Heartbeat-to-heartbeat cardiac tissue characterization
van den Boomen, Maaike
DOI:
10.33612/diss.128413796
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2020
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
van den Boomen, M. (2020). Heartbeat-to-heartbeat cardiac tissue characterization. University of Groningen. https://doi.org/10.33612/diss.128413796
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
der Harst, D.E. Sosnovik, R.J.H. Borra, N.H.J. Prakken – “Native T1reference values for
non-ischemic cardiomyopathies and populations with increased cardiovascular risk: A systematic review and meta-analysis,” Journal of Magnetic Resonance Imaging, 2018;47(4):891-912.
Chapter 2
Native T
1
reference values for non-ischemic
cardiomyopathies and populations with
increased cardiovascular risk
A systematic review and meta-analysis
AbstractThis chapter introduces myocardial native T1-mapping, which is widely used to
assess focal fibrosis in ischemic cardiomyopathies but is also increasingly used to diagnose diffuse fibrosis in non-ischemic cardiomyopathies. However, stud-ies reporting T1-values in healthy and diseased myocardium seem
contradict-ing, particularly in non-ischemic cardiomyopathies (NICM) and populations with increased cardiovascular risk. Therefore, this chapter provides diagnos-tic ranges of native myocardial T1-values in specifically those patients
popu-lations. The standard mean difference (SMD) for patients with hypertrophic cardiomyopathy (HCM), dilated cardiomyopathy (DCM), or myocarditis (MC) were significantly increased compared to healthy controls, which confirmed the diagnostic applicability of native T1-mapping for the assessment of fibrosis and
inflammation. Also, in a small number of studies were performed in iron over-load, amyloidosis and Fabry disease, where the T1-values showed to significantly
increase. Furthermore, in hypertension (HT) patients a distinction could be made between populations with and without left-ventricle hypertrophy (LVH), where only a significant increase in the SMD for HT patients with LVH was observed. These last results indicate that native T1-mapping can detect diffuse
fibrosis in NICM linked to the presence of left ventricular (LV) remodeling. Further research in populations with increased cardiovascular risk is needed to strengthen the diagnostic applicability of T1-mapping, which also helps to
2.1
Introduction
D
iseases defined as non-ischemic cardiomyopathies (NICM) are prevalentdis-eases characterized by different patterns of fibrosis in the myocardium that can eventually cause heart failure. According to the American Heart Association (AHA) and the National Institutes of Health (NIH), NICM comprises a heteroge-neous group of cardiac diseases presented as hypertrophic cardiomyopathy (HCM), dilated cardiomyopathy (DCM), or restrictive cardiomyopathy (RCM) (Wexler et al. 2009). Of these three subgroups HCM alone already affects 1/500 adults (Ommen et al. 2013) and its prevalence increases with age. Some populations also have an in-creased risk of developing NICM, according to the AHA. These populations include the one-third of the USA population that has a high blood pressure (Go et al. 2013), the approximately one-tenth that suffers from diabetes (Centers for Disease Con-trol and Prevention 2014); and the two-thirds that are either overweight (body mass index (BMI)ě25) or obese (BMIě30) (Wang et al. 2012, Flegal et al. 2012).
Early detection of NICM is of key importance in preventing major cardiac events. However, the subtle changes that are expected in the early stages of NICM are dif-ficult to detect and distinguish from normal variation. Cardiac magnetic resonance imaging (MRI) is commonly used to diagnose NICM by imaging standard parame-ters such as left ventricular (LV)-function, -mass, and myocardial fibrosis using late gadolinium enhancement (LGE) (Prakken et al. 2011, Shehata et al. 2008, Noureldin et al. 2012). In the more advanced stages of NICM, cardiac MRI can reveal fibrosis combined with either an increase in LV mass (HCM) or in dilatation of the ventric-ular cavity (DCM) (Prakken et al. 2009). However, in the earlier stages of NICM the increases in mass and dilation are less obvious, and the fibrosis patterns can be difficult to detect. This makes it challenging to recognize NICM at the onset of the disease (Puntmann et al. 2016). It is even more difficult to distinguish NICM from hypertension (HT), type 2 diabetes mellitus (DM) or obesity, because of their similarities in cardiac characteristics (Jellis and Kwon 2014a), especially when left-ventricle hypertrophy (LVH) is present. Some of their common characteristics in-clude: increased left ventricular wall-thickness (Treibel et al. 2015), diastolic dys-function (Wilmot et al. 2014), increased left ventricle mass (Olivotto et al. 2013), and infiltration of myocardial fat (Olivotto et al. 2013). These similarities may lead to incorrect interpretation and possible mistreatment, which stresses the need for ad-ditional diagnostic techniques to ensure accurate diagnosis of NICM.
T1-mapping has been proposed as a technique to aid earlier diagnosis of NICM
T1-mapping can differentiate between healthy myocardial tissue and pathologies
including HCM, myocarditis (MC), iron loading, amyloidosis and Fabry disease
(Germain et al. 2014). In addition, T1-values of myocardial tissue in HT patients
without LVH do not seem to change (Treibel et al. 2015, Kuruvilla et al. 2015), sug-gesting that it might be possible to differentiate the cardiovascular risk population with HT from NICM disease. However, further research is still needed to determine
whether T1-mapping can actually enable the earlier detection of these NICM.
Although there are concerns about the physical accuracy of T1-mapping, the
over-all precision and reproducibility are fairly high and of substantial clinical utility (Kellman et al. 2014). There is therefore currently an increasing demand for
norma-tive reference T1-values (Gai et al. 2011, Hamdy et al. 2016, Moon et al. 2013). These
reference values will be of particular importance for HT, DM and obese patients, be-cause they share cardiac MRI characteristics with NICM (Treibel et al. 2015, Wilmot et al. 2014, Olivotto et al. 2013). Since methodological differences can eventually af-fect the T1-values (Kellman et al. 2014, Moon et al. 2013), a meta-analysis is a suitable
approach to determine the normal myocardial T1-reference values and potential
co-variates.
2.2
Methods
Search Strategy
In June 2017, two reviewers (M. van den Boomen and E.V. Hulleman) systematically and independently searched for eligible studies published since 2011 in PubMed/
MEDLINE and EMBASE using cardiac T1-mapping in humans. The search was
restricted to studies to NICM, cardiac inflammatory or storage diseases and popu-lations with increased cardiovascular risk. Keywords that were used were: “cardio-myopathy”, “hypertension”, “obesity”, “diabetes mellitus”, “magnetic resonance
imaging”, and “T1-mapping” (See Appendix 2.A for full search term). Studies were
included if they met each of the following inclusion criteria:
1. Published results from randomized controlled trials or cohort studies; 2. Investigated human adults;
3. Included subjects with NICM, MC, iron overload, amyloidosis, HT, DM or
4. Contained native T1-values from either or both of the following sequences:
a. Modified Look-Locker Inversion-recovery (MOLLI) (Messroghli et al. 2004, Messroghli et al. 2006, Messroghli et al. 2007);
b. Shortened MOLLI (ShMOLLI) (Piechnik et al. 2010);
5. Excluded subjects with a history of coronary artery disease or myocardial in-farction;
6. Published in peer-reviewed journals, available in full text and written in En-glish.
The Preferred Reporting Items for Systemic Reviews and Meta-Analysis (PRISMA) statement (Moher et al. 2009) and the Cochrane Handbook for Systematic Review (Higgins and Green 2011) were used to perform and report this systematic review and the results from the meta-analysis.
Study Selection
Both reviewers (M. van den Boomen and E.V. Hulleman) independently assessed the title and abstract of the studies that were proposed by the databases. Full-text reports of the eligible studies were obtained and again independently assessed by these same authors. Any differences in opinion about inclusion between the two authors were resolved, which led to consensus about all included papers. Quality assessment was performed by using the Newcastle-Ottawa quality assessment scale (NOS), in which the quality of the study was appraised using three domains:
• Selection and definition of study populations (0-4 stars); • Comparability of populations (0-2 stars);
• Ascertainment of exposure/outcome (0-3 stars).
Either the cohort or the case control version of the NOS was used, which depended on the performed study type.
Data Collection
Data were extracted by the same authors noting: study population, age, gender,
BMI, native T1-value, magnetic field strength, vendor, imaging analysis method and
cardiac MRI sequence. No authors were contacted for additional information. The data were collected as reported (mean ˘ standard deviation (SD)). For studies that reported the median with interquartile range (IQR) or full range, the mean and SD were calculated using the approach of S.P. Hozo et al. (Hozo et al. 2005). For studies with multiple groups, only the data from the relevant populations was extracted. The data of healthy control groups (controls) were also extracted.
Data Analysis
The T1outcome values of the individual studies were combined in a random-effects
model, providing computations of standard mean difference (SMD) and 95%
confidence intervals (CI). I2 was used as a measure of heterogeneity and I2ě50%
with Pă0.05 on the χ2test was defined as a significant degree of heterogeneity. This
was further explored by a meta-regression, bias and sensitivity analyses for groups with sufficient (ą10) included studies (Higgins and Green 2011). A mixed-effect model was used for the meta-regression analysis and performed by including the
reported covariates to determine their association with the myocardial T1-value. A
backwards elimination approach with a removal criterion of Pą0.05 was used to determine the final significant covariates. Covariates that were at least included in this analysis were: gender, age, field strength, MRI vendor information and the used sequence, even though the MOLLI and ShMOLLI have already shown to have
good overall agreement in T1-values under 1200ms (Piechnik et al. 2010).
Further-more, funnel plots were used to define the missing studies, the Egger test (Stuck et al. 1998) were performed to determine publication bias, and a sensitivity analy-sis was conducted by omitting each study sequentially and recalculate the model. These statistical analyses were performed using Review Manager (RevMan, version 5.3, The Cochrane Collaboration, Copenhagen, Denmark) and the package “metafor” in R (version 3.22, R Foundation for Statistical Computing, Vienna, Austria). Lastly, the weighted mean and weighted SD were determined separately for all studied popu-lations and field strengths using the number of subjects as weight factor.
Figure 2.1: Overview of study review process according to the PRISMA flow dia-gram (Moher et al. 2009)
2.3
Results
Results of Literature Search
The search strategy identified 660 relevant abstracts on PubMed and EMBASE. In addition, eight handpicked papers were included. After removing the duplicates, a total of 557 abstracts were evaluated. In total, 49 articles remained for the meta-analysis; 305 studies were excluded based on title and abstract, 173 were excluded based on full text screening, and 30 were excluded based on the published data. More specific reasons for exclusion are listed in Figure 2.1. Eventually, a total of ten
Figure 2.2:Weighted mean T1-values with weighted standard deviation (SD) of all
included 1.5T studies with hypertrophic cardiomyopathy (HCM), dilated cardio-myopathy (DCM), myocarditis (MC), iron overload, amyloidosis, Fabry disease, hypertension (HT) with (LVH+) and without (LVH-) left ventricular hypertrophy, type 2 diabetes mellitus (DM), and obesity populations (black) and healthy controls (grey)
studies were included for HCM (Kuruvilla et al. 2015, Fontana et al. 2014, Goebel et al. 2016, Małek et al. 2015, White et al. 2013, Dass et al. 2012, Hinojar et al. 2015a, Puntmann et al. 2013, Wu et al. 2016, Wu et al. 2017), nine for DCM (Puntmann et al. 2016, Goebel et al. 2016, Dass et al. 2012, Puntmann et al. 2013, aus dem Siepen et al. 2015, Chen et al. 2016a, van Oorschot et al. 2017, Hong et al. 2015, Punt-mann et al. 2014), twelve for MC (Goebel et al. 2016, Bohnen et al. 2015, Ferreira et al. 2014, Ferreira et al. 2013, Hinojar et al. 2015, Luetkens et al. 2016, Luetkens et al. 2016a, Lurz et al. 2016, Radunski et al. 2014, Radunski et al. 2017, Luetkens et al. 2014, Toussaint et al. 2015), five for iron overload (Alam et al. 2015, Feng et al. 2013, Hanneman et al. 2016, Sado et al. 2015, Camargo et al. 2016), six for amyloidosis (White et al. 2013, aus dem Siepen et al. 2015a, Banypersad et al. 2015, Fontana et al. 2015, Gallego-Delgado et al. 2016, Karamitsos et al. 2013), two in Fabry disease (Sado et al. 2013, Pica et al. 2014), ten for HT (Treibel et al. 2015, Ku-ruvilla et al. 2015, Hinojar et al. 2015a, Wu et al. 2017, Edwards et al. 2015, Fer-reira et al. 2016, Rodrigues et al. 2016, Rodrigues et al. 2017, Roux et al. 2017, Am-bale Venkatesh et al. 2014), four for DM (Jellis et al. 2014, Jellis et al. 2011, Khan et al. 2014, Levelt et al. 2015) and one in obesity (Khan et al. 2014) (See Appendix
2.B). Since the field strength is known to influence the T1-values significantly (Pica
et al. 2014), results from studies performed on 1.5T or 3T are shown separately. Fur-thermore the field strength was also used as covariant in the meta-regression analy-sis.
Study Quality
One study (Hinojar et al. 2015a) received the maximum score in the NOS in all ar-eas and a total of two studies (Hinojar et al. 2015, Sado et al. 2015) received the full score in the category of study group selection. Since some studies did not include a healthy control group they automatically received a minimum score at the compara-bility area and a lower score in ascertainment. The studies that did include healthy control subjects, but had a poor description of patient and control subject selection, received a lower score in the selection category. A total of 23 studies reported the use of blinded analysis and evaluation by at least two analysts, which increased their score on ascertainment (NOS scores in Appendix 2.B).
Figure 2.3:Weighted mean T1-values with weighted standard deviation (SD) of all
included 3T studies with hypertrophic cardiomyopathy (HCM), dilated cardiomyo-pathy (DCM), myocarditis (MC), iron overload, hypertension (HT), type 2 diabetes militus (DM), and obesity populations (black) and healthy controls (grey). With no reported values for amyloidosis or Fabry disease at 3T.
Hypertrophic Cardiomyopathy
The weighted mean (Sh)MOLLI T1-values in HCM patients and controls, measured
at 1.5T, were 1002˘52ms and 962˘37ms, respectively (Appendix 2.B and Figure 2.2). At 3T these weighted means were 1166˘55ms and 1081˘45ms, respectively (Appendix 2.B and Figure 2.3). The meta-analysis showed a significant increase of
the myocardial T1-values for HCM patients (SMD=1.41, 95%CI 0.93-1.88, pă0.01,
I2=78%, Figure 2.4). Furthermore, the meta-regression determined the vendor and
the age of HCM patients as significant covariates, which accounted for the hetero-geneity in the meta-regression model, with no other remaining significant residual
factors (I2=0%). This automatically means that the SMD between HCM patients and
Figure 2.4: Standardized mean difference between native myocardial T1of
hyper-trophic cardiomyopathy (HCM) patients and healthy controls with its associated random effects model, CI=confidence interval, IV=inverse variance, Std=standardized.
as potential covariates but not significant. Interpretation of the significant covariates showed that younger HCM patients and the use of a Siemens MRI would decrease the SMD. Furthermore, no significant funnel asymmetry was found for the random or mixed effect models (Pă0.24 and Pă0.37, respectively). The sensitivity analysis demonstrated that one study (Puntmann et al. 2013) influenced the model, but this was not significant (Pą0.09). This specific study used a different scanner (Phillips 3T) and a relatively young HCM patient population (44˘11 years) compared to the other studies.
Dilated Cardiomyopathy
The weighted mean (Sh)MOLLI T1-values in DCM patients and controls, measured
at 1.5T were 1008˘48ms and 970˘130ms, respectively (Appendix 2.B and Figure 2.2). At 3T these were 1165˘64ms and 1080˘46ms, respectively (Appendix 2.B and
Figure 2.3) The meta-analysis confirmed this increase in T1-values in the
myocar-dium for DCM patients (SMD=1.48, 95%CI 0.86-2.10, pă0.01, I2=85%, Figure 2.5).
The heterogeneity and study bias could not be determined because there were less than ten published studies that compared DCM patients with controls. However, an exploratory meta-regression analysis indicates that the percentage men in the DCM population and the age of the subjects in the control population might be the source of heterogeneity.
Figure 2.5: Standardized mean difference between native myocardial T1of dilated
cardiomyopathy (DCM) patients and healthy controls with its associated random effects model. CI=confidence interval, IV=inverse variance, Std=standardized.
Myocarditis
The weighted mean (Sh)MOLLI T1-value in active/acute MC patients and controls,
measured at 1.5T, were 1054˘61ms and 949˘28ms, respectively (Appendix 2.B and Figure 2.2). At 3T these were 1193˘60ms and 1068˘36ms, respectively (Appendix 2.B and Figure 2.3). Studies that compared the active/acute MC patients with
con-trols showed a significant increase of the T1-value for MC patients. The
meta-analysis confirmed this significant increase (SMD=1.96; 95% CI 1.42-2.51; I2=91%,
Pă0.01, Figure 2.6). Furthermore, the significant covariates were the vendor and left ventricular ejection fraction (LVEF) of the MC patients, which accounted for the heterogeneity in the meta-regression model with no other remaining significant
residual factors (I2=0%, P=0.77). A significant funnel asymmetry was found for
the random effect model with one possible missing study (P=0.03), but not for the mixed effect model including correction with the two covariates (P=0.45). The sen-sitivity analysis demonstrated that one study (Hinojar et al. 2015) introduces some heterogeneity into the model, but only the part of that study performed on 1.5T had significant influence on the model fit (Pă0.05).
Iron Overload
The weighted mean (Sh)MOLLI T1-value, in iron overload patients and controls,
measured at 1.5T, were 814˘128ms and 980˘34ms, respectively (Appendix 2.B and Figure 2.2). At 3T these were 1010˘144ms and 1162˘42ms, respectively (Appendix 2.B and Figure 2.3). Only three studies restricted the inclusion to one specific iron overload patient population (Alam et al. 2015, Feng et al. 2013, Hanneman et al.
2016), the other two studies used a mixed population of patients (Sado et al. 2015, Camargo et al. 2016). The number of included studies was not sufficient to conduct a
meta-analysis, but all studies showed a same decrease in T1-values for iron overload
patients compared to healthy controls (Figure 2.7).
Figure 2.6:Standardized mean difference between native myocardial T1of
myocar-ditis (MC) patients and healthy controls with its associated random effects model, CI=confidence interval, IV=inverse variance, Std=standardized.
Figure 2.7:Standardized mean difference between native myocardial T1of patients
with iron overload and healthy controls with its associated random effects model, CI=confidence interval, IV=inverse variance, Std=standardized.
Amyloidosis
Amyloidosis is the most typical type of restrictive cardiomyopathy (Hare 2011).
The weighted mean (Sh)MOLLI T1-values were only measured at 1.5T and were
1140˘69ms for patients and 960˘29ms for controls (Appendix 2.B and Figure 2.2). Three studies compared amyloidosis patients with controls (White et al. 2013, Bany-persad et al. 2015, Karamitsos et al. 2013), and all concluded that there was a
signifi-cant increase of the T1for amyloidosis patients. Some studies divided the
amyloido-sis patient populations in immunoglobulin light chain (AL) or transthyretin (ATTR) (Fontana et al. 2014), or cardiac or no cardiac involvement amyloidosis (Gallego-Delgado et al. 2016, Karamitsos et al. 2013). Karamitsos et al. 2013 showed that all their sub-populations, including the amyloidosis patients without cardiac
involve-ment, had a significantly increased T1-value compared to healthy controls.
Further-more, no meta-analysis was performed, because of the small number of included studies. However, the overall effect direction was similar for all studies (Figure 2.8).
Figure 2.8:Standardized mean difference between native myocardial T1of patients
with amyloidosis (AM) and healthy controls with its associated random effects weight factors, CI=confidence interval, IV=inverse variance, Std=standardized.
Fabry Disease
Fabry disease is a less common restrictive cardiomyopathy and only two studies
that included T1-mapping were found. Nevertheless, the weighted mean ShMOLLI
T1-values at 1.5T were 875˘48ms for patients and 968˘23ms for controls (Appendix
2.B and Figure 2.2). However, it should be noted that both studies used the same pool of healthy controls. Furthermore, due to the small number of studies no meta-analysis or regression could be performed on this data (Figure 2.9).
Figure 2.9:Standardized mean difference between native myocardial T1of patients
with Fabry disease and healthy controls with its associated random effects model, CI=confidence interval, IV=inverse variance, Std=standardized.
Chronic Hypertension
The weighted mean (Sh)MOLLI T1-value measured by 1.5T was 1044˘41ms for
HT patients with LVH, 984˘41ms for HT patients without LVH and 975˘40ms for controls (Appendix 2.B and Figure 2.2). At 3T these were 1070˘68ms for HT patients and 1023˘41ms for controls (Appendix 2.B and Figure 2.3). Four studies compared HT patients with LVH to HT patients without LVH and with healthy controls (Treibel et al. 2015, Kuruvilla et al. 2015, Rodrigues et al. 2017, Rodrigues
et al. 2016). They all reported a significant increase of T1 of the HT with LVH
populations compared with healthy controls (Pă0.05) and three of them also re-ported a significant increase compared with HT patients without LVH, while this
last group had not significant change in T1-values compared to healthy controls
(Treibel et al. 2015, Rodrigues et al. 2017, Rodrigues et al. 2016). Two studies pared HT patients to HCM patients (Hinojar et al. 2015a, Wu et al. 2017). The
com-parison of HCM with HT without LVH showed a significant higher T1 value for
the HCM patients (Pă0.01) (Hinojar et al. 2015a), while the comparison between HCM and HT with LVH showed no significant difference between the two (Wu et al. 2017). The meta-analysis of all HT patients (with and without LVH) together
showed a significant difference between T1-values of healthy controls and HT
pa-tients (SMD:0.19; %CI 0.01-0.37; I2=61%; P=0.04, Figure 2.10). Furthermore, the
meta-regression analysis showed that in HT patients LVH was the only significant covariate that corrected the I2to 4%.
Figure 2.10: Standardized mean difference between native myocardial T1 of all
hypertension (HT) patients with and without left ventricular hypertrophy (LVH) and healthy controls with its associated random effects model, CI=confidence inter-val, IV=inverse variance, Std=standardized.
Figure 2.11:Standardized mean difference between native myocardial T1of
hyper-tension (HT) patients without left ventricular hypertrophy (LVH) and healthy con-trols with its associated random effects model, CI=confidence interval, IV=inverse vari-ance, Std=standardized.
A second meta-regression was performed excluding the HT patients with LVH.
This analysis showed no significant difference between the T1-values of healthy
con-trols and HT patients without LVH (SMD: 0.03; %CI -0.07–0.13; I2=2%; P=0.52,
Fig-ure 2.11). In a further analysis of the potential funnel asymmetry, missing studies and influencing studies, no significance was found for both groups of HT patients without LVH and all HT patients (HT without LVH: Pă0.83, P=0.5, and Pą0.05, respectively and all HT: P=0.09, P=0.5, Pą0.05, respectively).
Overweight/Obesity and Type 2 Diabetes Mellitus
DM and obese populations are studied less extensively with T1-mapping compared
with the other NICM and hypertension populations. The weighted mean MOLLI
T1-value measured on 1.5T was 853˘202ms for DM patients (Jellis et al. 2014, Jellis
et al. 2011, Khan et al. 2014), 963˘116ms for obesity subjects and 986˘87ms for the healthy controls (Khan et al. 2014) (Appendix 2.B and Figure 2.2). At 3T the only
measured T1-values were 1194˘32ms for DM patients and 1182˘28ms for controls
(Levelt et al. 2015) (Appendix 2.B and Figure 2.3). Furthermore, no meta-analysis could be performed for either of the two cardiovascular risk populations, because of the small number of studies (Figure 2.12 and 2.13).
Figure 2.12: Standardized mean difference between native myocardial T1 of type
2 diabetes melitis (DM) patients and healthy controls with its associated random effects model, CI=confidence interval, IV=inverse variance, Std=standardized.
Figure 2.13: Standardized mean difference between native myocardial T1of obese
subjects (OB) and healthy controls with its associated random effects model, CI=confidence interval, IV=inverse variance, Std=standardized.
2.4
Discussion
The findings of this systematic review and meta-analysis showed that native
myo-cardial T1-values changed significantly in patients with HCM, DCM, MC,
amyloi-dosis, and iron overload. This supports previously published work on the
diagnos-tic value of native T1-mapping to detect diffuse myocardial fibrosis, inflammation,
iron accumulation, and protein deposition (Germain et al. 2014, Captur et al. 2016).
HT patients without any LVH showed no significant change in T1-values, which
indicates the absence of tissue modifications, while HT patients with LVH had a
significantly increased T1. Insufficient number of studies has been conducted into
Fabry disease and the other populations with increased cardiovascular risk (DM
and obesity) to draw conclusions about changes in their myocardial T1-values.
While this meta-analysis confirmed the clinical potential of T1-mapping (Greulich
et al. 2016, Haaf et al. 2016), it also showed a strong variation in different reported T1
-values for healthy controls. Although the T1-values at 1.5T strongly seemed to vary,
MOLLI-based T1-value of 950˘21ms (Dabir et al. 2014). In studies performed at
3T none of the T1-values for controls were significantly different from the expected
1053˘23ms (Dabir et al. 2014). Moon et al. 2013 already stressed the need to improve
standardization of T1-mapping by describing protocol recommendations. However,
they also stated that there is no current standard for T1-mapping sequences, nor
for analysis and mapping methods. Since it is now recognized that the T1-value is
influenced by these factors, it already led to some inconsistencies in the reported T1-values (Hamdy et al. 2016).
In addition to the influence field strength and choice in sequence, the method
for post-processing of the T1-map can also introduce bias, errors, and loss of
pre-cision. This can especially be seen in protocols that have a difference in using regional ROIs, image segmentation, and variable slice orientations (Moon et al.
2013). Almost half of the included studies used ROIs to determine the T1 (White
et al. 2013, Puntmann et al. 2013, aus dem Siepen et al. 2015, Chen et al. 2016a, van Oorschot et al. 2017, Hong et al. 2015, Puntmann et al. 2014, Ferreira et al. 2013, Lurz et al. 2016, Radunski et al. 2017, Toussaint et al. 2015, Alam et al. 2015, Feng et al. 2013, Sado et al. 2015, Camargo et al. 2016, aus dem Siepen et al. 2015a, Banyper-sad et al. 2015, Fontana et al. 2015, Gallego-Delgado et al. 2016, Edwards et al. 2015, Rodrigues et al. 2016, Rodrigues et al. 2017, Roux et al. 2017, Ambale Venkatesh
et al. 2014). Nevertheless, Moon et al. 2013 recommended global myocardial T1
measurements and Puntmann et al. clearly showed the importance of this in their studies on DCM patients (Puntmann et al. 2016, Puntmann et al. 2013, Puntmann
et al. 2014). They reported the T1-values from a rectangular region of interest (ROI)
in the septum and compared it with an average T1 of the whole short axis (SAX).
The T1-value for the whole SAX showed no significant difference between DCM
pa-tients and controls (P=0.05), while the T1-values in the septal ROI were significantly
increased for DCM patients (Pă0.05). In addition to this, the T1-values of studies
that used the segment approach based on the AHA model could also be influenced
by similar averaging effects on the T1-values (Małek et al. 2015, aus dem Siepen
et al. 2015, Luetkens et al. 2016, Luetkens et al. 2016a, Luetkens et al. 2014, aus dem Siepen et al. 2015a, Fontana et al. 2015, Ferreira et al. 2016, Roux et al. 2017, Jellis et al. 2014, Jellis et al. 2011). Furthermore, some studies used the 4-chamber plane for
T1-mapping (Fontana et al. 2014, White et al. 2013, Banypersad et al. 2015, Fontana
et al. 2015, Gallego-Delgado et al. 2016, Karamitsos et al. 2013), which can lead to errors due to potential through-plane respiratory motion. All these factors, together
with the lack of standard protocols make it difficult to determine a normative T1
Fortunately, even with these analysis-based biases, the SMD between controls and the studied cardiac diseases are shown to be less variable across studies and sites. The SMD from this meta-analyses showed to be independent of the applied field strength and MR sequence, and only for the HCM and MC population the SMD depended on the system vendor. Moon et al. 2013 recommend correcting for vari-ation in the scanner’s characteristics and this meta-analysis demonstrates that this correction should mainly be based on vendor. Aside from the variation and lack of
standardization, the SMD shows that native T1has diagnostic value for most of the
included cardiac diseases.
NICM can have subtle and diffuse fibrosis patterns that are difficult to determine (Puntmann et al. 2016) and inclusion and study bias are a remaining concern in NICM studies. The funnel plots and egger tests showed that there was indeed some publication bias for the MC analysis, which should be kept in mind when inter-preting this SMD. However, none of the other populations showed this bias, and
heterogeneity in T1-values was only caused by the vendor, age and gender
covari-ates. These factors were already well known to influence myocardial T1-values and
are important to correct for (Moon et al. 2013, Biernacka and Frangogiannis 2011). In addition to this heterogeneity, some studies (White et al. 2013, Dass et al. 2012, Wu
et al. 2016, Hong et al. 2015) reported T1-values of LGE based ROIs. Since LGE is
known to be highly non-specific and can lead to missing the full representation of the disease (Moon et al. 2013, Moon 2007), these LGE based ROI data was excluded from the meta-analysis. After correcting the SMD for these heterogeneity factors,
the meta-analysis still showed that there are significant changes in T1and although
LGE is still the clinical standard to determine focal fibrosis, a change of native T1is
clearly associated with an increase in fibrotic tissue as well (Germain et al. 2014).
In addition to sensitivity for myocardial fibrosis, T1-values can also indicate edema
formation (inflammation), and deposition of substances like protein and iron, which makes it a nonspecific parameter (Germain et al. 2014, Greulich et al. 2016).
Nev-ertheless, the T1-mapping seemed to be sensitive enough to differentiate between
clinical disease stages of patients with myocarditis when an additional baseline scan and clinical records are provided (Hinojar et al. 2015, Lurz et al. 2016,
Hino-jar et al. 2016). Therefore, T1-values may help to follow disease progression and
treatment (Hinojar et al. 2016). However, this meta-analysis only confirmed the
sig-nificant changes in myocardial T1-values in the acute phase of MC.
Iron accumulation also changes myocardial T1-values by shortening the
relax-ation times significantly, which suggests T1-mapping could also be of value in the
included studies (Sado et al. 2015) evaluated the T2*of an iron overload patient
pop-ulation and concluded that one-third of them had a normal T2* while the T1-value
was decreased. They stated that T1-mapping might even be more sensitive to iron
accumulation than T2*imaging, but the amount of accumulated iron that correlates
to these T1-values still needs to be confirmed by human histology. Furthermore,
the differences in iron concentration of all included subjects in the different studies
might have caused a wide range in T1-values and more research to the correlation
between T1-values iron concentration in the myocardium is needed to determine
whether T1-mapping could also be used for monitoring.
All amyloidosis studies reported a significant increase in myocardial T1-values,
even for amyloidosis patients that had no biopsy or decreased cardiac function
that confirmed cardiac involvement. This meta-analysis showed that T1-mapping
is highly sensitive to increases of the interstitial space caused by myocardial protein
depositions in amyloidosis (Germain et al. 2014), which indicates that T1-mapping
could help with early detection of amyloidosis deposition in the heart. The signif-icant increased SMD was even found while multiple included studies used the
4-chamber imaging plane for T1-mapping which introduces more variation in T1but
is also commonly used to study amyloidosis patients (Fontana et al. 2014, White et al. 2013, Banypersad et al. 2015). Further research with cardiac axial slices is
needed to determine the classification potential of the T1-value in amyloidosis
pa-tients.
While HT and NICM patients seem to have several anatomical cardiac MRI read-outs in common, none of the included studies in this meta-analysis reported a
sig-nificant increase in T1-values for HT patients without LVH. Only patients with HT
in combination with LVH showed a significant change in T1-value (Rodrigues et al.
2016, Rodrigues et al. 2017), similar to HCM. However, all studies reported the mean
T1-value of the whole myocardium, which ignores the fact that HT might be
associ-ated with inhomogeneous T1distribution (Schelbert and Messroghli 2016). Further
research is needed to determine the ability of T1-mapping to image this
inhomo-geneity and whether it is applicable to follow HT progression.
Two studies reported clearly decreased T1-values for DM (Jellis et al. 2014, Jellis
et al. 2011), but had no healthy control population for comparison. A reason for this decrease might be that DM patients are known to develop myocardial steatosis due to their insulin resistance and the associated myocardial fat lowers the native
T1-value (Khan et al. 2014). However, the fat content of these myocardial steatosis
is much smaller than in Fabry disease, and the number and sizes of T1-mapping
other studies reported a smaller decrease in T1 for DM patients and did compare
them with healthy controls, which resulted in no significant difference (Khan et al. 2014, Levelt et al. 2015). Levelt et al. 2016 used healthy control subjects with a BMI of 28.6˘5.7, which raises the question whether healthy controls should also have a healthy weight (BMIă25). This concern is the same for the DM populations, because the DM patients in these included studies had a weighted mean BMI of 31˘5, which means at least part of the populations was also overweight or obese. Only one study (Boudina and Abel 2010) compared DM patients with a lean group of healthy controls and obese controls separately. However, the obesity subjects did not differ significantly from either of the two other populations in this study. Further research with lean controls and DM patients (BMIă25) is needed to confirm
the reported changes in T1-value, and whether it is possible to distinguish these
populations from NICM patients.
T1-mapping has numerous MRI dependent and methodological factors that can
influence the final T1-values (Camargo et al. 2016). As described before, the field
strength and sequence choice are two of these factors, but this meta-analysis showed
that both did not influence the SMD, even though the T1-values at 3T were about
100ms higher than at 1.5T. More research towards understanding the effect on
accu-racy, precision and reproducibility of T1-mapping is needed (Moon et al. 2013,
Kell-man et al. 2014). Without this knowledge, it remains unknown whether the variance
of the T1-maps is mainly caused by variability in physiological effects, or the
inac-curacy of the technique itself. However, the HCM, DCM, MC and HT patient popu-lations were studied in groups of sufficient size to assume that the significant SMD
of T1-values is caused by changes in tissue physiology. Further research should be
conducted on DM and obesity populations and on other possible factors associated
with variance in T1-mapping values.
The non-uniform reporting of data, heterogeneity of patient populations,
meth-ods for T1-mapping, differences in ROI placement, and the small number of studies
(for amyloidosis, iron overload, DM and OB), formed the major limitations of this meta-analysis. Most studies, particulary the studies with a greater sample sizes, did not publish their data per patient and therefore no conclusions could be drawn on a patient-by-patient basis. Future prospective studies should provide complete patient-level insight which may help mitigate selection bias for especially amyloi-dosis, iron overload, DM and obesity studies. In addition, the patient characteristics
should be published together with the T1-values to enable better correlation
anal-ysis. Finally, we had to compare the T1-values of a smaller number of
MC and HT diseases. However, since all studies to iron overload and amyloidosis
showed the same change in T1, this could be ascribed to the physiological alteration
associated with the diseases. For the DM and obesity populations this direction was less obvious.
In conclusion, this meta-analysis showed that native T1-mapping is a reliable way
to distinguish HCM, DCM, MC, iron overload, amyloidosis and HT patients with
LVH from healthy controls and HT patients without LVH. This indicates that T1
-mapping could help with the diagnosis of certain cardiomyopathies at an earlier stage than assessment of cardiac function alone. In addition, DM and obesity seem
to affect myocardial T1-values, although the change in T1 is opposite to what has
been reported in non-infiltrative NICM. Further research into these risk-populations
is needed to determine the degree of overlap in myocardial T1-values in the healthy,
cardiovascular risk and NICM population.
Conclusion
This chapter confirmed the applicability of native T1-mapping to assess
dif-fuse fibrosis in non-ischemic cardiomyopathies (NICM) and that it can even help to determine the presence of left-ventricular remodeling in hypertension
(HT) patients. However, the included populations that showed significant T1
changes were already selected based on clinical markers that confirmed the on-set of NICM, while for the populations without noticeable cardiac remodeling no change in native T1 could be detected. Therefore, it can be concluded that
T1-mapping could be used in addition to the assessment of cardiac function but
would not be appropriate as an early marker for cardiac remodeling. However, since oxygenation impairment is thought to be an important precursor of fibro-sis (Olivotto et al. 2006, Galati et al. 2016), this could potentially be used as an early marker for fibrosis paired remodeling.
2.A
Search term Native T
1Meta-Analaysis
The search term used in PubMed /MEDLIN for Literature Search:
(((”Cardiomyopathy, Hypertrophic”[Mesh] OR “hypertrophic cardiomyopathy” [tiab] OR (hypertrophic [tiab] AND cardiomyopathy [tiab]) OR HCM [tiab]) OR (”Fabry Disease” [Mesh] OR “Anderson Fabry” [tiab] OR “Fabry’s Disease” [tiab]) OR (”Obesity” [Mesh] OR ”Overweight” [Mesh] OR (body [tiab] AND weight [tiab]) OR (body [tiab] AND mass [tiab] AND index [tiab]) OR obes* [tiab] OR overweight [tiab]) OR (”Hypertension” [Mesh] OR ”blood pressure” [MeSH Terms] OR hyper-tens* [tiab] OR ”high blood pressure” [tiab]) OR (”Amyloidosis” [Mesh] OR Amyloi-dosis [tiab] OR amyloid [tiab]) OR (”Hypertrophy, Left Ventricular” [Mesh] OR “left ventricular hypertrophy” [tiab] OR (ventricular [tiab] AND hypertrophy [tiab]) OR LVH [tiab]) OR (”Cardiomyopathy, Dilated” [Mesh] OR “Dilated cardiomyopathy” [tiab] OR (dilated [tiab] AND cardiomyopathy [tiab]) OR DCM [tiab]) OR (”Myo-carditis” [Mesh] OR myocarditis [tiab] OR carditis [tiab] ) OR (”Hemochromato-sis” [Mesh] OR hemochromatosis [tiab] OR haemochromatosis [tiab]) OR (”Sclero-derma, Systemic” [Mesh] OR scleroderma [tiab] ) OR (”Diabetes Mellitus” [Mesh] OR diabet* [tiab]) OR (”Myocardial Ischemia” [Mesh] OR ”Myocardial Ischemia” [tiab] OR ”Myocardial Ischaemia” [tiab] OR ”Coronary Artery Disease” [tiab] OR (coronary [tiab] AND artery [tiab]) OR (coronary [tiab] AND disease [tiab]) OR “myocardial infarction” [tiab] OR (myocardial [tiab] AND infarction [tiab])) OR (”Cardiotoxicity” [Mesh] OR Cardiotoxic* [tiab]) OR (“Myocardial fibrosis” [tiab] OR (myocardial [tiab] AND fibrosis [tiab]) OR (cardiac [tiab] AND fibrosis [tiab])) OR (”Cardiomyopathies” [Mesh] OR cardiomyopathy* [tiab]) OR (”Iron Overload” [Mesh] OR (Iron [tiab] AND overload [tiab])) OR (Alcoholism [tiab] AND cardio-myopathy [tiab]) OR (”Heart Failure” [Mesh] OR (heart [tiab] AND failure [tiab]))) AND (”Magnetic Resonance Imaging” [Mesh] OR ”magnetic resonance” [tiab] OR ”MR imaging” [tiab] OR NMR [tiab] OR “nuclear magnetic resonance” [tiab] OR MRI [tiab] OR CMR [tiab] OR ((cardiovascular[tiab] OR cardiac [tiab]) AND mag-netic [tiab] AND resonance [tiab]) OR ((t1 [tiab] OR “t 1” [tiab]) AND (maps [tiab] OR mapping [tiab]))) AND (((t1 [tiab] OR “t 1” [tiab]) AND (maps [tiab] OR map-ping [tiab] OR value [tiab] OR values [tiab] OR time [tiab] OR times [tiab]) OR laxation rate” [tiab] OR “relaxation rates” [tiab] OR “relaxation time” [tiab] OR “re-laxation times” [tiab] OR “re“re-laxation value” [tiab] OR “re“re-laxation values” [tiab] OR ((r1 [tiab] OR “r 1” [tiab]) AND (value [tiab] OR values [tiab] OR time [tiab] OR times [tiab]))) AND (Heart [mesh] OR heart [ti] OR myocard* [tiab])) AND (”2011/01/01” [PDat] : ”2017/06/01” [PDat]) NOT (rabbit* OR pig OR PIGS OR mice OR rat OR rats OR children [mesh] OR child* [tiab]) NOT Case Reports [ptyp]).
2.B
Table of Included T
1Publications
Author (year) Disease/ Contr ol (n) T1 (ms) Disease T1 (ms) Contr ol P-value ROI place Study design Sequence NOS Population HYPER TROPIC CARDIOMYOP A THY 1.5T (Fontana et al. 2014) 46/52 1026 ˘64 967 ˘34 Mean base SAXor 4CH Pr osp. single center ShMOLLI Piechnik 2010 3,0,2 72% asym-metrical septal HCM, 60% L V outflow obstr uction, 76% LGE. (Goebel et al. 2016) 12/54 980 ˘43.6 955 ˘33.5 ă 0.05
Mean mid SAX
Retr
osp.
single center MOLLI 5(3)3 FA=35 TI=120- 4103
3,0,1 Subjects se-lected after CMR diagno-sis after image analysis (Kur uvilla et al. 2015) 20/22 996 ˘32.5 967.4 ˘35 ă 0.01
Mean base/ mid SAX
Pr
osp.
single center MOLLI Mess- roghli 2007
3,0,1 L VM for man ą 81g/m 2 for woman ą 61g/m 2, with BPM ą 140/90 mmHg
(Małek et al. 2015) 25/20 987 ˘52* 939.7 ˘47.9* ă 0.01 ă 0.01
Seg base- /mid- septal /lateral
Pr osp. single center ShMOLLI Piechnik 2010 2,0,1 HCM with L V hypertr ophy ď 15mm (White et al. 2013) 25/50 1058** 968**
4CH, septum base- /mid- LGE
Pr osp. single center ShMOLLI Piechnik 2010 3,0,2 Diagnostic, 80% asymmetrical septal HCM, max wall thick-ness 20 ˘ 4mm, 21 with LGE 3T (Dass et al. 2012) 28/12 1209 ˘28 1178 ˘13 ă 0.05 Mean 3SAX Pr osp. single center ShMOLLI Piechnik 2010 2,0,1 Pathogenic mu-tation or L V hy-pertr ophy ě 15 or ě 12mm fa-milial disease (Hinojar et al. 2015a) 95/23 1102 ˘58 1023 ˘44 Mean mid- SAX
Pr
osp.
multi center MOLLI Mess- roghli 2004
4,2,2 L V hypertr ophy ą 15mm, L V wall str ess or dilation, asym-metrical septal HCM
(Puntmann et al. 2013) 25/20 1254 ˘43 1070 ˘55 ă 0.01
Septal mid- SAX
Pr
osp.
single center MOLLI Mess- roghli 2007 2(3)5 FA=50
3,0,2 L V hypertr o-phy , no incr ease L V wall str ess. All asymmetric septal HCM (W u et al. 2016) 28/14 1241 ˘78.5 1114.6 ˘36.5 ă 0.05 ă 0.01
Mean base- /mid- SAX
Pr
osp.
single center MOLLI Mess- roghli 2004
2,0,1 L V wall thick-ness ě 15mm, LGE+ and LGE-seperate (W u et al. 2017) 11 1216 ˘26.5 Base- /mid- SAX
Pr
osp.
single center MOLLI Mess- roghli 2004
3,0,1 L V wall thick-ness ě 15mm by CMR, LGE+ and LGE-divided DILA TED CARDIOMYOP A THY 1.5T (aus dem Siepen et al. 2015) 29/56 1056 ˘62 1020 ˘40 ă 0.01
Mean mid- SAX AHA
Pr
osp.
Retr
osp.
single center MOLLI Mess- roghli 2004
HF suspected DCM, incr eased L VEDV/L VEDD and reduced L VEF (ď 45%)
(Chen et
al.
2016a)
21
1075 ˘83 ROI septum mid- SAX
Pr
osp.
single center MOLLI Mess- roghli 2004 3(3)5 FA=50
2,0,2 Referr ed for car diac resyn-chr onization (Goebel et al. 2016) 17/54 992 ˘37.3 855 ˘33.5 ă 0.01
Mean mid- SAX
Retr
osp.
single center MOLLI Mess- roghli 2004 5(3)3 FA=35 TI=120- 4103
3,0,1 Subjects se-lected after CMR diagnosis after image analysis (Puntmann et al. 2016) 357 SAX: 945 ˘141* Septal: 1004 ˘73* Septal and full mid- SAX
Pr
osp.
multi center MOLLI Malek 2015 3(3)3(3)5 FA=50
3,0,2 Diagnosied and incr eased L VEDV in-dexed to body surface ar ea and reduced EF (van Oorschot et al. 2017) 20/8 1166 ˘66 1026 ˘21 ă 0.01 ROI histo 3 mid-SAX Pr osp. single center MOLLI Mess- roghli 2004 FA=35
0,0,1
Explanted hearts
of
3T (Dass et al. 2012) 18/12 1225 ˘42 1178 ˘13 ă 0.01 Mean 3SAX Pr osp. single center ShMOLLI Piechnik 2010 2,0,1 echocar dio-graphy shows L VEF ă 45% and exclude CAD (Hong et al. 2015) 41/10 1247.5 ˘66.8 1205.4 ˘37.4 Not sig Mean Seg ROI
in 3 SAX Pr osp. single center MOLLI 3(3)3(3)5 FA=35 3,0,2 L V dilatation, L VEDD ě 6cm, systolic dys-function and L VEF ď 40% (excluding ischemic and restrictive CM) (Puntmann et al. 2013) 25/30 1254 ˘43 1070 ˘55 0.05
ROI septal mid- SAX
Pr
osp.
single center MOLLI Mess- roghli 2004 3(3)5 FA=50
3,0,2 Non-ischemic DCM, incr eased L V volume and reduced sys-tolic function (no LGE)
(Puntmann et al. 2014) 82/47 SAX: 1102 ˘72 ROI: 1145 ˘37 SAX: 1035 ˘47 ROI: 1055 ˘22 ă 0.01
Septal ROI /mid- SAX
Pr
osp.
single center MOLLI Mess- roghli 2004 3(3)5 FA=50
3,0,1 Incr eased L VEDV indexed to body surface ar ea, reduced L VEF , no LGE (Puntmann et al. 2016) 280 SAX: 1048 ˘127* Septal: 1111 ˘69* Septal /full mid- SAX
Pr
osp.
multi center MOLLI Mess- roghli 2004 3(3)3(3)5 FA=50
3,0,2 Diagnosised and incr eased L VEDV body surface ar ea and reduced EF MYOCARDITIS 1.5T (Bohnen et al. 2015) 16 of 31 1125 ˘93.5* ă 0.05 Mean 3SAX Pr osp. single center MOLLI Mess- roghli 2004 FA=35 TI=188- 3382
2,0,2 Recent-onset HF , L VEF ă 45%, no CAD, En-domyocar dial biopsy
(Ferr eira et al. 2014) 60/50 1011 ˘64 946 ˘23 ă 0.01
Mean base- /apex- SAX
Pr osp. multi center ShMOLLI Piechnik 2010 2,2,1 Suspected acute myocar ditis (Ferr eira et al. 2013) 50/45 1010 ˘65 941 ˘18 ă 0.01 ROI thr es-hold Pr osp. multi center ShMOLLI Piechnik 2010 2,2,1 Suspected myocar ditis, acute chest pain, elevation in tr oponin I level, recent viral disease, no ischemic (Goebel et al. 2016) A:19 C:26 /54 A:974 ˘35.9 C:965 ˘39.5 955 ˘33.5 ă 0.05 0.240
Mean mid- SAX
Retr
osp.
single center MOLLI 5(3)3 FA=35 TI=120- 4103
3,0,1 Established di-agnostic criteria (Hinojar et al. 2015) A:61, C:67 /40 A:1064 ˘37 C:995 ˘19 940 ˘20 ă .05 ă 0.05
Single mid- SAX
Pr
osp.
multi center MOLLI Mess- roghli 2004 3(3)3(3)5
3,0,1 Active viral myocar ditis, ă 1w symp-toms, ser olog-ical marker , no symptoms, no ser ological marker
(Luetkens et al. 2016) 34/50 MOLLI: 1048.6 ˘51.9 Sh- MOLLI: 887 ˘37.2 MOLLI: 966.9 ˘27.8 Sh- MOLLI: 831.4 ˘26.9 ă 0.01 ă 0.01 16AHA 3SAX Pr osp. single center MOLLI /
ShMOLLI Mess- roghli 2004/ Piechnik 2010
2,0,2 acute MC based on clinical observation and labora-tory . Contr ols wer e referr ed CMR with no abnormalities (Luetkens et al. 2016a) 24/45 1047.7 ˘44.0 965.1 ˘28.1 ă 0.01
diastolic SAX Seg
Pr
osp.
single center MOLLI Mess- roghli 2004 3(3)3(3)5 FA=35
3,0,2 acute chest pain, myo-car dial injury , viral infection, ser um marker (Lurz et al. 2016) A:43, C:48 A:1113 ˘67 C:1096 ˘64 ă 0.05
VLA, HLA, SAX manual ROI
Pr
osp.
single center MOLLI Mess- roghli 2004 FA=35 TI=108- 2965
1,0,1 Suspected MC, onset symp-toms, myocar -dial damage, viral disease, no CAD, with biopsy evi-dence A: ď 14d, C: ą 14d
(Radunski et al. 2014) 104/21 1098 ˘62* 1041 ˘42* ă 0.01 diastolic 3 SAX global Pr osp. single center MOLLI FA=35 TI=150- 3671
2,0,2 Recent infec-tion, elevated tr oponin, acute chest pain (n=38) or new onset heart failur e (n=66) (Radunski et al. 2017) 20/20 1225 ˘109* 1045 ˘34* ă 0.01 3 SAX LGE ROI Pr osp. single center MOLLI 3(3)5 FA=35 TI=88- 3382
1,0,1 Lake Louise Criteria, includ-ing CMR 3T (Hinojar et al. 2015) A:61 C:67 /40 A: 1189 ˘52 C:1099 ˘22 1045 ˘23 ă 0.05 ă 0.05
Single mid- SAX
Pr
osp.
multi- center MOLLI Mess- roghli 2004 3(3)3(3)5
3,0,1 A:V iral myocar -ditis, ă 1w after symptoms and C: no symptoms and no ser ologi-cal marker
(Luetkens et al. 2014) 24/42 1185.3 ˘49.3 1089.1 ˘44.9 ă 0.01 3 SAX Seg Pr osp. single center MOLLI Mess- roghli 2004
2,0,1 V iral infection, elevated ser um marker , myo-car dial injury , no CAD. Con-tr ols: referr ed for CMR no abnormalities (Lurz et al. 2016) A:43, C:48 A:1203 ˘71 C:1185 ˘78
VLA, HLA, SAX
Pr
osp.
single center MOLLI Mess- roghli 2004 FA=35 TI=108- 2965
1,0,1 Onset symp-toms, myocar -dial damage, viral disease, no CAD A: ď 14 days C: ą 14 days, excluding negative biopsy (T oussaint et al. 2015) 6 LGE ROI 1179.2 ˘48.3 ROIs LGE Pr osp. single center MOLLI Mess- roghli 2004
1,0,1 chest pain, fever , ECG changes, eleva-tion of car diac enzyme levels IRON OVERLOAD
1.5T (Alam et al. 2015) 53/20 939 ˘113* 1005 ˘40* 0.21 T ˚ 2
septum mid- SAX ROI
Pr
osp.
single center MOLLI Mess- roghli 2004 FA=35 TI=120- 280
2,2,2 Car diac sider o-sis scr eening or follow-up. W ide range of population (Feng et al. 2013) 52 653 ˘133 septal mid- SAX ROI
Pr
osp.
single center MOLLI Mess- roghli 2004 TI=100- 260
1,0,0 T ransfusion patients with thalassemia major receiv-ing therapy , T ˚ ă2 20ms (Hanneman et al. 2016) 19/10 850.3 ˘115.1 1006.3 ˘35.4 ă 0.01 3 SAX Pr osp. single center MOLLI 5(3)3 FA=35 TI=120- 4000
2,0,2 Thalassemia major patients receiving reg-ular blood transfusion and T ˚ ă2 20ms
(Sado et al. 2015) 88/67 827 ˘135 968 ˘32 ă 0.01 T ˚ 2 based ROIs Pr osp. single center ShMOLLI Piechnik 2010 4,0,2 β -thalassemia major (n=53) and the others had several other underly-ing diagnosis 3T (Alam et al. 2015) 53/20 1038 ˘167* 1155 ˘52* ă 0.01 T ˚ 2
based septal mid- SAX
Pr
osp.
single center MOLLI Mess- roghli 2004 FA=35 TI=100- 260
2,2,2 Car diac sider o-sis scr eening or follow-up. W ide dynamic range of ir on overload popu-lation (Camar go et al. 2016) 5/17 868.9 ˘120.2 1171.2 ˘25.5 ă 0.05
ROI mid- septum
Pr
osp.
single center MOLLI Mess- roghli 2004 FA=35
3,0,2 Referr ed for ir on quantifica-tion, all patients has T ˚ ă2 20ms
AMYLOIDOSIS 1.5T (aus dem Siepen et al. 2015a) 9 1009 ˘48* Mean mid- SAX
Pr
osp.
single center MOLLI FA=35 TI=100- 4400
2,2,2 Histologically proven TTR amyloid by en-domyocar dial biopsy , exclu-sion of any TTR gene variant (Banypersad et al. 2015) 100/54 1080 ˘87 954 ˘34 ă 0.01
4CH septal base ROI
Pr osp. single center ShMOLLI Piechnik 2010 3,0,2 Included 60 patients fr om baseline study (Fontana et al. 2015). Histological proof systemic AL amyloidosis
(Fontana et al. 2015) 250 (30 and 83) all:1082 ˘75 AL: 1150 ˘68 ATTR: 1113 ˘47 4CH base/ mid infer
o/ septum ROI Pr osp. single center ShMOLLI Piechnik 2010 2,0,1 Biopsy pr oven systemic AL, 91% histo-logical pr oof A TTR, 9 TTR mutations peo-ple with no evidence (Gallego- Delgado et al. 2016) 31 (5 and 26)
all:1197 ˘54 no:1265 ˘31 car: 1184 ˘47 4CH mid/ base SAX ROI
Pr osp. multi center MOLLI 1,0,1 Genetically proven TTR, car diac/no-car diac was CMR defined. AM was de-fined by uptake 99mTC-DPD tracer (Karamitsos et al. 2013) 14, 11 and 28 /36
no:1009 ˘31 pos: 1048 ˘48 def: 1140 ˘61 958 ˘20 ă 0.01 ă 0.01 ă 0.01
Mean mid- SAX and 4CH ShMOLLI Piechnik 2010 3,0,1 Histological confirmation of AM and echo-car diography for no, possi-ble (pos) and definite (def) car diac
(White et al. 2013) 20/50 1137** 968**
4CH LGE mid/ base ROI ShMOLLI Piechnik 2010 3,0,2 Pr oven noncar -diac biopsy and US with Mayo clinic classifica-tion 2/3 FABR Y DISEASE 1.5T (Pica et al. 2014) L VH- 25 LVH+ 38 /63 904 ˘46 853 ˘50 968 ˘32 Mean base/ mid- septum SAX
Pr osp. single center ShMOLLI 3,2,2 Genetically con-firmed diagno-sis fr om depart-ment of inher -ited car diovas-cular diseases (Sado et al. 2013) 44/67 882 ˘47 968 ˘32 Mean ROI base/ mid- SAX
Pr osp. single center ShMOLLI Piechnik 2010 Genetically proven patients fr om inherited car diac disease unit
CHRONIC HYPER TENSION 1.5T (Edwar ds et al. 2015) L VH- 43/43 956 ˘31 955 ˘30 Not sig Mean base- /mid- SAX septum
Pr osp. single center MOLLI 3(3)5 1,2,1 T reated HT pa-tients referr ed tohypertension clinic with no L VH (Ferr eira et al. 2016) L VH- 14/31 958 ˘23 954 ˘16 958 ˘19 Not sig 6 Seg SAX Pr osp. single center ShMOLLI Piechnik 2010 2,2,1 Essential HT , antihyperten- sive tr eatment ą 3 months, no sever e L V hypertr ophy (Kur uvilla et al. 2015) L VH- 23 LVH+ 20/22 974 ˘34 996 ˘33 967.4 ˘35 Not sig ă0.05 base- /mid- SAX
Pr
osp.
single center MOLLI Mess- roghli 2004 FA=35 TI=30- 1000
3,0,1 HT with and without L VH, HT sbp ą 140mmHg or dbp ą 90mmHg or taking medi-cation
(Rodrigues et al. 2017) L VH- 80 LVH+ 20/25 1035 ˘37 1070 ˘46 1026 ˘41 Not sig ă0.05 Mean mid- SAX septum
Pr osp. single center MOLLI FA=35 3,0,2 HT clinic, on SBP and DBP , no car dio-myopathy , no decr eased fil-tration rate, no sever e valvular heart disease (Rodrigues et al. 2016) L VH- 41 and 15 L VH+ 24 and 8 /29 1031 ˘35 1029 ˘45 1054 ˘41 1062 ˘41 1024 ˘41 Not sig ă0.05 mid- septum SAX
Observ . single center MOLLI FA=35 3,0,2 T ertiary HT clinic referr ed for CMR, no decr eased fil-tration rate, no sever e valvular heart disease (Roux et al. 2017) L VH- 10/10 952 ˘51 929 ˘80 Not sig Mean ROI 6 seg Pr osp. single center MOLLI 3(3)3(3)5 FA=35 1,0,2 Asymptomatic HT volunteers with no other car diovascular risks and no L VH
(T reibel et al. 2015) L VH- 40/50 948 ˘31 965 ˘38 Not sig Septum basal- SAX
Pr osp. single center ShMOLLI 3,1,1 HT patients without L VH but 35% showed L VH on MRI with BPM ě 140/90mmHg (Ambale Venkatesh et al. 2014) L VH- M:208 /415 F:196 /377 M:970 ˘38 F:984 ˘48 M:966 ˘37 F:986 ˘45 Not sig Manual ROI mid- SAX
Observ
.
cohort multi center MOLLI Mess- roghli 2004
1,0,2 MESA study of 6814 men and woman. HT with Joint Na-tional Commit-tee VI criteria 3T (Hinojar et al. 2015a) L VH- 69/23 1033 ˘68 1023 ˘41 Mid- SAX and septal ROI
Pr
osp.
single center MOLLI Mess- roghli 2004 3(3)3(3)5
4,2,2 T reated HT SBP ą 140mmHg DBP ą 95mmHg and concentric L VH ą 12mm no dilated L V (W u et al. 2017) L VH- 20 1197 ˘10.5 Base /mid ROI
Pr
osp.
single center MOLLI Mess- roghli 2004
DIABETES MELLITUS 1.5T (Jellis et al. 2014) 49 850 ˘293 881 ˘227 AHA 3SAX Pr osp. single center MOLLI FIEST A readout 2,0,1 DM with echo-car diography for myocar dial dysfunction (included) (Jellis et al. 2011) 13 and 54 Reg E: 786 ˘ 43 Irr eg E: 841 ˘ 185
Mean seg 3-SAX
Pr osp. single center MOLLI FIEST A readout 1,0,1 DM without vascular com-plications, valvular or ischemic heart disease (Khan et al. 2014) 11 /6 944 ˘93 985.5 ˘86.6 0.457
Single mid- SAX
Pr
osp.
single center MOLLI Mess- roghli 2004
2,2,1 DM without history of car -diovascular diseases fr om 1st and 2nd car e services.
3T (Levelt et al. 2015) 46/20 1194 ˘32 1182 ˘28 0.23
Single mid- SAX
Pr osp. single center ShMOLLI Piechnik 2010 2,2,1 Stable DM, no known complications. No history of car diovascular disease, chest pain, smoking, HT , or ischemic changes on electr ocar diog-raphy OBESITY 1.5T (Khan et al. 2014) 9 /6 962.3 ˘116.1 985.5 ˘86.6 Single mid- SAX
Pr
osp.
single center MOLLI Mess- roghli 2004
2,2,1 Obese, non-diabetic con-tr ols, excluding body mass ą 150kg
Table Appendix 2.B: The included T1-mapping publications, with Seg: segments, Prosp.:
prospective, Retrosp.: retrospective, SAX: single axial slice, LVH: left ventricular hyper-trophy, HCM: hypertrophic cardiomyopathy, DCM: dilated cardiomyopathy, DM: diabetes melitus, HT: hypertension, BPM: blood pressure, MOLLI: Modified Look-Locker Inversion-recovery, ShMOLLI: Shortened MOLLI, M: male, F:female, US: ultrasounds, LGE: late gadolinium enhancement, ROI: region of interest, VLA: vertical long axis, HLA: horizon-tal long axis, AHA: American Heart Association, AL: light-chain (amyloidosis), ATTR: transthyretin (amyloidosis), {beta-TM: beta-thalassemia major, CAD: coronary artery dis-ease, EMB: endomyocardial biopsy, HF: heart failure, LAX: long axis, LLC: Lake Louis criteria, LV: left ventricle, LVEF: left ventricular ejection fraction, LVEDV: left ventricular end diastolic volume, LVH: left ventricular hypertrophy, MI: myocardial infarction, MIO: myocardial iron overload, NS: non-significant, non-STEMI: non-ST-elevation myocardial infarction, PCI: percutaneous coronary intervention, NOS: Newcastle-Ottawa quality as-sessment scale, SAX: short axis, SD: standard deviation, Seq: MR sequence, SI: signal intensity, SLE: systemic lupus erythematosus, STEMI: ST-elevation myocardial infarction, TFE: turbo field echo, TI: thalassemia intermedia, TM: thalassemia major, TSE: turbo spin echo
