The prognostic value of T1 mapping and late gadolinium enhancement cardiovascular magnetic resonance imaging in patients with light chain amyloidosis

Background Cardiac impairment is associated with high morbidity and mortality in immunoglobulin light chain (AL) type amyloidosis, for which early identification and risk stratification is vital. For myocardial tissue characterization, late gadolinium enhancement (LGE) is a classic and most commonly performed cardiovascular magnetic resonance (CMR) parameter. T1 mapping with native T1 and extracellular volume (ECV) are recently developed quantitative parameters. We aimed to investigate the prognostic value of native T1, ECV and LGE in patients with AL amyloidosis. Methods Eighty-two patients (55.5 ± 8.5 years; 52 M) and 20 healthy subjects (53.2 ± 11.7 years; 10 M) were prospectively recruited. All subjects underwent CMR with LGE imaging and T1 mapping using a Modified Look-Locker Inversion-recovery (MOLLI) sequence on a 3 T scanner. Native T1 and ECV were measured semi-automatically using a dedicated CMR software. The left ventricular (LV) LGE pattern was classified as none, patchy, and global groups. Global LGE was considered when there was diffuse, transmural LGE in more than half of the short axis images. Follow-up was performed for all-cause mortality using Cox proportional hazards regression analysis and Kaplan-Meier survival curves. Results The patients demonstrated an increase in native T1 (1438 ± 120 ms vs. 1283 ± 46 ms, P = 0.001) and ECV (43.9 ± 10.9% vs. 27.0 ± 1.7%, P = 0.001) compared to healthy controls. Native T1, ECV and LGE showed significant correlation with Mayo Stage, and ECV and LGE showed significant correlation with echocardiographic E/E’ and LV ejection fraction. During the follow-up for a median time of 8 months, 21 deaths occurred. ECV ≥ 44.0% (hazard ratio [HR] 7.249, 95% confidence interval (CI) 1.751–13.179, P = 0.002) and global LGE (HR 4.804, 95% CI 1.971–12.926, P = 0.001) were independently prognostic for mortality over other clinical and imaging parameters. In subgroups with the same LGE pattern, ECV ≥ 44.0% remained prognostic (log rank P = 0.029). Median native T1 (1456 ms) was not prognostic for mortality (Tarone-Ware, P = 0.069). Conclusions During a short-term follow-up, both ECV and LGE are independently prognostic for mortality in AL amyloidosis. In patients with a similar LGE pattern, ECV remained prognostic. Native T1 was not found to be a prognostic factor.


Background
Immunoglobulin light chain (AL) type amyloidosis is characterized by monoclonal plasma cells and the deposition of insoluble fibrils formed by immunoglobulin light chains in various organs [1]. In approximately two-thirds of AL-type amyloidosis patients there is cardiac impairment at diagnosis, which is a major contributor to mortality [2]. Thus, early identification and risk stratification is of vital importance for timely clinical intervention that may improve the patients' prognosis. Current predictors of survival, such as serum biomarkers [3][4][5], electrocardiogram (ECG) [6], cardiac morphology and functional parameters [7][8][9][10] rely on measuring surrogates rather than direct markers of interstitial expansion.
Cardiovascular magnetic resonance (CMR) imaging with late gadolinium enhancement (LGE) is the most commonly performed non-invasive protocol for myocardial tissue characterization in a wide spectrum of cardiomyopathies. A typical pattern of global, predominately subendocardial LGE, serves not only as a diagnostic marker for cardiac AL amyloidosis but also as a prognostic marker for mortality [11][12][13][14]. However, because the recognition of LGE lesions involves delineation of abnormal tissue from normal tissue, early identification of mild cases can easily be missed in cardiac AL amyloidosis and other diffuse infiltrative cardiomyopathies [15][16][17].
Myocardial CMR T1 mapping methods are used for native (i.e., without use of gadolinium-based agents) and for post-contrast T1 measurements. In combination with the hematocrit, pre-and post-contrast measurements enable the quantification of the extracellular volume fraction (ECV). Native myocardial T1 values reflect a composite signal from both the intracellular (predominantly myocytes) and extracellular compartments [18][19][20]. Previous studies have shown that different sequences and field strengths yielded different native T1 and ECV values [21][22][23][24]. To date, only one study has examined the utility of a shortened Modified Look-Lockers Inversion-recovery (shMOLLI) sequence at 1.5 T to assess the prognostic value of native T1 and ECV in AL amyloidosis [25]. However, this study did not concurrently assess the utility of LGE in this population. In the present study, we examined a Chinese population with AL amyloidosis using a 3 T scanner with a MOLLI sequence and compared the prognostic value of T1 mapping parameters with LGE. This method of analysis of the prognostic values of native T1 and ECV for mortality in AL amyloidosis and its comparison with LGE have not been reported previously.

Study subjects
This prospective study was approved by the Institutional Ethnics Committee for Human Research at Peking Union Medical College Hospital (Beijing, China). All participants were required to provide written informed consent prior to recruitment. AL amyloidosis patients who were referred for CMR imaging at Peking Union Medical College Hospital between August 1, 2014 and August 31, 2016 were included in the study. Approximately 20% of the patients who had contraindications either to CMR imaging (i.e., CMRincompatible devices) or contrast administration (i.e., estimated glomerular filtration rate < 30 ml/min) were excluded.
Eighty-two AL amyloidosis patients (55.5 ± 8.5 years; 52 male) were consecutively recruited. All patients had biopsy evidence of AL amyloidosis with positive Congo red stain and light chain deposition confirmed by immunohistochemistry, immunofluorescence or mass spectrometer. The assays were performed in the tissues listed as follows: kidney (n = 29), myocardium (n = 19), bone marrow (n = 7), fat (n = 7), tongue (n = 7), liver (n = 4), upper gastrointestinal tract (n = 3), buccal mucosa (n = 3), lung (n = 1), rectum (n = 1) and skin (n = 1). All patients underwent laboratory examination of the cardiac biomarkers Troponin I (cTnI) and N-terminal pro-B-type natriuretic peptide (NT-proBNP), serum immunoglobulin free light chain difference (dFLC) at baseline and were categorized based on revised Mayo Stage published in 2012 [5]. All patients underwent transthoracic echocardiography (TTE) at baseline and the E: E′ ratio and E: A ratio were calculated to assess the left ventricular (LV) diastolic function. A hematologist and a cardiologist, both of whom were blinded to the results of CMR imaging, recorded the results of Mayo Stage and TTE, respectively.
Twenty healthy subjects (53.2 ± 11.7 years; 10 male) with normal CMR imaging results were recruited, who had neither history nor symptoms of cardiovascular disease or diabetes mellitus.

CMR scanning protocol
CMR was performed on a 3 T whole-body scanner (MAG-NETOM Skyra, Siemens Healthineers, Erlangen, Germany). The system is capable of operating at a maximum slew rate of 200 mT/m/ms and a maximum gradient strength of 45 mT/m. An 18-element body matrix coil and a 32-element spine array coil were used for data acquisition. A four-lead vector cardiogram was used for ECG gating.

CMR image analysis
CMR images were independently analyzed by two experienced radiologists. The LV LGE pattern was classified into three groups referred to Araoz Criteria [11] and Moon Criteria [12]: No LGE, when there were no areas of LGE; Patchy LGE, when there were discrete areas of LGE, or there were diffuse areas of LGE in less than half of the short axis images; Global LGE, when there was diffuse, transmural LGE in more than half of the short axis images. Discrepancies were resolved in consensus during a joint evaluation with a third radiologist.
Cardiac function, native T1 and ECV were measured semi-automatically using a dedicated CMR software cvi42 (version 5.3, Circle Cardiovascular Imaging, Calgary, Canada). Standard parameters of cardiac structure (i.e., inter-ventricular septum thickness, ventricle volume, LV mass and left atrium area with indexing for body surface area) and ventricle ejection fraction were measured by contouring the endocardium and epicardium on long-axis and short axis cine images at the end-systolic and enddiastolic stage. Native T1 and ECV of the 16 American Heart Association (AHA) segments and global LV were measured, by contouring the endocardium and epicardium and indicating the inter-ventricular septum on pre-contrast and post-contrast T1 mapping images with indexing for the hematocrit. Global LV native T1 and ECV were used for further analysis. The average values of native T1 and ECV measured by the two radiologists were used.

Clinical follow-up
A physician blinded to the results of CMR imaging conducted the telephone and clinical follow-up each month. Unless the outcome was death from any cause, patients were censored at the end of the study. If patients were lost to follow-up, their last clinic visit record was used. A follow-up CMR scan was performed after a complete standard course of chemotherapy, with an interval of about approximately one year.

Statistical analysis
Statistical analysis was performed using SPSS Statistics (version 21.0, International Business Machines, Inc., Armonk, New York, USA) and R programming language for statistical computing (version 3.0.1, The R Foundation for Statistical Computing). The agreement between two observers was assessed using the interclass correlation coefficient. Correlation between native T1 and ECV with continuous variables or categorical variables was assessed using the Pearson's r correlation or Spearman ρ correlation, respectively. Comparison between groups and the control was performed by one-way analysis of variance (ANOVA) with post-hoc Bonferroni correction. Statistical significance was defined as P < 0.05.
Survival was evaluated with Cox proportional hazards regression analysis, providing estimated hazard ratios (HR) with 95% confidence intervals (CI) and Kaplan-Meier curves. All variables were first analyzed with univariate Cox regression. Multivariate models were then used to evaluate the independent prognostic value of native T1, ECV or LGE above other clinically and statistically significant covariates. The median value of native T1 and ECV was used as cut-off values. The Harrell's C statistic was calculated for different models. Table 1 summarizes the characteristics of AL amyloidosis patients and healthy controls at baseline. At the time of CMR scanning, 9 (11%) patients had received triple chemotherapy for the first time with thalidomide or bortezomib, cyclophosphamide and dexamethasone (BCD or TCD), 2 (2%) had received autologous stem cell transplant (ASCT) and 71 (87%) had not received any chemotherapy. During the follow-up, 59 (83%) untreated patients received standardized treatment with chemotherapy or ASCT, and the rest did not receive any chemotherapy because of the expense or for personal or other reasons. At the time of last follow-up, 61 (74%) patients were alive, with a survival probability of approximately 75.6% at median follow-up time (8 months). Two patients were lost to follow-up. The follow-up time of one patient (female; 52 years; Mayo Stage, III; LVEF, 52.5%; native T1, 1575 ms; ECV, 51.4%; LGE pattern, global) was 5 months, and the other (male; 68 years; Mayo Stage, II; LVEF, 55.5%; native T1, 1512 ms; ECV, 41.6%; LGE pattern, global) was 18 months.

Clinical and biochemical markers of severity
All continuous variables were normally distributed (Kolmogorov-Smirnov test) and presented as the mean ± SD, except for cTnI, NT-proBNP and dFLC, which were log transformed for bivariate testing and presented as medians (quartiles 1-quartiles 3). As shown in Tables 1, 30 (37%),
Eight patients (Mayo stage I/II/III/IV, 1/1/3/3; no/ patchy/extensive LGE, 1/3/4) underwent follow-up CMR scans. The median interval between baseline and followup CMR scans was 12 months. All subjects completed a standard course of BCD chemotherapy and achieved a complete response (CR) or very good partial response (VGPR). For the patient with no LGE, the dynamic changes of LGE, native T1 and ECV are shown in Fig. 6. Another patient showed a significant regression of LGE as well as decreases of native T1 (1658 ms to 1490 ms) and ECV (62.7% to 51.4%). The other 6 patients showed no prominent progressions or regressions of LGE, and different trends of native T1 and ECV (increases in 2 patients, decreases in 2 patients, and no significant changes in 2 patients).

Discussion
In this study, we examined the prognostic value of CMR ECV, LGE and native T1 in a Chinese population with AL amyloid. To the best of our knowledge, this is the first study to concurrently assess the prognostic value of T1 mapping parameters with LGE in AL amyloid. Our findings indicate that, while ECV and LGE functioned as independent prognostic factors for mortality in AL amyloid patients, native T1 did not display prognostic value. We also showed that in subgroups with the same LGE pattern, ECV remained prognostic.
We found AL amyloid patients with no LGE demonstrated increased native T1 and ECV, highlighting the importance of native T1 and ECV over LGE in early   LGE Late gadolinium enhancement, ECV Extracellular volume detection of myocardial involvement in this disorder. In agreement with other studies [11,12], we showed that global LGE prognostic for mortality. We have also included a novel finding that subgroups with the same LGE pattern displayed ECV as a significant prognostic factor. LGE is the classic and most commonly performed CMR protocol for myocardial tissue characterization, and a typical pattern of global LGE serves as both diagnostic marker for cardiac AL amyloid and a prognostic marker for mortality [11][12][13][14]. However, early identification of mild cases of cardiac AL amyloid and other diffuse infiltrative cardiomyopathies are easily missed, [15,16] since the basis of LGE lesion identification involves demarcating the abnormal tissue amidst normal tissue. It is better to perform T1 mapping scanning together with LGE scanning in AL amyloid patients, for native T1 and ECV provide additional diagnostic and prognostic information.
The current study is also the second overall study that focuses on the prognostic value of T1 mapping parameters in AL amyloid. A previous study demonstrated the prognostic value of native T1 and ECV for mortality using a 1.5 T scanner with a shMOLLI sequence [25], but LGE was not assessed. In this study, using a 3 T scanner with a MOLLI sequence and found that, regardless of disease course and therapy status, ECV was an independently prognostic factor for mortality with a similar cut-off value as the previous study.
Moreover, we also found that native T1 did not act as a prognostic factor, which is controversial with previous study. Previous studies have shown variations in ECV values using different scanning sequences including MOLLI and shMOLLI [24], and variations in native T1 values with different equipment manufacturers, scanning sequences and undefined physiological status of the patients [21][22][23]. Despite the emerging importance of T1   ECV calculation requires the administration of an IV contrast agent. However, renal function impairment is often seen in AL amyloid patients, since the kidney is one of the most commonly involved organs [1,2]. This and other contraindications for the application of contrast agents may limit the use of ECV in this population. In this situation native T1 in combination with LVEF and inter-ventricular septum thickness seems to be the second best approach to detect diffuse myocardial involvement. Our data show that native T1 is not as prognostic as ECV, but still more sensitive than LGE for myocardial involvement in AL amyloid.
Our study has several limitations. One is the short follow-up with a median time of 8 months and a relatively low event proportion of 25.6%. Another limitation is that we do not have additional parameters to fully characterize the diastolic function. The third limitation is about the therapy status of the patients at baseline and during the follow-up, given the cardiotoxic effects of chemotherapy agents may be confounding factors. Besides, we found increased native T1 and ECV values in patients with no LGE, but only one such patient underwent myocardial biopsy verifying the result. Thus ours, like most studies in this area, suffer from the lack of diagnostic pathology.

Conclusion
For myocardial tissue characterization, while LGE is a classic and most commonly performed parameter, ECV is a recently developed quantitative CMR parameter. The current study is the first to compare the prognostic value of T1 mapping parameters with LGE in AL amyloid. During a short follow-up interval, we showed that both ECV and LGE were promising prognostic factors for mortality in AL amyloid. Further, in patients with the same LGE pattern, ECV remained prognostic, suggesting the merit of using T1 mapping scanning in conjunction with LGE in this population. Native T1, however, was found to be not as equally prognostic as ECV or LGE. Thus, for suspected AL amyloid patients without contraindications, it is better to perform contrast enhancement scanning.