Diagnostic Accuracy and Prognostic Value of Relative Apical Sparing in Cardiac Amyloidosis　― Systematic Review and Meta-Analysis ―

Abstract

Background: Although the relative apical sparing (RAPS) pattern of left ventricular (LV) longitudinal strain is a hallmark of cardiac amyloidosis, recent studies have raised concerns about its accuracy. The aim of this systematic review was to investigate diagnostic test accuracy (DTA) and prognostic impact of RAPS in cardiac amyloidosis.

Methods and Results: We searched PubMed, Embase, and Scopus for manuscripts that could potentially be used in the DTA arm and prognosis arm. Thirty-seven studies were used for DTA analysis. The pooled sensitivity, specificity, and diagnostic odds ratio were 61% (95% confidence interval [CI] 54–68%), 83% (95% CI 80–86%), and 8.9 (95% CI 6.1–13.1), respectively. These values did not differ regardless of the presence of aortic stenosis, but the diagnostic odds ratio differed significantly among analytical software packages. For the prognosis arm, 6 studies were dichotomously assessed for RAPS, and 5 were assessed quantitatively. The pooled proportion of RAPS was 49% and the pooled estimate of the RAPS ratio was 1.40. Although RAPS was associated with outcome (hazard ratio [HR] 1.87; 95% CI 1.15–3.04; P=0.011), its significance disappeared after trim and fill analysis (HR 1.42; 95% CI 0.85–2.38; P=0.184).

Conclusions: RAPS has a modest DTA with a significant vendor dependency and does not provide robust prognostic information.

The relative apical sparing (RAPS) pattern of left ventricular (LV) longitudinal strain (LS) in cardiac amyloidosis (CA) was first described by Phelan et al. in 2012,¹ and has become the most widely used echocardiographic feature for diagnosing CA. Subsequent studies revealed that a basal-to-apical LS gradient correlates with amyloid burden characterized by more abundant amyloid deposits in the basal and mid-myocardium on histopathology or more gadolinium enhancement at corresponding sites on cardiac magnetic resonance imaging.^2,3 However, high diagnostic test accuracy (DTA) of RAPS has not been consistently achieved in all studies, and some investigations have shown that less than half of CA patients have RAPS (RAPS ratio >1).^2,4 Although these conflicting results could be related to intervendor variability of regional LS measurements,^5,6 which require different cut-off values for the RAPS ratio when different software is used,⁷ recent studies have reported that RAPS is also found in symptomatic severe aortic stenosis (AS) without evidence of CA, and that the RAPS pattern disappeared after aortic valve replacement in most patients who had RAPS before surgery.⁸ Although RAPS is a red flag for suspected CA,⁹ it is not always selected as the parameter in the scoring system for CA diagnosis.^10–12 Thus, it is important to establish the clinical usefulness of RAPS in CA. Accordingly, the aims of this systematic review and meta-analysis were to investigate the DTA and prognostic impact of RAPS in patients with CA.

Methods

Study Design and Patient Involvement

This systematic review and meta-analysis followed Preferred Reporting Items for Systematic reviews and Meta-Analysis (PRSMA) guidelines and was prospectively registered with the University Hospital Medical Information Network (UMIN) Clinical Trial Registry (Registration no. UMIN000053051). Due to the aggregated analysis of published literature, institutional ethics review board approval was not required.

Search Strategy

Literature investigating the DTA of RAPS and the association between RAPS and outcomes was searched for in PubMed, Embase, and Scopus through December 11, 2023. Search key words used are listed in Supplementary Table 1.

Literature Selection

For DTA, we searched for studies that showed sensitivity and specificity of RAPS for diagnosing CA among patients with specific cardiac diseases (e.g., CA, AS, and hypertrophic cardiomyopathy) or patients who were suspected of having CA. According to the consensus from publication year of included articles, the diagnosis of CA was established through endomyocardial biopsy, extracardiac biopsy plus imaging features of cardiac involvement, and cardiac scintigraphy with technetium isotopes. For prognostic value, we searched for studies that described the hazard ratio (HR) and its 95% confidence interval (CI) of RAPS for specific outcomes. We carefully determined whether the authors used RAPS as a dichotomous outcome (yes/no) or a continuous outcome, and conducted separate meta-analyses.

Data Extraction

Each eligible study was reviewed by 2 authors (C.-Y.L., Y.N.). The data extracted from the full text are listed below.

(1) Article: first author’s name, year of publication, journal name, volume, and page number

(2) Demographics: number of study patients, their ages and sex, CA type, number of CAs, numbers of comparators, types of comparators, inclusion of AS patients in the study population, New York Heart Association Class ≥III, and software used for strain analysis

(3) Cardiac function parameters: LV ejection fraction, LV global LS, and RAPS ratio

(4) DTA parameters: RAPS cut-off value, method used to determine the cut-off value, and numbers of true positives, false negatives, false positives, and true negatives

(5) Prognosis: types of endpoints, follow-up duration and numbers of events, HR and 95% CI of RAPS

Statistical Analysis

Continuous variables are presented as the mean±SD, mean and range, or median and interquartile range, according to data distribution. Categorical variables are presented as numbers or percentages. A random effect model (DerSimonian–Laird) meta-analysis was used, and results were visualized using a forest plot. Trim-and-fill analysis was conducted when Egger’s test was statistically significant in at least 10 studies included in the meta-analysis, or when the number of studies was ≤10 as a sensitivity analysis.^13,14 Subgroup analysis and meta-regression analysis were also performed. For DTA analysis, pooled sensitivities, specificities, and diagnostic odds ratios (DOR) were calculated, and a summarized receiver operating characteristics (ROC) curve was constructed.¹⁵ Fagan’s nomogram and conditional probability plot were also created.^16,17 For prognostic analysis, pooled HRs were calculated, and forest plots were generated. Methodological quality assessment of included studies was performed using the Quality Assessment tool for Diagnostic Accuracy Studies (QUADAS) or the Quality In Prognosis Studies (QUIPS) tool.^18,19 Two-sided P<0.05 was used to determine statistical significance, whereas P<0.1 was used for Egger’s test. Statistical analyses were performed using R version 4.4.0 (R Foundation for Statistical Computing, Vienna, Austria) with the “meta,” “metafor,” “mada,” and “TeachingDemo” packages and RStudio (2023.09.1-494; Posit Software, PBC, Boston, MA, USA).

Results

A PRISMA flowchart is presented in Supplementary Figure 1. Regarding DTA, we retrieved 2,743 articles, from which 35 articles and 37 studies were used for the analysis. For the prognostic arm, we retrieved 2,207 articles, from which we ultimately selected 10 articles and 11 studies. Quality assessment of the included studies is summarized in Supplementary Figure 2.

DTA Arm

Patient characteristics are presented in Table 1. In all, 2,872 CA patients and 3,836 comparators were included. CA prevalence in each study was widely distributed, ranging from 5.3% to 79.5%, with a pooled estimate of 39.6%. At least 1,358 AS patients with or without CAs were included from 15 studies, in which 1,231 patients were diagnosed with severe AS. Three speckle tracking analytical software packages with echocardiography were used for strain analysis (GE Healthcare: EchoPAC; Philips: QLAB; and TomTec). A RAPS ratio cut-off of 1 was used in 24 studies. Eight studies used cut-off values of <1, and 5 studies used cut-off values of >1. Supplementary Table 2 summarizes the clinical characteristics in each study.

Table 1.

Characteristics of the Study Population in the DTA Arm (37 Studies, n=2,872)

Variable	Summary statistics
Age (years)	72.6 (58–88)
Male sex (n=2,631)	1,442 (55)
Prevalence of CA (%)	39.6 (5.3–79.5)
Type of CA
AL	5 (14)
ATTR	14 (38)
Mixed	16 (43)
Not specified	2 (5)
Type of ATTR
Wild type	10 (72)
Variant	1 (7)
Mixed	3 (21)
Diagnostic modality
Myocardial biopsy + extracardiac biopsy (%) (k=23)	70 (8–100)
Myocardial biopsy (%) (k=19)	33 (6–100)
Tc scintigraphy (%) (k=21)	90 (11–100)
NYHA Class III or IV (n=1,108; k=18)	439 (40)
No. comparators	3,836
Studies included AS	15/37 (41)
Study type
Prospective	13 (35)
Retrospective	24 (65)
Software used for analysis
EchoPAC (GE Healthcare)	26 (75)
QLAB (Philips)	5 (14)
TomTec	4 (11)
Unknown	2
LVEF (%) (k=34)	54.2 (47–62)
LVGLS (%) (k=32)	11.9 (8.4–18.7)
RAPS ratio (k=27)	1.31 (0.70–3.30)

Unless indicated otherwise, data are given as the mean (range) or n (%). AL, light chain amyloidosis; AS, aortic stenosis; ATTR, transthyretin amyloidosis; CA, cardiac amyloidosis; DTA, diagnostic test accuracy; k, number of studies; LVEF, left ventricular ejection fraction; LVGLS, left ventricular global longitudinal strain; NYHA, New York Heart Association; RAPS, relative apical sparing; Tc, technetium.

Sensitivity and Specificity of RAPS

The pooled sensitivity was 61% (95% CI 54–68%) with severe heterogeneity (I²=91%). There was no significant funnel plot asymmetry, as assessed by Egger’s test (P=0.666).

The sensitivity did not differ between studies that included AS as a comparator (60%, 95% CI 40–77%) and those that did not (62%, 95% CI 57–68%; P=0.872). The 3 analytical software packages showed different sensitivity values (TomTec, 30% [95% CI 8–66%]; Philips, 54% [95% CI 32–74%]; GE Healthcare, 67% [95% CI 62–72%]), and significant differences were observed between GE Healthcare and TomTec (P<0.001). Types of CA did not exhibit different sensitivities (light chain amyloidosis [AL]-CA, 66% [95% CI 59–72%]; transthyretin amyloidosis [ATTR]-CA, 56% [95% CI 36–75%]; mixed, 58% [95% CI 49–66%]; P=0.312).

The pooled specificity was 83% (95% CI 80–86%) with severe heterogeneity (I²=83%). There was significant funnel plot asymmetry (P=0.001). Trim-and-fill analysis added 14 studies and the specificity decreased to 76% (95% CI 72–81%). The specificity did not differ, regardless of whether AS was included as a comparator or which type of CA was used. However, the 3 analytical software packages showed different specificity (GE Healthcare, 81% [95% CI 77–85%]; Philips, 96% [95% CI 80–99%]; TomTec, 85% [95% CI 79–89%]), and meta-regression analysis revealed that specificity differed between GE Healthcare and Philips (P=0.034).

DOR of RAPS

The pooled DOR was 8.90 (95% CI 6.06–13.1) with severe heterogeneity (I²=84%). There was significant funnel plot asymmetry (P=0.036). Trim-and-fill analysis added 12 studies and the DOR decreased to 5.14 (95% CI 3.50–7.54). DOR did not differ between studies that included AS population as a comparator (7.34; 95% CI 3.18–17.0) and those that did not (9.48; 95% CI 6.77–13.3; P=0.579), the result was similar after adjusting for LVEF (P=0.358). DOR also did not differ among types of CA (P=0.219). Meta-regression analysis showed that the prevalence of ATTR-CA was not associated with DOR (P=0.331). However, significant differences in DOR were found among various software packages, with the highest DOR occurring with Philips software and the lowest with TomTec software (Figure 1). We performed meta-regression analysis including parameters that may potentially affect diagnostic test accuracy (Supplementary Table 3). DOR between software packages was still significantly different in most comparisons after adjustment. Specifically, DOR was significantly different among software packages after adjusting for the prevalence of ATTR-CA and the prevalence of NYHA Class III or IV (no. studies [k]=15; GE Healthcare vs. Philips, P=0.047; GE Healthcare vs. TomTec, P=0.009; Philips vs. TomTec, P<0.001).

Figure 1.

Forest plot of diagnostic odds ratios (ORs) for 3 strain analytical software packages: EchoPAC (GE Healthcare), QLAB (Philips), and TomTec. CI, confidence interval; FN, false negative; FP, false positive; TN, true negative; TP, true positive.

Because DOR differed significantly among software packages, we performed subgroup analysis in 26 studies in which GE Healthcare software was used for the RAPS calculation. DOR did not differ between studies with (9.22; 95% CI 5.27–16.1) or without (9.25; 95% CI 6.55–13.1) an AS population (P=0.991). DOR did not differ in regard to the type of CA (AL-CA, 15.3 [95% CI 6.47–36.1]; ATTR-CA, 8.18 [95% CI 5.71–11.7]; mixed, 7.03 [95% CI 4.58–10.8]; P=0.284). Meta-regression analysis revealed that there were no significant differences in DOR between AL-CA and ATTR-CA (P=0.308). No association was also noted between prevalence of ATTR-CA and DOR (P=0.316). There were negative associations between publication year and DOR (P=0.042; Supplementary Figure 3).

Figure 2 shows a summary ROC curve for the 3 vendor software packages, showing that the highest area under the curve occurred with Philips software. We created Fagan’s nomogram among the software packages when RAPS was observed in 40% of the pretest probability of CA. If we observed RAPS, the post-test probability of CA increased from 40% to 70% in GE Healthcare software, from 40% to 90% in Philips software, and from 40% to 57% in TomTec software, respectively. If we did not see RAPS, the post-test probability of CA decreased from 40% to 21% in GE Healthcare software, from 40% to 24% in Philips software, and from 40% to 35% in TomTec software, respectively (Supplementary Figure 4). Figure 3 shows a conditional probability plot, showing that Philips software had the highest discriminative ability.

Figure 2.

Summary receiver operating characteristic (SROC) curves comparing various strain analytical software packages: (A) EchoPAC (GE Healthcare), (B) QLAB (Philips), and (C) TomTec. Each circle represents a study. The size of a circle is adjusted by a weight function (1 / TP+1 / FN+1 / FP+1 / TN, where TP is true positive, FN is false negative, FP is false positive, and TN is true negative). The curve represents the summary curve for receiver operating characteristics for relative apical sparing. The black dot represents the summary estimate of test performance. The ellipsoid zone represents a 95% confidence interval (CI) region. AUC, area under the curve; DOR, diagnostic odds ratio.

Figure 3.

Conditional probability plots among various strain analytical software packages: (A) EchoPAC (GE Healthcare), (B) QLAB (Philips), and (C) TomTec. NLR, negative likelihood ratio; PLR, positive likelihood ratio.

Prognosis Arm

Patient characteristics are presented in Table 2. Ten studies, including 11 discrete study populations with 975 CA patients and 423 events, were used for the analysis. Six studies used RAPS as a dichotomous outcome and the other 5 studies used RAPS as a continuous variable; thus, we did not perform Egger’s test for funnel plot asymmetry.²⁰ Detailed clinical information from each study is presented in Supplementary Table 4. The pooled proportion of RAPS was 49% (95% CI 33–65%) with severe heterogeneity (I²=92%). Subgroup analysis showed that the proportion of RAPS differed significantly among software packages (GE Healthcare, 61% [95% CI 45–75%]; TomTec, 43% [95% CI 36–51%]; Philips, 12% [95% CI 6–22%]; P<0.001). The pooled estimate of the RAPS ratio was 1.40 (95% CI 1.05–1.75) with severe heterogeneity (I²=99.6%). Subgroup analysis revealed that there were significant differences in the RAPS ratio among software packages (P<0.001).

Table 2.

Characteristics of the Study Population in the Prognosis Arm (11 Studies, n=975)

Variable	Summary statistics
Age (years)	68.1 (50–81)
Male sex (n=929; k=10)	678 (73)
Type of CA
AL	504 (52)
ATTR	456 (47)
Others	15 (1)
NYHA Class III or IV (%) (k=10)	29 (0–59)
Comorbidities
Hypertension (%) (k=6)	47 (27–70)
Atrial fibrillation (%) (k=6)	39 (17–62)
Coronary artery disease (%) (k=8)	17 (4–45)
NT-proBNP (pg/mL) (k=7)	4,137 (1,670–7,720)
Endpoint
Death	8 (73)
Composite	3 (27)
No. events	423
Median [IQR] follow-up duration (months)	24 [16–31]
Software used for analysis
EchoPAC	6 (55)
Esaote	1 (9)
QLAB	1 (9)
TomTec	3 (27)
LVEF (%) (k=11)	56.4 (49–63.4)
LVGLS (%) (k=11)	12.8 (8.3–16.6)

Unless indicated otherwise, data are given as the mean (range) or n (%). IQR, interquartile range; NT-proBNP, N-terminal pro B-type natriuretic peptide. Other abbreviations as in Table 1.

RAPS had a significant association with outcomes (HR 1.87; 95% CI 1.15–3.04; P=0.011) with moderate heterogeneity (I²=52%). Trim-and-fill analysis added 2 studies and the pooled estimate had no association with outcomes (HR 1.42; 95% CI 0.85–2.38; P=0.184). The RAPS ratio was not associated with outcome (HR 1.28; 95% CI 0.82–2.01; P=0.275) with moderate heterogeneity (I²=66%). Trim-and-fill analysis added 2 studies and the HR further decreased to 0.98 (95% CI 0.60–1.59; P=0.928).

Subgroup analysis according to type of CA revealed that RAPS was significantly associated with outcomes (HR 1.87; 95% CI 1.03–3.41; P=0.041) with moderate heterogeneity (I²=47%) in AL-CA but not in ATTR-CA (HR 2.25; 95% CI 0.19–27.2; P=0.524; Figure 4A). Meta-regression analysis demonstrated that RAPS in AL-CA was still significantly associated with outcomes after adjusting for the prevalence of NYHA Class III or IV (P=0.012). In contrast, the RASP ratio was not associated with outcomes in both AL-CA and ATTR-CA (Figure 4B).

Figure 4.

Forest plot of hazard ratios (HRs) presenting relative apical sparing (RAPS) as (A) a dichotomous variable and (B) a continuous variable according to the type of cardiac amyloidosis (CA). One study (Tjahjadi et al.) presenting RAPS as a dichotomous variable was removed from analysis because it had a mixed group of CA patients. Another study (Pislaru et al.) with a mixed group of patients presenting RAPS as a continuous variable was classified as being a study on light chain amyloidosis (AL)-CA because most patients (40/48) had AL-CA.

Discussion

The main findings of this meta-analysis are summarized as follows: (1) RAPS had a modest DTA for diagnosing CA; (2) DTA did not differ between studies in which an AS population was included as a comparator and those in which it was not; (3) DTA differed among vendor analytical software packages, showing highest DOR with Philips software and lowest DOR with TomTec software; (4) DOR was negatively correlated with publication year in studies which GE Healthcare software was used for the RAPS analysis; (5) RAPS, but not the RAPS ratio, was associated with adverse outcomes; (6) RAPS was not associated with outcomes in ATTR-CA, but was in AL-CA.

Diagnostic Test Accuracy

Overall, the pooled sensitivity, specificity, and DOR were 61%, 83%, and 8.90, respectively, reflecting the moderate diagnostic accuracy of RAPS for CA, even though 40% of studies used optimal RAPS cut-off values according to ROC analysis. Meta-regression analysis revealed a greater DOR in earlier years followed by a decrease in DOR in recent years in 26 studies that used the same vendor software (GE Healthcare) for the strain analysis (Supplementary Figure 3), suggesting that there is increased awareness and recognition of CA and that we are now diagnosing patients at earlier disease stages using non-invasive technetium-labeled cardiac scintigraphy in ATTR-CA. Most importantly, we found that DOR differed among ultrasound speckle tracking software packages. DOR was highest (59.6) in Philips (QLAB), followed by GE Healthcare (EchoPAC; DOR=9.3), and lowest in TomTec (DOR=2.8). Although different baseline characteristics, such as types and clinical stage of CA, as well as NYHA classification may affect DTA, the trend did not change after adjusting for several parameters that may potentially affect DTA (Supplementary Table 3). This supports the notion that the specific software package used for analysis, and not baseline profile, mainly affects the DTA. Our results are concordant with those of a prior study.⁷ Jiang et al. reported that the average basal LS and average mid-LS were similar between GE Healthcare and TomTec software in 48 patients with CA; however, average apical LS was significantly lower in TomTec compared with GE Healthcare software, resulting in a significantly lower magnitude of RAPS measured by TomTec software.⁷ This was associated with a lower diagnostic accuracy of TomTec than GE Healthcare software for the detection of CA. Although efforts of the American Society of Echocardiography and the European Association of Cardiovascular Imaging Task Force have reduced intervendor variability of LV global LS measurements, segmental and regional LS measurements still exhibit high variability among software vendors.^5,6

Because the sensitivity, specificity, likelihood ratios, and DOR depend on the prevalence of CA, we measured the post-probability of CA when we observed the RAPS pattern under the pretest prevalence of CA, which was assumed to be 40% (which was equivalent to the pooled estimate of the CA incidence in this meta-analysis) with each type of software. When we observed RAPS, the post-probability increased to 90% in Philips software and 70% in GE Healthcare software, but remained at 57% in TomTec software. Thus, even though we observe a red flag (RAPS), the post-test probability of the CA could differ not only from the pretest probability of CA, but also the software used for the strain analysis.

AS is a disease of aging, and the coexistence of CA has been reported at a rate of 13–15%.^21–23 A previous meta-analysis of the echocardiography diagnosis of CA in AS patients reported that LV wall thickness and LV mass were significantly higher, and that LV and right ventricular function were significantly lower, in AS patients with than without CA.²⁴ However, only 2 studies addressed RAPS, and there were no significant differences in the RAPS ratio in the 2 groups (0.97 vs. 0.90). Among the 26 studies that used the same vendor software, 8 included AS as a comparator. Meta-regression analysis revealed no differences in DOR between the groups with and without AS as a comparator, even after adjusting for LVEF. Thus, AS itself does not adversely affect the DTA of RAPS for diagnosing CA.

A recent study investigated the impact of tafamidis on echocardiographic parameters in patients with ATTR-CA.²⁵ No significant changes were observed in any parameters in that study, including global LS and RAPS, after the administration of tafamidis,²⁵ which implies that changes in the RAPS ratio may not reflect the effectiveness of specific therapy for ATTR-CA.

Prognostic Value

Several studies have described the prognostic usefulness of echocardiography and technetium scintigraphy findings in patients with CA. Using tissue Doppler echocardiography, Koyama et al.²⁶ showed that the reduction in LV basal strain was an independent predictor of both cardiac and overall deaths in patients with AL-CA. In other studies, 2- or 3-dimensional speckle tracking analysis revealed that LV global LS was associated with outcome in AL-CA.^27,28 Regarding ATTR-CA, technetium pyrophosphate myocardial uptake could stratify patients into those with a high and low risk of death.²⁹

In the present meta-analysis, 2-dimensional speckle tracking-derived RAPS was significantly associated with outcome when we assessed RAPS as a dichotomous, but not as a continuous, variable. These contradictory results may be explained, in part, by the fact that many studies regarding RAPS as a dichotomous variable used an “optimal” cut-off value based on ROC analysis, which provided a best-case scenario in each study, resulting in the overestimation of true effect size. A significant association between the presence of RAPS and outcome was lost when we conducted crude publication bias adjusted analysis, trim-and-fill analysis. These results suggest that RAPS is not a robust marker for predicting outcomes in CA. It is interesting to note that, in subgroup analysis, RAPS was associated with outcomes only in AL-CA, but not ATTR-CA. Currently, whether this result could be explained by pathogenic and prognostic differences between the 2 subtypes is unknown.

Clinical Implications

The RAPS “cherry on top” of the bull’s eye plot is visually attractive, and is the most famous echocardiographic hallmark for diagnosing CA. This meta-analysis showed that its DTA was modest at best, and that its accuracy varied among ultrasound speckle tracking software packages. When we used the best software under a CA incidence of 40%, the presence of RAPS increased the post-test probability of CA from 40% to 90%, but the absence of RAPS resulted in a decrease in the post-test probability of CA of from 40% to 24%. RAPS was not robustly associated with outcome.

Clinically, RAPS should be used with caution due to the following factors. First, high heterogeneity (I²=84%, τ²=1.08) of DOR was noted in our results; this may be attributed to interobserver variability in speckle tracking analysis, the heterogeneity of the control groups in previous studies, and different software packages being used, suggesting the diagnostic power of RAPS may vary in different clinical scenarios. Second, speckle tracking analysis is not available in every echocardiography laboratory, and its reliability is further reduced when image quality is poor. Therefore, rather than relying on RAPS as the sole indicator of CA, we suggest the use of a prediction model/score based on simple and widely available clinical and echocardiographic parameters, such as LV wall thickness (septal thickness ≥16 mm), which is more clinically relevant for assessing CAs, and performing additional diagnostic tests, but only in patients who are at high risk of CA.¹²

Study Limitations

This study has several limitations. First, the observed results were influenced by the variability of the original studies. Second, the cut-off value of RAPS for the DTA arm was not uniform and the authors used the best cut-off value according to ROC analysis in 16 studies. The application of a fixed cut-off value (e.g., a RAPS ratio of 1) may further reduce the DTA. However, this approach may not be applicable because we found vendor variability in regional strain measurements. Third, we observed heterogeneity in the study cohorts and comparator groups. We also could not adjust for the disease stage of CA, because very few studies addressed clinical stage (k=3). A meta-analysis at the level of individual patient data is required to overcome this problem. Fourth, the result that DTA differed among different speckle tracking software packages should be interpreted with caution even though the results were significant, as shown by meta-regression analysis, because only 4 studies used TomTec software and 5 studies used Philips software for analysis. A randomized controlled study or comparison of the DTA in the same study patients may overcome this problem, but such studies have seldom been conducted.⁷ Finally, the number of studies evaluating prognosis was relatively small because some authors used RAPS as a dichotomous outcome, whereas others used it as a continuous variable. Thus, we could not perform a detailed subgroup or meta-regression analysis to determine whether the results would remain the same according to different ultrasound software packages in specific types of CA. A future meta-analysis should be conducted to validate whether RAPS is associated with outcomes in more accumulated publications.

Conclusions

The diagnostic accuracy of RAPS for CA was modest, and results differed among strain analytical software packages used for the analysis. RAPS was not robustly associated with outcomes in the types of CA and criteria used for analysis. Thus, RAPS is not a sensitive marker for CA.

Sources of Funding

This research was supported by National Science and Technology Council (NSTC 113-2314-B-002-138), Taipei, Taiwan and National Taiwan University Hospital (113-IF0004), Taipei, Taiwan.

Disclosures

The authors have no conflicts of interest to declare.

Supplementary Files

Please find supplementary file(s);

https://doi.org/10.1253/circj.CJ-24-0472

References

• 1.   Phelan D, Collier P, Thavendiranathan P, Popović ZB, Hanna M, Plana JC, et al. Relative apical sparing of longitudinal strain using two-dimensional speckle-tracking echocardiography is both sensitive and specific for the diagnosis of cardiac amyloidosis. Heart 2012; 98: 1442, doi:10.1136/heartjnl-2012-302353.
• 2.   Ternacle J, Bodez D, Guellich A, Audureau E, Rappeneau S, Lim P, et al. Causes and consequences of longitudinal LV dysfunction assessed by 2D strain echocardiography in cardiac amyloidosis. JACC Cardiovasc Imaging 2016; 9: 126–138, doi:10.1016/j.jcmg.2015.05.014.
• 3.   Williams LK, Forero JF, Popovic ZB, Phelan D, Delgado D, Rakowski H, et al. Patterns of CMR measured longitudinal strain and its association with late gadolinium enhancement in patients with cardiac amyloidosis and its mimics. J Cardiovasc Magn Reson 2017; 19: 61, doi:10.1186/s12968-017-0376-0.
• 4.   Cuddy SAM, Datar Y, Ovsak G, Saith S, Murphy SP, Bay CP, et al. Optimal echocardiographic parameters to improve the diagnostic yield of Tc-99 m-bone avid tracer cardiac scintigraphy for transthyretin cardiac amyloidosis. Circ Cardiovasc Imaging 2022; 15: e014645, doi:10.1161/circimaging.122.014645.
• 5.   Mirea O, Pagourelias ED, Duchenne J, Bogaert J, Thomas JD, Badano LP, et al. Intervendor differences in the accuracy of detecting regional functional abnormalities. JACC Cardiovasc Imaging 2018; 11: 25–34, doi:10.1016/j.jcmg.2017.02.014.
• 6.   Mirea O, Pagourelias ED, Duchenne J, Bogaert J, Thomas JD, Badano LP, et al. Variability and reproducibility of segmental and longitudinal strain measurement. JACC Cardiovasc Imaging 2018; 11: 15–24, doi:10.1016/j.jcmg.2017.01.027.
• 7.   Jiang MN, Simkowski J, Hangouche NE, Lee J, Marion M, Vaitenas I, et al. Relative apical sparing of longitudinal strain in cardiac amyloidosis: An intervendor software variability assessment. J Am Soc Echocardiogr 2023; 36: 254–256, doi:10.1016/j.echo.2022.09.011.
• 8.   Abecasis J. Prevalence and significance of relative apical sparing in aortic stenosis: Insights from an echo and cardiovascular magnetic resonance study of patients referred for surgical aortic valve replacement. Eur Heart J Cardiovasc Imaging 2023; 24: 1033–1042, doi:10.1093/ehjci/jead032.
• 9.   Garcia-Pavia P, Rapezzi C, Adler Y, Arad M, Basso C, Brucato A, et al. Diagnosis and treatment of cardiac amyloidosis: A position statement of the ESC Working Group on Myocardial and Pericardial Diseases. Eur Hear J 2021; 42: 1554–1568, doi:10.1093/eurheartj/ehab072.
• 10.   Boldrini M, Cappelli F, Chacko L, Restrepo-Cordoba MA, Lopez-Sainz A, Giannoni A, et al. Multiparametric echocardiography scores for the diagnosis of cardiac amyloidosis. JACC Cardiovasc Imaging 2020; 13: 909–920, doi:10.1016/j.jcmg.2019.10.011.
• 11.   Davies DR, Redfield MM, Scott CG, Minamisawa M, Grogan M, Dispenzieri A, et al. A simple score to identify increased risk of transthyretin amyloid cardiomyopathy in heart failure with preserved ejection fraction. JAMA Cardiol 2022; 7: 1036–1044, doi:10.1001/jamacardio.2022.1781.
• 12.   Arana-Achaga X, Goena-Vives C, Villanueva-Benito I, Solla-Ruiz I, Jimenez AR, Gaspar TI, et al. Development and validation of a prediction model and score for transthyretin cardiac amyloidosis diagnosis T-Amylo. JACC Cardiovasc Imaging 2023; 16: 1567–1580, doi:10.1016/j.jcmg.2023.05.002.
• 13.   Harrer M, Cuijpers P, Furukawa TA, Ebert DD. Doing meta-analysis with R: A hands-on guide. 1st edn. Boca Raton: CRC Press, 2022.
• 14.   Mavridis D, Salanti G. How to assess publication bias: Funnel plot, trim-and-fill method and selection models. Evid Based Ment Health 2014; 17: 30, doi:101136/eb-2013-101699.
• 15.   Shim SR, Kim SJ, Lee J. Diagnostic test accuracy: Application and practice using R software. Epidemiol Health 2019; 41: e2019007, doi:10.4178/epih.e2019007.
• 16.   Fagan TJ. Letter: Nomogram for Bayes’s theorem. N Engl J Med 1975; 293: 257, doi:10.1056/nejm197507312930513.
• 17.   Kotanidis CP, Bazmpani MA, Haidich AB, Karvounis C, Antoniades C, Karamitsos TD. Diagnostic accuracy of cardiovascular magnetic resonance in acute myocarditis: A systematic review and meta-analysis. JACC Cardiovasc Imaging 2018; 11: 1583–1590, doi:10.1016/j.jcmg.2017.12.008.
• 18.   Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The development of QUADAS: A tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol 2003; 3: 25, doi:101186/1471-2288-3-25.
• 19.   Hayden JA, Côté P, Bombardier C. Evaluation of the Quality of Prognosis Studies in Systematic Reviews. Ann Intern Med 2006; 144: 427, doi:10.7326/0003-4819-144-6-200603210-00010.
• 20.   Sterne JA, Sutton AJ, Ioannidis JP, Terrin N, Jones DR, Lau J, et al. Recommendations for examining and interpreting funnel plot asymmetry in meta-analyses of randomised controlled trials. BMJ 2011; 343: d4002, doi:10.1136/bmj.d4002.
• 21.   Treibel TA, Fontana M, Gilbertson JA, Castelletti S, White SK, Scully PR, et al. Occult transthyretin cardiac amyloid in severe calcific aortic stenosis: Prevalence and prognosis in patients undergoing surgical aortic valve replacement. Circ Cardiovasc Imaging 2016; 9: 245, doi:10.1161/circimaging.116.005066.
• 22.   Castano A, Narotsky DL, Hamid N, Khalique OK, Morgenstern R, DeLuca A, et al. Unveiling transthyretin cardiac amyloidosis and its predictors among elderly patients with severe aortic stenosis undergoing transcatheter aortic valve replacement. Eur Heart J 2017; 38: 2879–2887, doi:10.1093/eurheartj/ehx350.
• 23.   Scully PR, Treibel TA, Fontana M, Lloyd G, Mullen M, Pugliese F, et al. Prevalence of cardiac amyloidosis in patients referred for transcatheter aortic valve replacement. J Am Coll Cardiol 2018; 71: 463–464, doi:10.1016/j.jacc.2017.11.037.
• 24.   Jaiswal V, Ang SP, Chia JE, Abdelazem EM, Jaiswal A, Biswas M, et al. Echocardiographic predictors of presence of cardiac amyloidosis in aortic stenosis. Eur Heart J Cardiovasc Imaging 2022; 23: 1290–1301, doi:10.1093/ehjci/jeac146.
• 25.   Ichikawa Y, Oota E, Odajima S, Kintsu M, Todo S, Takeuchi K, et al. Impact of Tafamidis on echocardiographic cardiac function of patients with transthyretin cardiac amyloidosis. Circ J 2023; 87: 508–516, doi:10.1253/circj.CJ-22-0683.
• 26.   Koyama J, Falk RH. Prognostic significance of strain Doppler imaging in light-chain amyloidosis. JACC Cardiovasc Imaging 2010; 3: 333–342, doi:10.1016/j.jcmg.2009.11.013.
• 27.   Buss SJ, Emami M, Mereles D, Korosoglou G, Kristen AV, Voss A, et al. Longitudinal left ventricular function for prediction of survival in systemic light-chain amyloidosis: Incremental value compared with clinical and biochemical markers. J Am Coll Cardiol 2012; 60: 1067–1076, doi:10.1016/j.jacc.2012.04.043.
• 28.   Lei C, Zhu X, Hsi DH, Wang J, Zuo L, Ta S, et al. Predictors of cardiac involvement and survival in patients with primary systemic light-chain amyloidosis: Roles of the clinical, chemical, and 3-D speckle tracking echocardiography parameters. BMC Cardiovasc Disord 2021; 21: 43, doi:10.1186/s12872-021-01856-3.
• 29.   Castano A, Haq M, Narotsky DL, Goldsmith J, Weinberg RL, Morgenstern R, et al. Multicenter study of planar technetium 99 m pyrophosphate cardiac imaging: Predicting survival for patients with ATTR cardiac amyloidosis. JAMA Cardiol 2016; 1: 880–889, doi:10.1001/jamacardio.2016.2839.