NH-IMRangio and GLS that would support the hypothesis that part of the ventricular dysfunction in TTS patients could be a mechanism to minimize myocardial damage during the acute phase of the disease. Nevertheless, these data are only hypothesis generating and would require a multicenter, prospective validation.
Furthermore, we found an alteration of the LS longitudinal strain in the three LV slices (basal, mid ventricular and apical) with the apical region showing the worst LS values. Our results are aligned with previous studies that reported a global LV involvement despite the visual appearance of basal hypercontractility in TTS patients, being the apical GLS the most severely affected [30][31][32].
Finally, in our study, NT-proBNP levels at admission were positively correlated with GLS values while we found that performing bedside echocardiography reflects standard clinical practice and will facilitate its reproducibility by other groups in the future.

Conclusion
GLS provided an incremental prognostic value to the degree of impaired MR in TTS patients. GLS is globally impaired in these patients with the apical segments being the most affected. NT-proBNP but not MR were associated with the GLS values in TTS patients. Future studies will be needed to validate these results and to fully understand the etiopathogenesis of ventricular dysfunction in TTS patients.