Relative apical sparing in cardiac amyloidosis is not always explained by an amyloid gradient

Abstract

Aims

Myocardial longitudinal strain (LS) by two-dimensional (2D) speckle-tracking echocardiography has a diagnostic and prognostic role in cardiac amyloidosis (CA). Typically, the apical segments of the left ventricle (LV) are less affected by LS abnormalities, a finding called relative apical sparing (RELAPS). Whether a variable burden of CA might explain the RELAPS remains unknown.

We aimed to evaluate the extent, distribution, and deposition pattern of amyloid in autopsy hearts of CA patients and to correlate the histopathology findings with 2D echocardiography.

Methods and results

This is a retrospective study of whole heart specimens of CA patients who died and underwent autopsy and 2D echocardiography. Amyloid burden quantification was assessed by histomorphometry in each segment at different LV levels. The LS analysis results were compared with the amyloid burden and the base-to-apex distribution.

Histopathology investigation of 27 hearts with CA [immunoglobulin light chains (AL) 17 cases and transthyretin (ATTR) 10 cases] demonstrated an amyloid base-to-apex gradient. In 11 CA patients with 2D echocardiography, analysis of LS and histological amyloid burden allowed to identify different patterns: RELAPS (8 cases, 73%), with (2) or without (6) amyloid gradient, normal or mildly reduced LS with diffuse low amyloid (2, 18%), and severely reduced LS with diffuse high amyloid (1, 9%).

Conclusion

The typical RELAPS pattern at echocardiography is not always explained by a base-to-apex gradient of amyloid burden at histopathology, suggesting that RELAPS might be an epiphenomenon of complex interactions among amyloid infiltration, myocardial structure, and adaptation.

Graphical Abstract

Graphical Abstract.

Introduction

Amyloidosis includes an under-diagnosed but increasingly recognized group of disorders characterized by the extracellular deposition of misfolded proteins in one or more organs. Cardiac involvement is the major contributor to prognosis in patients with systemic amyloidosis^1,2 and is mainly caused by the immunoglobulin light chains (AL) and transthyretin (ATTR) misfolding and amyloidogenesis.^3,4 A high index of suspicion is needed to achieve the correct diagnosis, and two-dimensional (2D) echocardiography is considered a cornerstone for the identification and follow-up of patients with known or suspected cardiac amyloidosis (CA).^5,6 The most renowned echocardiographic markers of CA are the increased left ventricular (LV) wall thickness with severe LV diastolic dysfunction and restrictive filling pattern.⁵ In early disease stage of CA, LV ejection fraction (LVEF) is generally preserved. However, LV longitudinal strain (LS), measurable using 2D speckle-tracking echocardiography,⁷ can be significantly impaired and also characterized by a typical relative apical sparing (RELAPS) pattern.⁸ RELAPS is common in patients with CA and is included among the ‘red flags’ useful for the differential diagnosis with other causes of increased LV wall thickness.⁹ Despite the increasing focus on this specific marker of CA, an etiological clarification for this echocardiographic sign is yet to be determined. A base-to-apex gradient in myocardial amyloid burden has been previously proposed as a possible explanation of the RELAPS pattern,^10,11 but serial clinicopathological correlations are missing. Hence, the aim of the present study was to investigate the histopathological basis of the RELAPS phenomenon in whole heart specimens of patients affected by CA.

Methods

Study design and population

The study included consecutive patients who died at the University Hospital of Padua with an established in vivo diagnosis of CA from July 2005 to November 2021 and underwent complete autopsy. Exclusion criterion was non-availability of the whole heart specimen or inadequate preservation. Written informed consent was obtained from all participants for echocardiography. Consent to autopsy by the patient’s family is not required in Italy.

Pathology

During the post-mortem dissection, a first mid-ventricular transverse section (a so-called short-axis section) was made in the unfixed heart. A standard gross examination of the heart was performed. The hearts were routinely fixed with 10% buffered formalin. After fixation, two additional transverse sections at the level of the ventricles were performed so that, overall, a basal, mid-ventricular, and apical slice were obtained. Each transverse section for all specimens was entirely processed for histology, with LV samples matching the standard model used for 2D echocardiography.¹² The samples were routinely processed for histology and paraffin embedded. For evaluation of amyloid deposition, three µm-thick serial sections were obtained for haematoxylin–eosin and sulphated alcian blue stain, as previously described.¹³ For each cardiac specimen, the localization of amyloid fibrils was determined, in terms of prevalent interstitial pattern (PIP) or prevalent vascular pattern (PVP). Amyloid distribution was assessed by a histomorphometric quantitative analysis. The extent of amyloid was expressed as percentage of the examined myocardial area. Each LV sample (except for the septal ones) was subdivided in four layers: subepicardial, mid-mural, subendocardial, and trabecular. The interventricular septum (IVS) samples were instead partitioned in three layers: right ventricular (RV)-side, mid-mural, and LV-side. The presence of replacement-type fibrosis was similarly quantified in each ventricular sample by a histomorphometric quantitative analysis and expressed as a percentage of myocardial area.

Amyloid typing

Amyloid typing was obtained by immune electron microscopy on formalin-fixed paraffin-embedded blocks after dewaxing and resin-embedding. Selected sections were then processed for post-embedding immunogold. The primary antibodies used were anti-human kappa light chains, anti-human lambda light chains, and anti-human TTR.

Echocardiography

Echocardiographic images were acquired using a Vivid 7 or Vivid 9 ultrasound system (GE Medical Systems, Milwaukee, USA), and analysis was independently carried out in post-processing by a trained cardiologist blinded to the pathology findings. Inadequate image quality was defined in the presence of a frame rate <40 frames per second or inability to visualize or perform adequate speckle-tracking analysis on more than two myocardial segments. IVS thickness, LV end-diastolic diameter (LVEDD), and posterior wall (PW) thickness were measured on 2D images at the level of the mitral valve leaflet tips in diastole. LVEF was calculated using the biplane Simpson method. Diastolic function was evaluated using the E/A and E/e′ ratios, with conventional and tissue Doppler traces recorded in the apical four-chamber view. All measurements were in accordance with the American Society of Echocardiography guidelines.¹² Peak systolic LS assessment was performed using the EchoPAC software v. 113 (GE Medical Systems, Milwaukee, USA). The three standard apical four-chamber, two-chamber, and long-axis views were used for LS measurements. A polar plot of segmental peak systolic LS values was generated by the software. Global LS (GLS) was calculated using the average of the segmental LS values. Strain values for the six basal, six mid, and five apical segments of the LV were averaged to obtain three regional LS values (basal, mid, and apical). Relative apical LS was calculated as relative apical LS = average apical LS/(average basal LS + average mid LS), and a value of 1 was used as a cut-off for the presence of RELAPS.⁸ Qualitative assessment (i.e. visually identified RELAPS) was not considered.

Considering that echocardiographic RELAPS is a pattern of LS and to avoid any bias due to single segmental analysis, an adjunctive pattern analysis of the relationship between LS and segmental amyloid deposition was performed; accordingly, every LS polar plot was visually compared with the correspondent LV segmental amyloid deposition polar diagram.

Statistical analysis

Continuous variables were expressed as median with 25th and 75th percentiles (Q1–Q3) and analysed by Student’s t-test (or analysis of variance) or Mann–Whitney U test for parametric and non-parametric variables, respectively. Categorical variables were expressed as absolute numbers and percentages and were compared using the χ² test or Fisher exact test, when appropriate. Pearson and Spearman correlation analysis were used to test the association between continuous variables. Friedman test was applied for comparison of means between subgroups. All statistical analyses were performed using IBM SPSS Statistics 27.0 package and Jamovi (version 2.3).

Results

Among 2932 consecutive autopsies performed during the study period, CA was diagnosed in 67 patients (2.3%) and the heart specimen was available for further examination in 27 cases (18 males; median age 71 years, range 45–93 years). Amyloid typing by immunogold technique on transmission electron microscopy performed on formalin-fixed paraffin-embedded cardiac samples after autopsy identified 17 cases of AL-CA (63%) and 10 ATTR-CA (37%).

Histopathology

Detailed gross and histopathological data are reported in Table 1. At gross examination, heart weight ranged from 400 to 1400 g, with a median value of 605 g. The basal LV section had the greatest median ventricular wall thickness (15 mm), compared with the mid-ventricular (13 mm) and apical section (9 mm). At histology, amyloid deposition was observed with PIP in 21 cases (78%), whereas a PVP was documented in 6 cases (22%). Pattern of deposition was further characterized in interstitial segmental in 18 cases (67%) and interstitial diffuse in 9 (33%). No evidence of major epicardial coronary artery involvement by amyloid deposits was observed.

Table 1.

General and histopathological characteristics of the study population

	All (Study population)	PIP	PVP	P	AL-CA	ATTR-CA	P
	n = 27	n = 21	n = 6		n = 17	n = 10
Age (years) (range)	71 (45–93)	71 (45–88)	69 (61–93)	0.64	63 (57–66)	83 (79–88)	0.001
Sex	M: 18 (67) F: 9 (33)	M: 13 (62) F: 8 (38)	M: 5 (83) F: 1 (17)	0.63	M: 13 (76) F: 4 (24)	M: 5 (550) F: 5 (50)	0.22
CA subtype	AL: 17 (63) ATTR: 10 (37)	AL: 12 (57) ATTR: 9 (43)	AL: 5 (68) ATTR: 1 (17)	0.36	AL: 17 (100) ATTR: 0	AL: 0 ATTR: 10 (100)	—
Gross analysis
Heart weight, g	605 (510–700)	605 (510–700)	600 (500–650)	0.88	575 (505–640)	685 (550–750)	0.10
LV thickness
Basal, mm	15 (13–16)	16 (15–16)	13 (11–14)	0.012	15 (13–16)	16 (13–16)	0.60
Mid-ventricular, mm	13 (11–14)	13 (12–15)	11 (9–12)	0.015	12 (10–13)	13 (13–14)	0.29
Apical, mm	9 (7–11)	10 (8–11)	8 (7–8)	0.24	9 (8–11)	9 (7–10)	0.46
Histomorphometric analysis
Amyloid burden per segment, %	25.30 (6.72–57.54)	46.16 (14.64–60.57)	3.69 (2.54–6.10)	<0.001	24.13 (6.81–51.56)	43.58 (7.71–49.18)	0.07
Fibrosis burden per segment, %	0.50 (0.00–1.33)	0.31 (0.00–1.11)	1.81 (0.96–3.68)	<0.001	0.52 (0.00–1.32)	0.45 (0.00–1.54)	0.69

Categorical values are reported as n (%), and continuous values are reported as median (25th–75th). M, male; F, female; AL, light-chain amyloidosis; ATTR, transthyretin amyloidosis; CA, cardiac amyloidosis; LV, left ventricular; PIP, prevalent interstitial pattern; PVP, prevalent vascular pattern.

Histopathology: segmental analysis

Overall median amyloid infiltration per each ventricular segment was 25.3% (6.72–57.54), and it was greater in patients with PIP compared with those with PVP (46.16% vs. 3.69%, P < 0.001) and in ATTR-CA vs. AL-CA (43.58% vs. 24.13%) (although not statistically significant, P = 0.07). Fibrosis per each ventricular segment was 0.5% (0–1.33), similar between ATTR- and AL-CA (0.45% vs. 0.52%, P = 0.69), and greater in patients with PVP compared with those with PIP (1.81% vs. 0.31%, P < 0.001).

Amyloid burden quantification analysis divided per basal, mid-ventricular, and apical segments and subepicardial, mid-mural, subendocardial, and trabecular layers is reported in Table 2.

Table 2.

Amyloid burden quantification at histology

	n	Basal	Mid-ventricular	Apical	P
All	27	25.38 (11.35–59.04)	26.70 (6.90–56.07)	18.80 (4.02–59.35)	0.03
Trabecular		29.75 (19.38–65.25)	27.00 (9.88–61.38)	31.67 (5.17–60.83)	0.11
Subendocardial		22.75 (10.69–61.19)	25.75 (8.06–60.50)	19.31 (4.50–59.31)	0.07
Mid-mural		26.25 (9.25–53.56)	24.13 (6.88–53.19)	17.38 (3.81–55.06)	0.001
Subepicardial		26.25 (9.50–53.75)	28.38 (8.31–54.13)	20.00 (3.56–56.69)	0.011
PIP	21	42.60 (17.02–59.71)	46.55 (16.27–58.02)	45.96 (11.53–62.82)	0.09
Trabecular		48.50 (28.00–67.75)	55.50 (22.00–63.00)	50.70 (14.08–67.42)	0.16
Subendocardial		48.25 (16.38–62.50)	53.38 (16.13–63.00)	50.25 (13.75–63.69)	0.14
Mid-mural		42.38 (16.13–57.25)	41.50 (14.38–53.63)	39.63 (7.38–58.38)	0.09
Subepicardial		33.63 (15.00–56.00)	38.38 (20.21–56.12)	38.50 (9.13–59.56)	0.07
PVP	6	4.03 (2.93–7.94)	4.07 (2.28–6.44)	3.48 (2.42–4.15)	0.14
Trabecular		5.25 (3.50–14.31)	5.13 (3.31–13.31)	5.17 (3.42–5.92)	0.61
Subendocardial		3.88 (2.69–7.69)	4.06 (2.66–6.31)	3.75 (2.81–4.50)	0.40
Mid-mural		3.50 (2.06–5.41)	3.56 (2.25–4.59)	2.63 (2.06–3.56)	0.30
Subepicardial		4.31 (3.31–5.69)	3.63 (2.66–4.50)	2.38 (2.00–3.31)	0.31
ATTR-CA	10	49.15 (13.94–71.57)	35.81 (7.71–64.82)	37.13 (4.44–65.61)	0.008
Trabecular		60.63 (24.31–80.19)	42.88 (8.63–69.63)	44.17 (4.00–76.08)	0.048
Subendocardial		47.50 (13.40–73.00)	35.19 (6.88–66.88)	38.50 (3.50–68.56)	0.007
Mid-mural		46.81 (11.81–68.22)	30.88 (6.68–60.34)	29.63 (4.25–66.06)	0.014
Subepicardial		41.06 (11.25–70.31)	36.00 (8.72–62.63)	32.63 (3.25–62.13)	0.027
AL-CA	17	23.03 (7.94–53.26)	26.70 (9.84–51.83)	18.80 (4.03–48.66)	0.14
Trabecular		28.00 (14.25–60.50)	27.00 (15.75–57.50)	24.67 (5.95–54.58)	0.31
Subendocardial		21.75 (8.75–55.00)	25.75 (10.00–53.50)	19.88 (4.50–53.56)	0.74
Mid-mural		24.13 (5.63–44.37)	24.13 (8.60–43.50)	17.38 (3.63–47.25)	0.62
Subepicardial		20.13 (5.88–51.25)	28.38 (9.13–51.50)	20.00 (3.68–47.56)	0.83

All values are reported in median (25th–75th) of percentages.

AL, light-chain amyloidosis; ATTR, transthyretin amyloidosis; CA, cardiac amyloidosis; PIP, prevalent interstitial pattern; PVP, prevalent vascular pattern.

In the whole population, amyloid burden was significantly greater in the basal (25.38%) and mid-ventricular levels (26.70%) compared with the apical ones (18.80%) (P = 0.030) (Figure 1).

Figure 1.

Longitudinal and transmural amyloid distribution at histology. (A) Amyloid burden at histology in different ventricular levels (basal vs. mid-ventricular vs. apical, P = 0.03). (B) Transmural distribution of amyloid burden at histology in the basal level, (C) in the mid-ventricular level, and (D) in the apical level (B, C, D, P < 0.001). trab, trabecular layer; sub-endo, subendocardial layer; sub-epi, subepicardial layer.

The amyloid distribution was diffusely low in 11, diffusely high in 7, heterogeneous in 4, and with a base-to-apex gradient in 5.

In patients with ATTR-CA, amyloid burden was significantly greater in the basal level (49.15%), compared with the mid-ventricular (35.81) and apical ones (37.13%) (P = 0.008), and the difference was confirmed in all layers. However, when considering separately patients with PIP, PVP, or with AL-CA only, no differences among segments were detected, neither overall nor according to different layers (Table 2). In all groups, a transmural gradient in amyloid burden was found, with more deposits in the trabecular and subendocardial layers compared with the mid-mural and subepicardial ones (Table 2 and Figure 1).

Echocardiography study

Detailed 2D echocardiography data including LS evaluation were available in 11/27 patients (n = 8 with PIP and n = 3 with PVP; n = 7 with AL-CA and n = 4 with ATTR-CA) and are showed in Table 3. No LV segments were excluded in the analysis. No differences in all echo parameters among PIP vs. PVP groups or AL-CA vs. ATTR-CA were detected. The speckle-tracking analysis in all patients revealed a reduced GLS (−8%), with a RELAPS phenomenon observed in 8/11 (73%). A significant base-to-apex gradient of GLS was found in the whole echo population (P < 0.001), PIP (P = 0.010), AL-CA (P = 0.021), and ATTR-CA (P = 0.039) subgroups (Table 3).

Table 3.

Echocardiographic results

	All (echo subgroup)	PIP	PVP	P	AL-CA	ATTR-CA	P
	n = 11	n = 8	n = 3		n = 7	n = 4
LVEDVi, mL/m2	51 (47–66)	49 (47–57)	66 (62–96)	0.13	62 (47–66)	49 (47–94)	0.85
LVEF, %	54 (41–56)	55 (48–56)	41 (27–59)	0.77	53 (41–56)	56 (40–57)	0.64
IVSd, mm	17 (14–18)	18 (14–19)	14 (11–18	0.38	17 (14–18)	17 (13–21)	0.63
LVEDD, mm	45 (39–57)	42 (37–52)	51 (45–59)	0.28	44 (35–51)	52 (43–61)	0.26
PWTd, mm	13 (11–17)	14 (12–18)	12 (9–16)	0.38	13 (11–17)	14 (11–18)	0.85
RWT	0.63 (0.46–0.72)	0.66 (0.55–0.75)	0.53 (0.31–0.63)	0.13	0.63 (0.53–0.77)	0.56 (0.42–0.69)	0.71
LV mass/BSA (g/m2)	146 (112–194)	148 (104–191)	132 (122–175)	0.84	137 (112–150)	188 (108–414)	0.35
E/e′	20 (14–33)	15 (13–23)	33 (18–35)	0.27	20.2 (13.6–33.0)	18.8 (13.1–33.0)	1.00
E/A	1.8 (1.2–3.0)	1.8 (1.2–3.6)	1.5 (0.5–2.5)	0.64	1.8 (1.1–2.5)	3.1 (1.2–4.9)	0.51
GLS (−%)	8.0 (4.3–12.3)	4.9 (2.8–7.5)	1.7 (0.5–13.2)	0.63	3.7 (1.7–7.3)	5.4 (1.9–9.9)	0.79
Basal LS (−%)	5.5 (2.5–7.5)	6.0 (3.0–7.5)	3.5 (1.0–12.5)	1.00	5.5 (3.5–7.5)	5.0 (0–9.5)	0.79
Mid-ventricular LS (−%)	8.0 (5.5–9.0)	8.0 (6.5–8.5)	5.5 (1.5–14.0)	0.49	7.0 (5.5–9.0)	8.0 (6.3–11.0)	0.65
Apical LS (−%)	12.0 (11.0–18.0)	12.0 (11.0–16.0)	15.0 (9.0–19.0)	0.63	14.0 (11.0–18.0)	11.5 (8.5–15.0)	0.41
RELAPS	8 (73)	6 (75)	2 (67)	1.00	6 (86)	2 (50)	0.49

RELAPS is expressed in n (%) and all other values in median (25th–75th).

AL, light-chain amyloidosis; ATTR, transthyretin amyloidosis; BSA, body surface area; CA, cardiac amyloidosis; GLS, global longitudinal strain; IVSd, interventricular septum in diastole; LVEDD, left ventricular end-diastolic diameter; LVEDVi, left ventricular end-diastolic volume index; LVEF, left ventricular ejection fraction; PIP, prevalent interstitial pattern; PVP, prevalent vascular pattern; PWTd, posterior wall thickness in diastole; RWT, relative wall thickness; RELAPS, relative apical sparing.

Echocardiography–histopathology correlation

A significant correlation between amyloid burden and IVS and PW thickness was observed (r = 0.83 and r = 0.73, respectively; P < 0.001 for both; see Figure 2). While no significant inverse correlation between global amyloid burden and LVEF was found (r = −0.267, P = 0.428), a direct correlation between segmental amyloid burden and GLS in each segment was evident (r = 0.409, P = 0.018), with even more significant values when only PIP patients were considered (r = 0.616, P = 0.001) (Figure 2E and F).

Figure 2.

Clinicopathological correlations between echocardiographic parameters and amyloid burden at histology. (A) Correlation between histologically assessed amyloid deposition at the level of the IVS and the IVS thickness evaluated at echocardiography. (B) Correlation between LV amyloid burden at histology and LVEF. (C–D) The same correlations as in A–B but restricted to patients with a PIP of amyloid deposition. (E) Correlation segment by segment between LV amyloid deposition at base, mid-ventricular, and apical level and LS at echocardiography. (F) The same correlation as in E but restricted to patients with a PIP of amyloid deposition.

When comparing values of LS and histological amyloid burden in the LV (Figure 3, Table 4), two patients, who had normal or mildly reduced LS and no evidence of RELAPS at 2D echocardiography, were characterized by diffuse low percentages of amyloid deposition. One patient, who had severely reduced LS and no evidence of RELAPS at 2D echocardiography, showed diffuse high percentages of amyloid deposition. Among the eight patients with the RELAPS phenomenon at 2D echocardiography, two (25%) showed a gradient of amyloid deposition, greater in the basal and middle than in the apical segments, and six patients had no histological gradient of amyloid deposition. More in detail, three patients had a diffusely high amyloid burden; two patients had a diffusely low amyloid burden, and one patient had a heterogeneous amyloid build-up. Furthermore, we assessed the correlation between transmural gradient and RELAPS or subendocardial-only content and RELAPS (see Supplementary data online, Table) without evidence of statistically significant differences.

Figure 3.

Echo-histological patterns of amyloid distribution. Different patterns of correlation between echocardiographic findings and histologically assessed amyloid distribution are represented: normal or mildly reduced LS, diffuse low amyloid, without RELAPS at echocardiography, and with diffuse low percentages of amyloid deposition at histology; severely reduced LS, diffuse high amyloid, without RELAPS at echocardiography, and with diffuse high percentages of amyloid deposition at histology; RELAPS, amyloid gradient, with RELAPS phenomenon and a base-to-apex gradient of amyloid deposition; and RELAPS, no amyloid gradient, with diffusely low amyloid deposition, heterogeneous amyloid deposition, and diffuse high amyloid deposition. All patterns are illustrated with a sequence of echocardiographic LS polar plot, histological polar plot equivalent of amyloid deposition, panoramic view of basal and apical LV sections, and close-up of the same sections (scale bar 100 µm).

Table 4.

Clinical and pathological details of the study population (11 patients with echocardiography)

Pattern	Sex	Age of death	Time echo—death (months)	Amyloid type	Specific therapy	History of ACS or CAD	Cause of death	HR (bpm)	SBP (mmHg)	DBP (mmHg)	IVS thickness (mm)	PW thickness (mm)	iEDV (mL/mq)	EF (%)	Echo pattern	Amyloid histological pattern	Amyloid histological burden
Normal or mildly reduced LS—diffuse low amyloid	F	82	5	ATTRwt	No	Yes	Septic shock (endocarditis)	80	130	80	11	9	45	46	LS normal or mildly reduced	PIP	Diffuse low
	M	65	10	AL lambda	CyBorD	No	Septic shock	63	140	70	14	12	62	59	LS normal or mildly reduced	PVP	Diffuse low
Severely reduced LS—diffuse high amyloid	M	75	5	ATTRwt	No	No	Acute heart failure	65	135	75	26	21	138	25	LS diffusely reduced	PIP	Diffuse high
RELAPS—amyloid gradient	F	88	1	ATTRwt	No	Yes	ACS-STE	65	145	70	14	13	51	57	RELAPS	PIP	Base-to-apex gradient
	M	79	2	AL lambda	No	No	Acute heart failure	92	125	80	18	16	66	41	RELAPS	PVP	Base-to-apex gradient
RELAPS—no amyloid gradient	F	73	< 1a	AL lambda	No	No	Cardiogenic shock	80	120	80	11	9	96	27	RELAPS	PVP	Diffuse low
	F	83	10	ATTRwt	No	No	Acute heart failure	96	130	80	22	15	47	56	RELAPS	PIP	Diffuse high
	M	57	1	AL lambda	CyborD	No	Acute heart failure	93	100	60	18	19	52	43	RELAPS	PIP	Diffuse high
	F	59	1	AL lambda	CyBorD	No	SCD (refractory FV)	86	110	70	14	11	38	53	RELAPS	PIP	Diffuse low
	M	66	1	AL lambda	BorD	No	SCD (EMD)	56	115	80	17	17	50	54	RELAPS	PIP	Heterogeneous
	M	73	< 1b	AL lambda	No	No	Acute peritonitis	90	100	60	18	13	63	56	RELAPS	PIP	Diffuse high

Two weeks before death.

Three weeks before death.

F, female, M, male; AL, light chain amyloidosis; ATTR, transthyretin amyloidosis; ATTRwt, ATTR wild-type; CyBordD, cyclophosphamide, bortezomib, and dexamethasone; ACS, acute coronary syndrome; CAD, coronary artery disease; DBP, diastolic blood pressure; HR, heart rate; IVS, interventricular septum; PW, posterior wall of the left ventricle; iEDV, indexed end-diastolic volume; EF, ejection fraction; EMD, electromechanical dissociation; PIP, prevalent interstitial pattern; PVP, prevalent vascular pattern; SBP, systolic blood pressure.

Discussion

The aim of our clinicopathological study was to investigate the histopathological basis of the RELAPS phenomenon in whole heart specimens of patients affected by CA. The main study results were (i) the amyloid burden as assessed at the microscope was significantly greater in the basal and mid-ventricular than in the apical levels (base-to-apex gradient); (ii) a transmural gradient in amyloid burden was also present, with more deposits in the trabecular and subendocardial compared with the mid-mural and subepicardial layers; (iii) the amyloid deposition was observed both in the interstitium and in the intramyocardial vessels, but in about one-fourth of cases, the deposition was prevalent in the vessels, a finding predominantly observed in AL-CA; and (iv) 2D echocardiography data, besides confirming a significant correlation between amyloid burden and LV wall thickness, showed a RELAPS phenomenon in nearly three-quarters of cases and a significant base-to-apex gradient of LS in the whole echo population. A direct correlation between segmental amyloid burden and LS in each segment was evident, with even more significant values when only patients with PIP were considered. Myocardial mass was higher than normal, with thickened LV ventricular wall in nearly all cases. At echocardiography, increased LV wall thickness with preserved EF and severe diastolic dysfunction was found, in keeping with the classical phenotype reported for CA.^5,6,9

Despite their intrinsic heterogeneity, AL-CA and ATTR-CA did not manifest significant differences except for the age at presentation. The well-known dissimilarities mainly related to age at diagnosis (higher in ATTR-CA), LV wall thickness (higher in ATTR-CA), and survival (worse in AL-CA)^4,14 are mostly unidentifiable in our study population, most probably due to its small size and study design. However, a non-significant trend is recognizable with ATTR-CA having greater mean heart weight and mass and a slightly more severe amyloid burden per segment. A greater difference in the amyloid and fibrosis burden per segment can be clearly recognized when comparing the two deposition patterns (PIP and PVP), with PVP related to less amyloid accumulation but more severe (even if minimally) myocardial fibrosis.

Three main hypotheses have been advanced to explain the RELAPS phenomenon so far: (i) differences in amyloid level deposition, greater in the base than in the apex; (ii) differences in myocytes and extracellular matrix orientation between base and apex; and (iii) greater tendency towards apoptosis and remodelling in basal regions, due to turbulent flow in the LV outflow tract.^10,15 According to the hypothesis of amyloid base-to-apex gradient, the presence of less amyloid deposition at the apex may preserve cardiac myocyte contraction and deformation properties, leading to the relative sparing of apical LS.

Phelan et al. first suggested the amyloid base-to-apex gradient theory, after the echocardiographic observation of a thicker LV wall at the basal and mid-ventricular segments than at the apex in patients with CA. However, no histology validation analysis was available. Similar results were proposed by Ternacle et al., who studied with 2D strain echocardiography and cardiac magnetic resonance a series of patients with CA, showing that LV LS impairment reflected the amyloid burden and late-gadolinium enhancement was associated with LV LS impairment. Of note, a strong negative correlation between the amyloid burden measured by histopathology and segmental LV LS in different types of CA was also highlighted. Although only three explanted hearts were available, amyloid deposits were more abundant in the basal and mid-ventricular sections, which were those with greater LGE and poorer LS. A similar histopathological explanation for the RELAPS substrate has been provided only by Sawada et al.¹¹ with a single case report showing a base-to-apex gradient of amyloid deposition, even if with limited methodology (a single histological section was sampled at each level, namely basal, mid-ventricular, and apical). Bravo et al.¹⁶ by studying a series of patients with AL-CA with 2D echocardiography, ¹⁸F-florbetapir positron emission tomography, and cardiac magnetic resonance postulated that the RELAPS is explained by regional differences in total mass of the amyloid deposition rather than the proportion of amyloid deposits. In other words, the total amyloid mass was disproportionately greater towards the basal and mid-ventricular segments than the apex.

In the present clinicopathological study, we specifically addressed the role of the amyloid deposition carrying out for the first time a detailed histomorphometric analysis matching the standard model used for 2D echocardiography. At the level of any single LV segment—not considering if basal, medium, or apical—a significant correlation between percentage of amyloid deposition and LS was found. In addition, in our study, the GLS analysis showed more propensity of AL-CA to develop RELAPS. A contributing role of intramyocardial vessel amyloid deposition is possible, suggesting that RELAPS might be an epiphenomenon of complex interactions among different patterns of deposition, myocardial structure, and consequent adaptation.

Overall, we found that the amyloid burden assessed at histology was significantly greater in the basal and mid-ventricular levels than in the apical one, irrespectively of deposition pattern (PIP or PVP). In detail, the comparative analysis of LS and amyloid burden led to identify four echo-histological patterns, i.e. normal or mildly reduced LS with diffuse low amyloid burden, severely reduced LS with diffuse high amyloid burden, RELAPS with amyloid gradient, and RELAPS without amyloid gradient, the latter with diffuse low, heterogenous, or diffuse high amyloid deposition. Thus, while a base-to-apex gradient of amyloid burden always accounts for the RELAPS phenomenon at 2D echocardiography, this marker was not always explained by a gradient of amyloid in the heart, as demonstrated by histopathological correlation. Furthermore, by comparing the three cases without RELAPS vs. the eight cases with RELAPS, our results are not supporting the theory that the absence of RELAPS should be considered a marker of advanced disease due to the apical LS impairment.

A possible role of transmural amyloid gradient with prevalent subendocardial involvement together with the peculiar vulnerability of the subendocardium to ischaemia¹⁷ should be taken into consideration. A previous extensive histomorphometric analysis demonstrated a similar transmural gradient of amyloid deposition, with trabecular and subendocardial layers being the most infiltrated.¹⁸ This region of the heart is crucial in regulating the contraction and mechanics, suggesting this additional pathophysiological pathway to explain the RELAPS phenomenon. As previously hypothesized by Dorbala et al.,¹⁹ since the majority of longitudinal fibres are subendocardial and this area of the myocardium is most vulnerable to ischaemia, predominant amyloid deposition in this region could play a role in longitudinal impairment.

Furthermore, as suggested by Rapezzi and Fontana,¹⁵ it is possible that the diverse fibre orientation at the apex compared with the base and the preferential involvement of specific fibre subtypes could contribute to the echocardiographic phenomenon of RELAPS. Anatomic studies of the LV fibre architecture have clearly demonstrated the physiological regional variation, showing the clock-wise spiralling of the fibres forming the apex.²⁰

The presence itself of the misfolded proteins has been hypothesized as a cause of direct damage to the cardiomyocytes via apoptotic cell death leading to cardiac dysfunction and subsequent heart failure.²¹ Unfortunately, apoptosis is difficult to evaluate at histology in a quantitative way, since the most frequently used method (called TUNEL) is burdened by high controversy.²² Moreover, our retrospective study of fixed hearts coming from the pathology archives does not allow to perform a reliable investigation of cardiomyocyte apoptosis in the current population. Recently, the RELAPS pattern has been reported as quite frequent in patients with severe symptomatic aortic stenosis,^23,24 even in the absence of pathologically demonstrated CA and with reversibility after surgery.²⁵ These findings are supportive of a putative role of pronounced LV remodelling with predominant IVS thickening at basal level. The role of myocardial fibrosis remains controversial, since the percentage of replacement-type fibrosis could not be reversed by surgery. The recent demonstration of GLS improvement and RELAPS disappearance in treated AL-CA²⁶ could further support the role of LV remodelling and toxic myocardial damage from circulating amyloid precursors.²⁷ Further clinicopathological correlation studies are needed to elucidate the pathological substrates of RELAPS in different clinical settings.

Limitations

Our study has some limitations inherent to its retrospective nature with a relatively small number of heart specimens from patients in which GLS analysis was available. Unfortunately, echocardiogram was available in less than half of the population. However, the population is unique, taking into account the need to find whole preserved specimens with the possibility to investigate multiple sections from base to apex. Overall, by considering the skewed distribution of RELAPS (yes/no), the different aetiology of CA (AL/ATTR), and level of severity of amyloid deposition, it is hard to express any definite conclusion about negative findings.

Conclusion

The present clinicopathological study demonstrated that amyloid is variably distributed in the heart of patients affected by CA, with vascular involvement being more frequent in AL-CA. The amyloid burden at histology was significantly greater in the basal and mid-ventricular levels compared with the apex and in the subendocardial layers compared with the subepicardial ones (i.e. longitudinal and transmural gradients). A direct correlation between segmental amyloid burden and LS in each segment is evident. The RELAPS phenomenon at echocardiography is present in nearly half of cases and is not always explained by a base-to-apex gradient of amyloid burden at histopathology, suggesting that RELAPS might be an epiphenomenon of complex interactions among amyloid infiltration, myocardial structure, and adaptation.

Supplementary data

Supplementary data is available at European Heart Journal - Cardiovascular Imaging online.

Funding

Pfizer Inc. provided financial support to the University of Padova for the present study. Pfizer Inc. had no role in the study design, data analysis, and results interpretation of the present study. Other sources of funding were the Registry for Cardio-Cerebro-Vascular Pathology, Veneto Region, Venice, Italy; Target Projects RSF-2016-2016-02363774, Ministry of Health, Rome, Italy; and University Research grant BIRD2222_01, Padua, Italy.

Data availability

The data underlying this article cannot be shared publicly due to privacy of individuals that participated in the study. The data underlying this article will be shared on reasonable request to the corresponding author.