European Heart Journal – Cardiovascular Imaging (2015) 16, 14–22 doi:10.1093/ehjci/jeu182

Histological validation of cardiac magnetic resonance analysis of regional and diffuse interstitial myocardial fibrosis Leah M. Iles 1,2*, Andris H. Ellims 1,2, Huw Llewellyn 3, James L. Hare1,2, David M. Kaye 1,2, Catriona A. McLean 3, and Andrew J. Taylor 1,2 1

Department of Cardiovascular Medicine, Alfred Hospital, Melbourne, Australia; 2Baker IDI Heart and Diabetes Institute, Melbourne, Australia; and Department of Anatomical Pathology, Alfred Hospital, Melbourne, Australia

3

Received 8 April 2014; accepted after revision 19 August 2014; online publish-ahead-of-print 29 October 2014

Aim

Myocardial fibrosis is fundamental in the pathogenesis of heart failure. Late gadolinium enhancement (LGE) with cardiac magnetic resonance (CMR) imaging is commonly assumed to represent myocardial fibrosis; however, comparative human histological data are limited, especially in non-ischaemic cardiac disease. Diffuse interstitial myocardial fibrosis is increasingly recognized as central in the pathogenesis of cardiomyopathy and can be quantified using newer CMR techniques such as T1 mapping. We evaluated the relationship of CMR assessment of regional and diffuse fibrosis with human histology. ..................................................................................................................................................................................... Methods Eleven patients on the waiting list for heart transplantation (43.5 + 7.6 years, 64% male) and eight patients undergoing and results surgical myectomy for obstructive hypertrophic cardiomyopathy (57.1 + 8.6 years, 63% male) were recruited and underwent CMR prior to cardiac transplantation or myectomy. Quantification of fibrosis in explanted hearts using digitally analysed Masson-trichrome-stained slides was validated against picrosirius red-stained slides analysed using Image J, with an excellent correlation (R ¼ 0.95, P , 0.0001). Significant correlations were observed between LGE and histological fibrosis across a range of signal intensity thresholds in the explanted hearts (range: 2– 10 standard deviations above reference myocardium), with maximal accuracy at a threshold of 6 SD (R ¼ 0.91, P , 0.001). Assessment of interstitial myocardial fibrosis with post-contrast T1 times demonstrated a significant correlation on both segmental (R ¼ 20.64, P ¼ 0.002) and per-patient (R ¼ 20.78, P ¼ 0.003) analyses. ..................................................................................................................................................................................... Conclusion CMR provides accurate, non-invasive assessment of regional myocardial fibrosis using LGE, while diffuse interstitial myocardial fibrosis is accurately assessed with post-contrast T1 mapping.

----------------------------------------------------------------------------------------------------------------------------------------------------------Keywords

Cardiomyopathy † Magnetic resonance imaging † Pathology † myocardial fibrosis

Introduction Myocardial fibrosis is fundamental in the pathogenesis of heart failure, regardless of aetiology. It may be regional, such as replacement fibrosis observed following myocardial infarction, or a more diffusely distributed interstitial fibrosis as observed in advanced cardiomyopathy.1 Although both regional and diffuse interstitial myocardial fibrosis commonly coexist, the extent and distribution exhibit significant inter-individual variation. Accurate detection and quantification of myocardial fibrosis, both regional and diffuse, is therefore critical to

understanding pathophysiology, investigating possible therapies and predicting prognosis. Regional myocardial fibrosis following experimental myocardial infarction in animals has been shown to correlate with areas of late gadolinium enhancement (LGE) on cardiac magnetic resonance (CMR) imaging,2,3 and studies have confirmed the utility of CMR-LGE in the assessment of myocardial viability.4,5 LGE has also been observed in non-ischaemic cardiomyopathy (NICM), where it has been linked to adverse cardiac remodelling and increased risk of malignant arrhythmia.6,7 Although a qualitative relationship between CMR-LGE and

* Corresponding author. Tel: +61 390763263; fax: +61 390762461. Email: [email protected] Published on behalf of the European Society of Cardiology. All rights reserved. & The Author 2014. For permissions please email: [email protected].

15

Myocardial fibrosis and CMR

human histology has been observed,8 no study has quantitatively evaluated the accuracy of CMR-LGE in assessing regional myocardial fibrosis in humans. Consequently, the optimal method for quantitative CMR-LGE remains unresolved. CMR-LGE is often used to quantify regional fibrosis by setting a standard deviation threshold above the mean signal intensity of the reference myocardium. Until recently, guidelines recommended a 2 SD threshold;9 however, this is not consistently utilized, with authors reporting probable overestimation of regional fibrosis.10 A 5 or 6 SD threshold has been advocated;10,11 however, recent guidelines have not included a recommendation for clinical quantification of LGE due to lack of expert consensus.12 Even without regional fibrosis, increased extracellular collagen is found in many cardiac conditions.13,14 While CMR-LGE is limited by its requirement for non-involved reference myocardium, newer techniques to independently quantify diffuse interstitial fibrosis have been developed.15,16 A significant relationship between myocardial post-contrast T1 time and extent of diffuse histological fibrosis has been shown,16 – 18 using endomyocardial biopsies. In this study, we evaluated the relationship between CMR assessment of myocardial fibrosis (with CMR-LGE and post-contrast T1 mapping) and histological fibrosis in the left-ventricular (LV) myocardium of explanted hearts, as well as surgical myectomy specimens of patients with obstructive hypertrophic cardiomyopathy (HCM). We also describe for the first time a novel histological technique for whole human heart analysis, validated against standard histological techniques, allowing accurate correlation of histology at both a patient and segmental level.

Methods Study design Research was conducted at The Alfred Hospital, Melbourne, Australia. Subjects on the heart transplant waiting list were assessed for prospective CMR; additionally those with previous clinical CMR were invited to participate. For the evaluation of between-subject differences, subjects with obstructive HCM with CMR including post-contrast T1 mapping prior to surgical myectomy were recruited (Ellims et al., unpublished data, with permission). We used phantom studies to derive a heart rate (HR) correction algorithm. We then utilized novel histological methodology validated against established techniques to investigate CMR assessment of both regional and diffuse myocardial fibrosis. Informed consent was obtained from all participants, and the study was conducted under the guidelines of the Alfred Hospital Ethics Committee.

CMR imaging All CMR scans were performed using a clinical 1.5 T CMR scanner (Signa HD 1.5 T, GE Healthcare, Waukesha, WI, USA). Sequences were acquired during a 10 –15 s breath hold. LV function was assessed by a steady-state-free precession pulse sequence (TR: 3.8 ms, TE: 1.6 ms, 30 phases, slice thickness: 8 mm). Delayed hyper-enhancement was evaluated 10 min following a bolus of gadolinium – diethylene triamine penta-acteic acid (DTPA) (0.2 mmol/kg BW Magnevistw, Schering, Germany) to identify regional fibrosis using an inversion-recovery gradient echo technique (TR: 7.1 ms; TE: 3.1 ms; TI individually determined to null the myocardial signal, range: 180 – 250 ms, slice thickness: 8 mm, matrix: 256 × 192, number of acquisitions ¼ 2).

For the evaluation of interstitial fibrosis, a prototype sequence was used to cycle through acquisition of images over a range of preparation times as previously described.16 The sequence consisted of an ECGtriggered, inversion-recovery prepared, 2D fast gradient echo sequence employing variable temporal sampling of k-space (Global Applied Science Laboratory, GE Healthcare). Ten images were acquired 15 min following gadolinium – DTPA bolus with imaging parameters as previously published.16 All cine CMR sequences were performed in three standard short-axis slices (apical, mid, and basal), kept identical for each sequence throughout the CMR examination. From an end-diastolic four-chamber long-axis view, five equally spaced slices were planned, with the two outer slices lined up with the tip of the apex or the mitral annulus. The two outer slices were then deleted, leaving three slices corresponding to typical basal, mid, and apical short-axis views. Delayed-enhancement imaging was performed in both long- and short-axis views. For T1 mapping, the middle short-axis slice was utilized.

Phantom studies The influence of HR on post-contrast T1 times was investigated using agarose gel phantoms with different amounts of nickel chloride and known T1 times (range: 102– 512 ms), validated using conventional goldstandard inversion-recovery spin-echo imaging. HRs were generated ranging from 50 to 120 b.p.m. using an external HR simulator. T1 values were obtained using the technique and parameters outlined above, and results were compared with known phantom T1 times determined by spin-echo imaging. In line with prior modified Look-Locker (MOLLI) T1 mapping studies, we found that derived T1 mapping values were accurate at relatively short T1 times (,400 ms) and HRs ,80 b.p.m.; however, the combination of longer T1 times and HR .80 b.p.m. resulted in an overestimation of T1 time. This effect was explored using curve fitting of phantom data across all ranges of HR and phantom T1 times, where the following relationship was identified between T1 values obtained from the gold-standard spin-echo sequence (T1SE) and those measured with the in vivo sequence (T1measured) values T1meas = a.eb.T 1SE , Where a and b are fitting constants, which varied minimally with HR according to the following formulae: a = 88.68 − (0.2241 × HR), b = 0.0032 + (0.00002 × HR). This derived formula was applied to all T1 mapping sequences in this study.

CMR image analysis Evaluation of LV function and regional fibrosis LV function was evaluated globally using the biplane area-length method using two- and four-chamber long-axis views. Regional fibrosis was identified by myocardial LGE, defined quantitatively by myocardial postcontrast signal intensity at a number of SD (range: 2 – 10) above that within a reference region of remote myocardium within the same slice. In addition to global quantification of LGE at this level, the mid-short-axis slice was divided into six equal segments, starting from the anterior rightventricular insertion point, resulting in segments corresponding to the six standard AHA segments.19

Evaluation of interstitial fibrosis with T1 mapping Following image acquisition, the 10 short-axis images of varying inversion times were analysed using commercially available software (CMR42; Circle Cardiovascular Imaging, Inc., Calgary, Canada). To analyse regions of interest (ROI) to find average T1 for that area, data acquired

16

L.M. Iles et al.

at various preparation times were fitted to the exponential curve Mz (t ¼ TP) ¼ M0 [A– B (e – t/T1)] relating the sample magnetization Mz observed at the time t ¼ TP to the equilibrium magnetization M0 and sample T1 where TP denotes preparation time for an inversionrecovery experiment. In contrast to the more commonly utilized MOLLI T1 mapping sequences, each inversion pulse in our sequence was followed by a single readout phase (the ‘single-point’ technique), enabling measurement of true T1 rather than T1*. There was therefore no requirement to mathematically convert the fitted T1 value to a true T1 value, which is not the case for MOLLI sequences. For each image, a ROI was drawn around the entire myocardium, to calculate postcontrast myocardial T1 time for each subject. Post-contrast myocardial T1 time was also calculated separately for each of the six segments identified during LGE analysis. Patients with surgical myectomy for HCM had regional T1 time calculated in the area of surgical resection to best correlate with histology. All data were corrected for HR, renal function (corrected to glomerular filtration rate of 90 mL/min/ 1.73 m2), and time post-gadolinium administration (corrected to 15 min) as previously published.20 For all analyses, the threshold for appropriate curve fit was determined by including only those ROIs with R 2 ≥ 0.95.

Histological analysis Hearts were fixed in 10% formaldehyde, then an initial full cross-section mid-way between the tip of the apex and the atrioventricular groove was performed. This mid-section was then sectioned in its entirety. Blocks were processed according to normal protocols, sectioned at 6 mm and stained with Masson-trichrome (M-T) and picrosirius red. The M-T-stained slides were then scanned on a Microtek 1000XK lit-back scanner at 600PPI as A4 .tiff images. These .tiff images were loaded into Adobe Photoshop CS6 Extended (PS6). Individual slide sections were then isolated and pericardium was removed to avoid excessive detection of pericardial fibrosis (frequently seen following insertion of LV assist devices). PS6 action PixelCount.atn . Adjust brightness was run to aid in visual differentiation of the M-T stain colours (green ¼ fibrosis, red ¼ myocardial fibres). PS6 action PixelCount.atn . Remove background was run to remove non-tissue in background pixels. PS6 action PixelCount.atn . PixelCount was run to separate fibrosis pixels from

Table 1

remaining red/myocardial fibre pixels into two layers. The fibrosis pixel count was divided by the pixel count of both layers to determine percentage fibrosis. For reconstruction of whole heart images, individual sections were arranged in PS6 to match pre-sectioned heart photographic images and flattened to form the heart in cross-section. The LV was then divided into six segments to match CMR segmentation, and the above process was repeated to determine segmental percentage fibrosis. Sections stained with picrosirius red were photographed in their entirety at ×200 magnification. Sections were then analysed using the ImageJ software (Bethesda, MD, USA), as previously described.21,22

Statistical analysis All data were analysed blindly with independent analysis of histological and CMR results. Data are expressed as mean + SD unless otherwise indicated. For all comparisons, a P-value ,0.05 was considered significant; all reported P-values are two-tailed. Comparison of continuous baseline variables utilized the unpaired Student’s t-test, and comparison of categorical data, the x2 test or Fisher’s exact test. Correlations of variables were determined by calculating the Pearson Product Moment. Bland – Altman analysis was used to compare histological fibrosis with CMR-LGE. All analyses were conducted using Stata software version 11.1 (StataCorp, College Station, TX, USA).

Results Clinical and demographic data During the study period (January 2008–December 2012), 131 patients were waitlisted for heart transplantation with 101 transplants successfully performed. No patient on the transplant waiting list could undergo prospective CMR due to device implantation (pacemaker, defibrillator, or mechanical ventricular assist device). Eleven patients had CMR prior to device implantation and all consented to participate before heart transplantation. Clinical characteristics are outlined in Table 1. The mean age at CMR was 43.5 + 7.6 years, seven (64%) were male and the mean LV ejection fraction

Clinical characteristics

Age (years) Gender Aetiology Time (CMR-Tx, Weight (kg) Height (m) BMI (kg/m2) LVEF (%) LVEDVI LVMI (g/m2) days) (mL/m2)

............................................................................................................................................................................... 1* 2

48 44

M M

3

47

F

DCM

964

4* 5

54 49

M M

ICM DCM

815 521

6*

41

F

Restrictive

563

7 8

50 34

M M

Congenital DCM

228 2267

9

28

F

DCM

323

M F

DCM DCM

604 529

10 38 11* 45

DCM DCM

435 411

95 111

1.78 1.88

30.0 31.4

15.5 9.0

216.4 263.6

81.7 80.9

90

1.68

31.9

28.4

117.3

64.8

100 72

1.68 1.83

35.4 21.5

31.9 14.8

166.6 253.3

65.0 109.1

72

1.74

23.8

43.3

77.0

39.4

77 135

1.70 1.80

26.6 41.7

34.3 16.8

194.5 126.7

113.4 61.6

51

1.60

19.9

13.6

188.6

69.7

83 49

1.76 1.55

26.8 20.4

5.6 24.8

209.0 218.5

48.6 96.2

Tx, transplant; BMI, body mass index; LVEDVI, left-ventricular end-diastolic volume indexed to body surface area; LVMI, left-ventricular mass indexed to body surface area; DCM, dilated cardiomyopathy; ICM, ischaemic cardiomyopathy. *Patients with post-contrast T1 mapping performed are indicated with an asterisk.

17

Myocardial fibrosis and CMR

(LVEF) was 21.6 + 11.7%. Owing to the inability to re-scan patients on the waiting list, the median time from CMR to transplantation was 529 days (inter-quartile range: 423– 710 days). As post-contrast T1 mapping images were acquired in only 4 of the 11 patients during their clinically indicated CMR, the sample size was too small for between-patient analysis. Therefore, an additional eight patients with HCM who underwent CMR with post-contrast T1 mapping before surgical myectomy were evaluated. The mean age of this group was 57.1 + 8.6 years, five (63%) were male with preserved LVEF (66.3 + 7.8%).

Validation of histological analysis technique Histological quantification of collagen has been well published using picrosirius red staining8,16,23 analysed with ImageJ software (Bethesda, MD, USA).21 In this study, we reconstructed the mid-LV short-axis slice to analyse histology in individual AHA myocardial segments,19 Figure 1. To reconstruct this mid-LV short-axis slice and allow segmental analysis, we utilized M-T staining quantified digitally with pixel-by-pixel analysis using PS6. To validate this novel pathological method, we compared results from 24 slides using picrosirius

red-staining analysed with ImageJ with results from the same tissue using M-T staining and PS6. An excellent correlation was observed between the two groups (Figure 1, R ¼ 0.95, P , 0.0001). For the remaining analysis, M-T staining and PS6 were utilized, allowing an accurate comparison of segmental histology with in vivo CMR results.

Comparison of LGE with histological fibrosis Table 2 demonstrates LGE comparison across a range of SD thresholds (2– 10 SD) with histological fibrosis quantification. Both global (at the mid-LV short-axis slice) and segmental (six segments per patient) myocardia were analysed (Figures 2 and 3). A strong, statistically significant correlation was found at all measured SD, with the absolute value closest to histological fibrosis (11.3 + 12.1%) at a threshold of 6 SD (12.3 + 11.0%, R ¼ 0.91, P , 0.001). Minimal bias (by Bland –Altman analysis) was also found at 6 SD threshold, both on global (bias: 20.9991, 95% limits of agreement: 210.89 to 8.896) and segmental (bias: 20.9089, 95% limits of agreement: 225.46 to 23.64) analysis. Underestimation of histological fibrosis was observed with higher thresholds, whereas lower thresholds were found to overestimate histological fibrosis.

Figure 1: Histological validation of digital image analysis. For analysis of the mid-LV short-axis slice, histological images were reconstructed (A) to match the CMR image (B). The mid-LV short-axis slice was reconstructed using PS6 of Masson (green ¼ fibrosis)-Trichrome (red ¼ myocardial fibres) stains of the entire cross-section (A). Myocardial tissue was also stained with picrosirius red (C). Collagen was quantified using PS6 for Massontrichrome slides and ImageJ for picrosirius red slides. Excellent correlation was observed between the two techniques (R ¼ 0.95, P , 0.0001, D).

18

L.M. Iles et al.

Table 2 Correlations between histological myocardial fibrosis and CMR quantification of LGE at a range of signal intensity thresholds Quantification

% Global quantification (n 5 11)

Difference (LGE-histology)

R

% Segmental quantification (n 5 66)

Difference (LGE-histology)

R

............................................................................................................................................................................... Masson trichrome

11.3 + 12.1

–

–

11.8 + 16.4

–

LGE 2 SD LGE 3 SD

49.6 + 12.5 35.8 + 12.6

38.3 24.5

0.70* 0.78†

49.0 + 26.5 35.9 + 25.4

37.2 24.1

0.52§ 0.59§

LGE 5 SD

17.1 + 11.7

5.8

0.88‡

18.7 + 21.7

6.9

0.72§

‡

–

LGE 6 SD LGE 8 SD

12.3 + 11.0 6.6 + 8.2

1.0 24.7

0.91 0.92‡

12.8 + 18.8 6.8 + 13.9

1.0 25.0

0.74§ 0.76§

LGE 10 SD

3.5 + 5.4

27.8

0.93§

3.6 + 9.5

28.2

0.78§

*P , 0.05. † P , 0.01. ‡ P , 0.001. § P , 0.0001.

Figure 2: Histological analysis of myocardial fibrosis. (A) Mid-LV short-axis slice reconstructed using PS6 of Masson (green ¼ fibrosis)-Trichrome (red ¼ myocardial fibres) stains of entire cross-section. (B) Equivalent CMR image, showing good geographical correlation with LGE (white) (C) Pixel counted for fibrosis alone. (D) Pixels counted for myocardial fibres alone, both counted in six AHA segments within the slice to give per cent fibrosis.

Correlation of regional fibrosis and LGE In theory, LGE would best identify replacement rather than interstitial fibrosis, which may be better identified using T1 mapping. In line with the previous literature,11 we divided histological segments into two groups to assess segments with and without replacement fibrosis separately. Segments with ≥1 focus of uninterrupted fibrosis

measuring ≥2 mm were defined as containing replacement fibrosis and remaining segments were considered to have no histological replacement fibrosis.11 Of 66 segments analysed, 35 had ≥1 focus of replacement fibrosis (RF+), compared with 31 segments without replacement fibrosis (RF2). As expected, the mean histological fibrosis was higher in

19

Myocardial fibrosis and CMR

Figure 3: LGE and histological fibrosis. (A) and (B) demonstrate a significant correlation between CMR-LGE (6SD) and histological fibrosis at both a per-patient and per-segment level, respectively. The correlation is similar in segments with histological replacement fibrosis (C), but no correlation is observed in segments with only histological interstitial fibrosis (D).

RF+ compared with RF2 segments (18.5 + 18.9 vs. 4.3 + 5.3%, P ¼ 0.0001) with more LGE (6 SD threshold) identified in the RF+ than in the RF2 group (18.5 + 22.9 vs. 6.2 + 5.8%, P ¼ 0.004). In RF+ segments, the correlation between histological collagen and LGE remained strong (R ¼ 0.73, P , 0.0001), whereas in RF2 segments no significant correlation was identified (R ¼ 0.08, P ¼ 0.68) (Figure 3).

Effect of time delay between CMR and transplant A limitation of this study is the delay between CMR and transplant. For each patient, we calculated the difference in histological fibrosis and LGE (6 SD) and compared the result with the time delay between CMR and transplant. We found no correlation between time delay and difference in CMR and histological fibrosis (R ¼ 20.42, P ¼ 0.19).

Correlation between post-contrast T1 mapping and histological fibrosis Four out of 11 patients who underwent transplantation had postcontrast T1 mapping performed [median delay CMR-transplant ¼ 546 days (inter-quartile range: 506– 626 days)]. For each patient, ROIs were drawn to calculate post-contrast T1 times in each of the six segments in the mid-LV short-axis slice, giving a total of 24 segments. Three segments were excluded due to poor curve fit (R 2 ,

0.95), leaving 21 analysable segments. All T1 times were corrected for HR, renal function, and image acquisition time following gadolinium contrast administration.20 Post-contrast T1 times (330.3 + 34.8 ms) correlated well with histological fibrosis (16.0 + 18.2%) on a per-segment analysis (R ¼ 20.64, P ¼ 0.002, Figure 4A). For between-subject analysis, an additional eight patients with HCM and CMR post-contrast T1 mapping performed before surgical myectomy were analysed. The interventricular septum from the heart transplant group was considered most comparable with myectomy samples, and histological myocardial fibrosis in both groups was compared with corresponding regional post-contrast T1 mapping. The mean post-contrast T1 time was 341.9 + 57.2 ms, and the mean histological fibrosis was 10.6 + 6.7%. On a one segment per-subject analysis (n ¼ 12), postcontrast T1 time shortened as histological fibrosis increased (R ¼ 20.78, P ¼ 0.003, Figure 4B). When the 10 segments without replacement fibrosis were analysed separately (as defined earlier), a significant correlation between post-contrast T1 times and histological fibrosis persisted (R ¼ 20.71, P ¼ 0.02).

Discussion This study is the first to our knowledge to comprehensively evaluate both regional and diffuse myocardial fibrosis (using LGE and post-contrast T1 mapping, respectively) with whole human

20

L.M. Iles et al.

Figure 4: Post-contrast T1 times and histological fibrosis. Post-contrast T1 times correlated well with histological fibrosis in segmental analysis of explanted hearts (n ¼ 21, A). An additional eight patients with surgical myectomy for HCM were analysed together with interventricular septal segments from explanted hearts (combined n ¼ 12). On per-patient analysis, there was a significant correlation between post-contrast T1 times and histological fibrosis (B).

heart histological fibrosis. Prior studies have validated similar CMR methods for the quantification of interstitial fibrosis with histology, using post-contrast T1 mapping or other similar CMR techniques;16,17,22,23 however, only one has used whole human hearts,23 and in this study no analysis of LGE was reported. We report a novel methodology to allow for reconstruction of the whole mid-LV short-axis slice using standard histological methods and available digital technology validated against conventional single-section histological analysis with an excellent correlation. This allowed for accurate comparative quantification between histology and CMR findings at both segmental and patient level.

Identifying regional myocardial fibrosis CMR-LGE has been histologically validated to represent myocardial fibrosis following myocardial infarction2,3 and in HCM,11 and a qualitative association has been recently reported in NICM.8 Our findings provide evidence for the common assumption that LGE observed in NICM also represents myocardial fibrosis, and additionally we present data supporting a signal intensity of 6 SD above the reference myocardium as the most accurate cut-off for the quantification of myocardial fibrosis. Prior studies have utilized a wide range of thresholds for the quantification of LGE, and we found that while all signal intensity thresholds correlate with histological fibrosis, the minimum bias was found at 6 SD, suggesting that this is the best threshold for diagnostic accuracy. This finding provides important information for accurate non-invasive quantification of regional myocardial fibrosis, which is believed to be an important prognostic factor in cardiomyopathy.8 Some have advocated the ‘full-width half-maximum’ (FWHM) method,15 where the threshold is taken using a reference area of LGE rather than normal myocardium. This technique is most readily applicable to individuals with a dense area of LGE available as a reference myocardium, such as an area of infarct, or in some cases of HCM, where promising results using FWHM have been reported.10,15 In our population, areas of LGE were less discrete with diffusely abnormal myocardium, consistent with the predominant non-ischaemic aetiology, making the FWHM method technically difficult.

Identifying diffuse myocardial fibrosis LGE, while being an invaluable clinical CMR technique, is inherently unable to evaluate diffusely distributed fibrosis. This study provides whole heart histological confirmation of the correlation between histological fibrosis and post-contrast T1 mapping. Importantly, in contrast to LGE analysis, a significant correlation between postcontrast T1 mapping and histological fibrosis persisted even in the absence of histological replacement fibrosis. This is in keeping with our proposition that LGE identifies replacement fibrosis, whereas post-contrast T1 mapping quantifies all patterns of myocardial fibrosis. Additionally, it confirms that T1 mapping reflects not only large areas of scar, but also is independently associated with interstitial fibrosis. These findings are consistent with prior data from our group demonstrating that post-contrast T1 mapping quantifies diffuse myocardial fibrosis, validated histologically with endomyocardial biopsy samples obtained from the right ventricle.16 Subsequent studies confirmed histological validation of this and other quantitative methods using surgical samples or right-ventricular endomyocardial biopsy tissue.17,22 Only one other study to date has provided results using whole human hearts. In this study,23 six explanted hearts were analysed on both a per-segment and per-patient basis. Using both postcontrast T1 mapping and the ECV technique previously published by this group of investigators,22,24 a good correlation between CMR and histological myocardial fibrosis was observed. However, on a between-subject analysis, only the ECV technique remained significant. Our between-subject findings using post-contrast T1 mapping are not in agreement with this study, we suggest for a number of reasons. First, in a small study (n ¼ 6), the power of a betweensubject analysis is low; hence, the lack of a significant correlation may be due to a type-II error (a false-negative result, when a positive correlation exists but is not observed due to lack of statistical power). Also, known inter-individual variables such as time following contrast administration and renal function were not corrected in the previous study, whereas these were corrected in our results, utilizing previously published methods.20 Ideally, we would have compared our results with the ECV method; however, as our CMR data were

21

Myocardial fibrosis and CMR

mostly acquired prior to publication of the ECV technique, precontrast T1 mapping sequences required for this method were not performed. Similarly, while robust data have recently emerged in the utility of non-contrast T1 mapping across a range of cardiac diseases characterized by diffuse myocardial fibrosis,25 this promising T1 mapping technique could not be evaluated in this study. Finally, as MOLLI sequences are sensitive to changes in T2 as well as in T1,26 correction for T2 effect via the ECV calculation may be more important when MOLLI sequences are used as opposed to ‘single-point’ T1 mapping sequences we and others16,27 have utilized. Theoretically, pre- or post-contrast T1 mapping may be best suited to single-point T1 mapping sequences, whereas when MOLLI sequences are used, analysis with a ratio such as the ECV calculation may be more appropriate.

Limitations The most significant limitation is the time delay between CMR and acquisition of histology. In our institution, rate of device implantation is high, with most patients receiving devices well ahead of consideration for cardiac transplantation. However, the lack of correlation between the difference in CMR and histological fibrosis and time delay suggests the time difference is unlikely to have significantly altered the results. If anything, progression of fibrosis over time would be likely to reduce, not enhance, the strength of our results; hence, our findings may underestimate the correlation between CMR and histological assessment of fibrosis. Owing to difficulties in access to patients with both CMR and whole heart histology, study numbers are small. This is a problem inherent to investigators seeking to correlate clinical analysis with anatomical pathology, and large studies of this kind are unlikely to be conducted. The addition of myectomy samples allowed a per-patient post-contrast T1 mapping analysis, although not equivalent to whole heart histology; these are LV samples not confined to subendocardium, reducing the sampling error associated with biopsy specimens.

Conclusions CMR-LGE and post-contrast T1 mapping both accurately quantify histological myocardial fibrosis. LGE is strongly associated with replacement fibrosis, with a threshold signal intensity of 6 SD above reference myocardium most closely approximating histological fibrosis. Post-contrast T1 mapping demonstrates excellent agreement with histological fibrosis, even in areas without histological replacement fibrosis, identifying diffuse interstitial fibrosis.

Acknowledgements The authors thank Dr Chris Neil, Western Health, Melbourne, Australia for supplying phantoms. Conflict of interest: none declared.

Funding L.M.I. is supported by a National Health and Medical Research Council Postgraduate Research Scholarship, Melbourne, Australia and a combined National Health and Medical Research Council and National Heart Foundation Early Career Fellowship, Melbourne, Australia. A.H.E. is supported by a combined Heart Foundation of Australia and National

Heart and Medical Research Council Postgraduate Research Scholarship, Melbourne, Australia. J.L.H. is supported by a Cardiac Society of Australia and New Zealand Research Investigatorship. D.M.K. is supported by a National Health and Medical Research Council program grant. A.J.T. is supported by a National Health and Medical Research Council project grant.

References 1. Mewton N, Liu CY, Croisille P, Bluemke D, Lima JA. Assessment of myocardial fibrosis with cardiovascular magnetic resonance. J Am Coll Cardiol 2011;57:891 – 903. 2. Kim RJ, Fieno DS, Parrish TB, Harris K, Chen E-L, Simonetti O et al. Relationship of MRI delayed contrast enhancement to irreversible injury, infarct age, and contractile function. Circulation 1999;100:1992 –2002. 3. Schelbert EB, Hsu L-Y, Anderson SA, Mohanty BD, Karim SM, Kellman P et al. Late gadolinium-enhancement cardiac magnetic resonance identifies postinfarction myocardial fibrosis and the border zone at the near cellular level in ex vivo rat heart. Circ Cardiovasc Imaging 2010;3:743–52. 4. Bodi V, Sanchis J, Lopez-Lereu MP, Losada A, Nunez J, Pellicer M et al. Usefulness of a comprehensive cardiovascular magnetic resonance imaging assessment for predicting recovery of left ventricular wall motion in the setting of myocardial stunning. J Am Coll Cardiol 2005;46:1747 –52. 5. Kim RJ, Wu E, Rafael A, Chen EL, Parker MA, Simonetti O et al. The use of contrast-enhanced magnetic resonance imaging to identify reversible myocardial dysfunction. N Engl J Med 2000;343:1445 –53. 6. Iles L, Pfluger H, Lefkovits L, Butler MJ, Kistler PM, Kaye DM et al. Myocardial fibrosis predicts appropriate device therapy in patients with implantable cardioverterdefibrillators for primary prevention of sudden cardiac death. J Am Coll Cardiol 2011;57:821 –8. 7. McCrohon JA, Moon JCC, Prasad SK, McKenna WJ, Lorenz CH, Coats AJS et al. Differentiation of heart failure related to dilated cardiomyopathy and coronary artery disease using gadolinium-enhanced cardiovascular magnetic resonance. Circulation 2003;108:54– 9. 8. Gulati A, Jabbour A, Ismail TF, Guha K, Khwaja J, Raza S et al. Association of fibrosis with mortality and sudden cardiac death in patients with nonischemic dilated cardiomyopathy. JAMA 2013;309:896 –908. 9. Hundley WG, Bluemke D, Bogaert JG, Friedrich MG, Higgins CB, Lawson MA et al. Society for Cardiovascular Magnetic Resonance guidelines for reporting cardiovascular magnetic resonance examinations. J Cardiovasc Magn Reson 2009;11:5. 10. Maron MS. Clinical utility of cardiovascular magnetic resonance in hypertrophic cardiomyopathy. J Cardiovasc Magn Reson 2012;14:13. 11. Moravsky G, Ofek E, Rakowski H, Butany J, Williams L, Ralph-Edwards A et al. Myocardial fibrosis in hypertrophic cardiomyopathy: accurate reflection of histopathological findings by CMR. JACC Cardiovasc Imaging 2013;6:587 –96. 12. Schulz-Menger J, Bluemke DA, Bremerich J, Flamm SD, Fogel MA, Friedrich MG et al. Standardized image interpretation and post processing in cardiovascular magnetic resonance: Society for Cardiovascular Magnetic Resonance (SCMR) Board of Trustees Task Force on Standardized Post Processing. J Cardiovasc Magn Reson 2013;15:35. 13. Kasner M, Westermann D, Lopez B, Gaub R, Escher F, Kuhl U et al. Diastolic tissue Doppler indexes correlate with the degree of collagen expression and cross-linking in heart failure and normal ejection fraction. J Am Coll Cardiol 2011;57:977 –85. 14. Weber KT, Brilla CG. Pathological hypertrophy and cardiac interstitium. Fibrosis and renin-angiotensin-aldosterone system. Circulation 1991;83:1849 –65. 15. Flett AS, Hasleton J, Cook C, Hausenloy D, Quarta G, Ariti C et al. Evaluation of techniques for the quantification of myocardial scar of differing etiology using cardiac magnetic resonance. JACC Cardiovasc Imaging 2011;4:150 – 6. 16. Iles L, Pfluger H, Phrommintikul A, Cherayath J, Aksit P, Gupta SN et al. Evaluation of diffuse myocardial fibrosis in heart failure with cardiac magnetic resonance contrast-enhanced T1 mapping. J Am Coll Cardiol 2008;52:1574 –80. 17. Sibley CT, Noureldin RA, Gai N, Nacif MS, Liu S, Turkbey EB et al. T1 Mapping in cardiomyopathy at cardiac MR: comparison with endomyocardial biopsy. Radiology 2012;265:724 –32. 18. Mascherbauer J, Marzluf BA, Tufaro C, Pfaffenberger S, Graf A, Wexberg P et al. Cardiac magnetic resonance postcontrast t1 time is associated with outcome in patients with heart failure and preserved ejection fraction. Circ Cardiovasc Imaging 2013;6:1056 – 65. 19. American Heart Association Writing Group on Myocardial Segmentation and Registration for Cardiac Imaging Segmentation AHAWGoM, Imaging: RfC, Cerqueira MD, Weissman NJ, Dilsizian V, Jacobs AK, Kaul S et al. Standardized myocardial segmentation and nomenclature for tomographic imaging of the heart: a statement for healthcare professionals from the Cardiac Imaging Committee of the Council on Clinical Cardiology of the American Heart Association. Circulation 2002; 105:539–42.

22

L.M. Iles et al.

20. Gai N, Turkbey EB, Nazarian S, van der Geest RJ, Liu CY, Lima JA et al. T1 mapping of the gadolinium-enhanced myocardium: adjustment for factors affecting interpatient comparison. Magn Reson Med 2011;65:1407 – 15. 21. Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of image analysis. Nat Methods 2012;9:671 –5. 22. Flett AS, Hayward MP, Ashworth MT, Hansen MS, Taylor AM, Elliott PM et al. Equilibrium contrast cardiovascular magnetic resonance for the measurement of diffuse myocardial fibrosis: preliminary validation in humans. Circulation 2010; 122:138 – 44. 23. Miller CA, Naish JH, Bishop P, Coutts G, Clark D, Zhao S et al. Comprehensive validation of cardiovascular magnetic resonance techniques for the assessment of myocardial extracellular volume. Circ Cardiovasc Imaging 2013;6:373 – 83.

24. White SK, Sado DM, Fontana M, Banypersad SM, Maestrini V, Flett AS et al. T1 Mapping for myocardial extracellular volume measurement by CMR: bolus only versus primed infusion technique. JACC Cardiovasc Imaging 2013;6:955 –62. 25. Puntmann VO, Voigt T, Chen Z, Mayr M, Karim R, Rhode K et al. Native T1 mapping in differentiation of normal myocardium from diffuse disease in hypertrophic and dilated cardiomyopathy. JACC Cardiovasc Imaging 2013;6:475 –84. 26. Chow K, Flewitt JA, Green JD, Pagano JJ, Friedrich MG, Thompson RB. Saturation recovery single-shot acquisition (SASHA) for myocardial T mapping. Magn Reson Med 2014;71:2082 –95. 27. Slavin GS, Stainsby JA. True T1 mapping with SMART1Map (saturation method using adaptive recovery times for cardiac T1 mapping): a comparison with MOLLI. J Cardiovasc Magn Reson 2013;15(Suppl 1):P3.

IMAGE FOCUS

doi:10.1093/ehjci/jeu160 Online publish-ahead-of-print 3 September 2014

.............................................................................................................................................................................

A long way from the heart Marina Leitman1*, Eli Peleg1, Miriam Weinberger2, Mehrzad Cohenpour3, and Zvi Vered1 1 Department of Cardiology, Assaf Harofeh Medical Center and Sackler School of Medicine, Tel Aviv University, Zerifin 70300, Israel; 2Infectious Diseases Unit, Assaf Harofeh Medical Center and Sackler School of Medicine, Tel Aviv University, Zerifin, Israel; and 3Department of Nuclear Medicine, Assaf Harofeh Medical Center and Sackler School of Medicine, Tel Aviv University, Zerifin, Israel

* Corresponding author. Tel: +972 89779736; Fax: +972 89778412, Email: [email protected]

A 45-year-old man was admitted due to fever and chills, and blood cultures were positive for Streptococcus mitis. Echocardiography revealed a mobile vegetation of 6 mm on the ventricular aspect of the anterior mitral leaflet, a three-leaflet aortic valve with a lack of co-optation, and a moderate aortic regurgitant jet directed towards the anterior mitral leaflet (Panel A, and see Supplementary data online, Video S1A and B). The patient was treated with ampicillin and gentamycin for 1 week and was discharged for homebased ceftriaxone treatment. One week later, he was admitted again with recurrent chills, temperature of 38.58C, and C reactive protein of 153 mg/ L, and CT scan of the head and abdomen was normal. Echocardiography revealed significant aortic regurgitation, the vegetation on the mitral valve grow up to 1.1 cm (Panel B, and see Supplementary data online, Video S2A and B). Urgent surgery was considered, but patients’ condition improved with switch to penicillin and gentamycin. A follow-up echocardiogram before planned discharge detected a mobile echodense vegetation (Panel C, and see Supplementary data online, Video S3A and B). On the next day, the 30th day since the initial presentation, the patient complained of severe pain in his right thigh. Echocardiography was remarkable for the absence of vegetation (Panel D, and see Supplementary data online, Video S4A and B). Septic emboli was suspected and the Gallium-67 citrate SPECT and low-dose CT showed pathological uptake in the right side of sacrum and in the right thigh (Panels E–G), consistent with pelvic osteomyelitis and infection of soft tissue of the right thigh. The patient was treated conservatively. This is an unusual case of late septic emboli 1 month after the beginning of appropriate antibacterial therapy. Supplementary data are available at European Heart Journal – Cardiovascular Imaging online. Published on behalf of the European Society of Cardiology. All rights reserved. & The Author 2014. For permissions please email: [email protected].