Introduction

Amyloidosis is a disease that results from the deposition of amyloid fibrils in the extracellular space, destroying the structure of tissues and consequently causing damage to multiple organ systems. The common clinical types are immunoglobulin light-chain (AL) amyloidosis and transthyretin (ATTR) amyloidosis, accounting for 59.1% and 32.9% of patients with amyloidosis. AL amyloidosis is due to excessive secretion of monoclonal light chains by bone marrow plasma cells and aggregation into amyloid fibrils deposited in tissues, often involving the liver, lung, kidney, lower urinary tract, bone marrow, and rarely the heart [1]. ATTR amyloidosis is caused by the misfolding of transthyretin proteins, which aggregate into amyloid fibrils and accumulate in tissues, mainly involving the heart, carpal tunnel, spinal ligaments, and peripheral nerves [2,3,4]. When ATTR or AL amyloid is deposited in the myocardium, it causes thickening of the ventricular wall, diastolic dysfunction, restricted systolic function, impaired transmission, and ultimately heart failure. Research studies have shown that approximately 20% of heart failure and myocardial wall thickening patients have wild-type transthyretin cardiac amyloidosis [5].

In the past 20 years, cardiac amyloidosis has often been considered a rare disease and has not received much attention. Patients with early-onset ATTR-CA tend to often present with decreased exercise tolerance and slowly progressive decreased diastolic heart function, and these nonspecific symptoms can mostly lead to delayed diagnosis or miss the best time for treatment. The earliest examinations to suspected heart-related discomforts are electrocardiogram (ECG) and echocardiography. ECG is performed as a routine exam. Approximately 34–46% of patients with amyloid cardiomyopathy exhibit low voltage in the limb leads, with mutant ATTR amyloid cardiomyopathy (ATTRm-CA) and wild-type ATTR amyloid cardiomyopathy (ATTRwt-CA) accounting for 38% and 18%, respectively [6, 7]. The characteristic manifestations of ATTR-CA diagnosed by echocardiography are mainly preserved ejection fraction and left ventricular hypertrophy, which is nonspecific and can also occur in hypertensive hypertrophic heart disease and hereditary hypertrophic cardiomyopathy [8]. A further examination, cardiac MRI, has good diagnostic sensitivity (95%) and specificity (98%) for amyloid cardiomyopathy but does not provide a specific differential diagnosis between ATTR-CA and other cardiomyopathies [9]. In addition, the gold standard for the diagnosis of ATTR-CA is an endomyocardial biopsy, which is invasive, risky, and needs to be performed by a physician with excellent expertise.

Current treatment modalities for ATTR-CA include chemotherapy, transthyretin protein reduction, and liver transplantation, and the prognosis after treatment is much better than that of AL amyloid cardiomyopathy (AL-CA) [10]. Therefore, early diagnosis and treatment are crucial. 99mTc-labelled bone tracer has been used for a long time for the diagnosis of ATTR-CA. The advantage of bone imaging agents is the specificity in differentiating ATTR-CA from AL-CA and other types of cardiomyopathies. Published studies in various countries have shown good diagnostic efficacy of technetium-99m diphosphono-1,2-propanodicarboxylic acid (99mTc-DPD), technetium-99m pyrophosphate (99mTc-PYP), and technetium-99m hydroxymethylene diphosphonate (99mTc-HMDP) imaging for ATTR-CA, but most studies included too few patients. In addition, the clinical application of bone scans for the diagnosis of ATTR-CA is still in the early clinical trial stage in many countries and regions. Therefore, we included multiple studies in our meta-analysis to assess diagnostic performance from overall diagnostic efficacy to specific differential applications (different imaging time, different diagnostic criteria, different bone-seeking tracers), thus providing more comprehensive and more adequate diagnostic evidence for the use of bone scintigraphy in cardiac disease.

Materials and methods

Search strategy

This review followed the Preferred Reporting Items for a Systematic Review and Meta-Analysis (PRISMA) statement [11]. Ethical approval or informed consent was not required by conducting a meta-analysis of published studies without reference to specific patients for inclusion.

We searched the published literature by utilizing the PubMed database and EMBASE database with a search deadline of August 10, 2021. The search formula used was: (((amyloid[Title/Abstract] OR amyloidosis[Title/Abstract]) AND (TTR[Title/Abstract] OR ATTR[Title/Abstract] OR transthyretin[Title/Abstract])) AND (scintigraphy[Title/Abstract] OR scan[Title/Abstract] OR SPECT[Title/Abstract] OR SPET[Title/Abstract] OR bone[Title/Abstract] OR skeletal[Title/Abstract] OR skeleton[Title/Abstract] OR PYP[Title/Abstract] OR DPD[Title/Abstract] OR HMDP[Title/Abstract] OR MDP[Title/Abstract] OR HDP[Title/Abstract])). There was no restriction on the language of the article.

Study selection and exclusion

We included studies that met the following criteria: (1) studies in which 99mTc-PYP or 99mTc-DPD or 99mTc-HMDP imaging were used for the diagnosis of ATTR-CA; (2) diagnostically relevant data could be extracted, such as true-positive (TP), false-positive (FP), false-negative (FN), true-negative (TN), sensitivity, and specificity; (3) studies with no less than ten cases. The following types of literature were excluded: (1) non-human studies; (2) case report, review, editorial, letter, comment, conference proceedings, conference abstract, and articles without full-text.

Articles that did not meet the criteria were first excluded through the database filters, and then the remaining articles that did not meet the criteria were further excluded by carefully reading the title, abstract, and full text. Two authors decided the final included articles, and any disagreements were resolved through consensus discussions.

Data extraction and quality assessment

We extracted relevant information from the included literature, including first author, year of publication, country, study design, number of patients, age of patients, type of patients, diagnostic modality, type of imaging agent, and reference standard. TP, FP, FN, and TN were extracted directly or calculated indirectly by reading the full text. Two authors assessed the methodology of each study using the entries in the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) questionnaire [12].

Statistical analysis

First, we assessed the quality of the included studies using RevMan software (Review Manager, version 5.4). The included studies’ visual assessment method, quantitative ratio (i.e., H/CL, H/WB, H/M) analysis, and myocardial SUV analysis may have used different thresholds in the diagnostic process. We used a Bayesian bivariate random-effects model analysis to fully consider the variation among studies and the correlation between pooled sensitivity and specificity. We obtained the pooled sensitivity, specificity, LLR  +, LLR −, and LDOR using the “INLA” package [13] (the integrated nested Laplace approximation based on INLA to combine the data) and the “meta4diag” package [14] (the bivariate meta-analysis based on the Bayesian framework principle) in R (R for Windows, version 4.1.0). The fitted SROC curves obtained from the modelling based on the Logit transformation were used to assess the overall efficacy of bone scan for ATTR-CA diagnosis. In addition, for subgroup data with more zero values and insufficient data that could not be pooled by R software, the pooled sensitivity and specificity were calculated after correction by adding 0.5 to the cells with zero values using MetaDiSc software (Meta-analysis of Diagnostic and Screening Tests, version 1.4) [15]. Finally, publication bias was assessed by plotting funnel plots using the R package described above [16].

Results

Literatures search and study characteristics

We retrieved 383 articles on PubMed, 1021 articles on EMBASE, and two articles by manual search. First, a total of 1177 articles were excluded by the automatic filters of the database; then, 167 articles were excluded by reading the titles and abstracts of the remaining 229 articles; further, 23 articles were excluded by carefully reading the full text of the remaining 63 articles. Finally, 39 articles were included in our meta-analysis [5, 17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54]. The detailed search and selection process are shown in Fig. 1.

Fig. 1
figure 1

Flow chart of literature search using PRISMA 2020 method

Full size image

We included 39 studies published between 2002 and 2021, 28 retrospective and 11 prospective, with a total of 3636 patients and a mean or median age span of 54.4–86 years. There were 12 studies using 99mTc-DPD, eight using 99mTc-HDMP, 16 using 99mTc-PYP, and three using multiple bone imaging agents. Four studies were analyzed quantitatively by measuring cardiac SUVmax or SUVpeak, and the remaining 35 studies were analyzed by semi-quantitative visual score or combined H/CL (heart to contralateral chest retention) or H/WB (heart to whole-body retention) or H/M (heart to mediastinum retention) ratios. The reference standard for 28 of these studies was subendocardial biopsy or combined with extracardiac tissue biopsy; for 8 studies, the reference standard was extracardiac tissue biopsy, gene sequencing, and immunohistochemistry; and for the other 3 studies, the reference standard was typical imaging image presentation, clinical features, and immunohistochemistry. The essential characteristics of the included studies are detailed in Table 1.

Table 1 Characteristics of included studies
Full size table

Methodological qualitative analysis

Figure 2a shows the risk bias and clinical usability issues for each included study regarding patient selection, experimental methodology, reference standard, and study flow. Figure 2b presents a summary assessment of all included studies in terms of methodology. Fourteen studies reported that they included patients on a consecutive basis [19, 23, 34, 35, 37, 39, 42,43,44,45, 51,52,53,54], and 25 studies did not report whether they were consecutive [5, 17, 18, 20,21,22, 24,25,26,27,28,29,30,31,32,33, 36, 38, 40, 41, 46,47,48,49,50]. Regarding the index test and reference standard, 16 studies were blinded [18, 19, 21,22,23, 28, 29, 36, 37, 40, 42, 45, 50, 51, 54], 21 did not report whether they were blinded, and two studies were unblinded [31, 47], while ten studies had a reference standard other than pathology [5, 18, 19, 34, 35, 38,39,40,41, 47], thus introducing a high risk in the index test and the reference standard. Further, Asif and Bellevre’s study used the visual score of bone scintigraphy as a reference standard, leading to the concern of clinical applicability [34, 47]. Since Löfbacka’s study included patients with known ATTRm-CA and positive bone scan, there is a clinical applicability concern in patient selection [49].

Fig. 2
figure 2

Quality evaluation of the methodology of each included study and overall assessment of risk bias and applicability concerns for included studies based on the QUADAS-2 method

Full size image

Overall pooled diagnostic performance of bone scintigraphy

Figure 3 shows a forest plot of the pooled sensitivity and specificity of bone scan for the diagnosis of ATTR-CA. The high pooled sensitivity (0.97, 95% CI 0.95–0.99) and specificity (0.96, 95% CI 0.94–0.98) of the 39 included studies indicate the excellent performance of bone scan for the diagnosis of ATTR. Figure 4 shows forest plots of positive likelihood ratio, negative likelihood ratio, and diagnostic odds ratio after logit transformation. The pooled estimates of LLR  +, LLR −, and LDOR were 3.22 (95% CI 2.76–3.80), − 3.59 (95% CI − 4.58 to − 2.86), and 6.81 (95% CI 5.87–7.93), respectively. Figure 5 shows the SROC curve and the estimate of AUC (0.99, 95% CI 0.95–0.99). The dark blue SROC line shown in Fig. 5 is particularly close to the upper left corner, suggesting that the diagnostic value of bone scintigraphy is exceptionally high.

Fig. 3
figure 3

Forest plot for the pooled sensitivity (true positive rate) and specificity (true negative rate)

Full size image
Fig. 4
figure 4

Forest plot for the pooled LLR −, LLR +, and LDOR

Full size image
Fig. 5
figure 5

SROC curve for the evaluation of the overall performance of bone scan for the diagnosis of ATTR-CA

Full size image

Pooled diagnostic performance of each subgroup set

Table 2 shows the pooled sensitivity, specificity, LLR  +, LLR −, LDOR, and 95% CI for each subgroup.

Table 2 Comparison of diagnostic performance between subgroups
Full size table

Visual scoring, quantitative ratio (i.e., H/CL, H/WB, H/M) analysis, and myocardial SUVmax/peak analysis showed high pooled sensitivities of 0.97 (95% CI 0.94–0.98), 0.98 (95% CI 0.96–1.00), and 1.00 (95% CI 0.95–1.00), respectively, for diagnosing ATTR-CA. The pooled specificities were higher for visual scoring (0.96, 95% CI 0.94–0.98) and quantitative ratio analysis (0.96, 95% CI 0.92–0.99) and lowered for myocardial SUVmax/peak (0.87, 95% CI 0.79–0.93). The pooled LLR  +, LLR − and LDOR of quantitative ratio analysis were higher than visual scoring [3.33 (95% CI 2.45–4.59) vs. 3.21 (95% CI 2.71–3.83); − 4.06 (95% CI − 5.45 to − 3.05) vs. − 3.41 (95% CI − 4.15 to − 2.80); 7.39 (95% CI 5.97–9.23) vs. 6.62 (95% CI 5.84–7.48)], indicating that quantitative ratio analysis may be superior to visual scoring in identifying ATTR-CA.

On the other hand, when the diagnostic threshold for visual scoring was score 1 (five data sets), the pooled sensitivity, specificity, LLR  +, LLR −, and LDOR were 0.99 (95% CI 0.97–1.00), 0.93 (95% CI 0.78–0.99), 3.02 (95% CI 1.35–5.27), − 4.62 (95% CI − 6.33 to − 3.44), and 7.64 (95% CI 5.58–10.35), respectively. When the diagnostic threshold was score 2 (32 data sets), the pooled sensitivity, specificity, LLR  +, LLR −, and LDOR 0.96 (95% CI 0.93–0.98), 0.96 (95% CI 0.94–0.98), 3.16 (95% CI 2.66–3.79), − 3.12 (95% CI − 3.91 to − 2.51), and 6.28 (95% CI 5.43–7.23), respectively.

Regarding imaging time, the pooled sensitivity, specificity, LLR  +, LLR −, and LDOR for studies in which the time from radiotracer injection to scanning was 30 min–1 h (14 data sets) were 0.97 (95% CI 0.95–0.99), 0.95 (95% CI 0.93–0.97), 3.03 (95% CI 2.68–3.49), − 3.72 (95% CI − 4.90 to − 2.08), and 6.75 (95% CI 5.68–8.04), respectively. The pooled sensitivity, specificity, LLR  +, LLR −, and LDOR for studies in which the imaging time was 2.5–4 h after radiotracer injection (33 data sets) were 0.98 (95% CI 0.95–0.99), 0.97 (95% CI 0.94–0.99), 3.49 (95% CI 2.82–4.35), − 3.91 (95% CI − 5.22 to − 2.97), and 7.40 (95% CI 6.27–8.79), respectively. Overall, pooled sensitivity, specificity, LLR  +, LLR − and LDOR were higher at an imaging time of approximately 3 h than at approximately 1 h.

When grouped according to the type of imaging agent, the pooled sensitivity, specificity, LLR  +, LLR −, and LDOR for imaging with 99mTc-DPD (12 data sets) were 0.98 (95% CI 0.96–1.00), 0.94 (95% CI 0.88–0.98), 2.82 (95% CI 2.06–3.79), − 4.52 (95% CI − 7.55 to − 3.03), and 7.34 (95% CI 5.47–10.47), respectively. The pooled sensitivity, specificity, LLR  +, LLR −, and LDOR for imaging with 99mTc-PYP (16 data sets) were 0.95 (95% CI 0.90–0.98), 0.95 (95% CI 0.93–0.97), 3.02 (95% CI 2.53–3.71), − 2.96 (95% CI − 4.07 to − 2.16), and 5.98 (95% CI 4.94–7.35), respectively. The pooled sensitivity, specificity, LLR  +, LLR −, and LDOR for imaging with 99mTc-HMDP (9 data sets) were 1.00 (95% CI 0.98–1.00), 0.98 (95% CI 0.96–1.00), 4.24 (95% CI 3.07–6.27), − 6.38 (95% CI − 11.72 to − 3.80), and 10.62 (95% CI 7.63–15.55), respectively.

Publication bias

By visual analysis of the funnel plot (Fig. 6), we found eight studies scattered outside the 95% confidence interval of the funnel plot, indicating heterogeneity among our included studies. In addition, a large number of studies were distributed at the bottom of the funnel plot, indicating that we included a large number of small sample studies. Observing the studies within the 95% confidence interval as a whole, we found that these studies were roughly symmetrically distributed on both sides of the LDOR estimates, thus indicating that there was no significant publication bias in the inclusion of our meta-analysis.

Fig. 6
figure 6

Funnel plot on bone scintigraphy for the diagnosis of ATTR-CA

Full size image

Discussion

Bone-seeking radiopharmaceuticals have been used for the diagnosis of ATTR-CA for 20 years. Most clinical studies based on the diagnosis of ATTR-CA have a small number of cases due to the limited therapeutic effect of amyloid cardiomyopathy and the lack of awareness of the disease among clinicians. The previous meta-analysis has shown good diagnostic performance with bone scans [55]. Our study is a Bayesian bivariate meta-analysis based on a previous study by adding many new publications. In addition, a meticulous subgroup meta-analysis was performed to assess the difference in diagnostic efficacy of bone scintigraphy for ATTR-CA.

Electrocardiography and echocardiography are the most common examinations for patients with suspect amyloid cardiomyopathy or other cardiac diseases. Low voltage in the limb leads of the ECG combined with septal thickening (>  12 mm) on echocardiography can distinguish ATTR-CA from other diseases associated with septal thickening with a sensitivity and specificity of 100% and 95%, respectively [56, 57]. However, only 25% of patients with ATTR-CA showed low voltage in the limb leads, and the lack of specificity in the presentation of septal thickening made it difficult to differentiate between ATTR-CA and AL-CA in clinical practice [58]. A meta-analysis study showed that late gadolinium enhancement cardiovascular MRI has a high diagnostic value in diagnosing amyloid cardiomyopathy, with a sensitivity and specificity of 85% (95% CI 77–91%) and 92% (95% CI 83–97%), respectively, but fails to diagnose ATTR-CA subtype and AL-CA subtype specifically [59]. Our meta-analysis showed that bone scan had excellent diagnostic efficacy for ATTR-CA, with sensitivity, specificity, and AUC of 0.97, 0.96 and, 0.99, respectively.

First, it is noteworthy that the results of our analysis suggest better diagnostic performance with 3-h imaging compared to 1 h, which is consistent with the findings of previous studies. Singh et al. found that the concentration of bone-seeking tracer in the cardiac cavity blood pool was still higher at 1 hour after the radiopharmaceutical injection and therefore delayed imaging to 3 h [42]. Masri’s study also demonstrated that increased blood pool activity affects visual diagnostic results. In his study, images with an imaging time of 1 h and 3 h were scored differently in nine patients according to visual scoring criteria (i.e., nine false positives appeared in 1-h imaging). Therefore, 1-h imaging is recommended to be combined with SPECT/CT imaging to accurately differentiate between myocardial tracer uptake and ventricular blood pool tracer aggregation to reduce false positives [38].

Second, the predominant method for diagnosing ATTR-CA is the Perugini grading system, a semi-quantitative visual analysis method [18]. Our results suggest that the diagnostic efficacy of the visual scoring method is comparable to that of the quantitative ratio (H/CL or H/WB, or H/M) analysis methods. It has been shown that there is no difference in diagnostic accuracy between the visual scoring method and the quantitative ratio method, and both methods can be well mastered and applied by experienced readers and novices [42]. In a retrospective study, the Perugini visual score (≥  2) of planar imaging had a better diagnostic performance for ATTR-CA and was almost comparable to 99mTc-PYP SPECT imaging. In contrast, the planar image’s H/CL ratio (≥  1.5) performed poorly as a diagnostic criterion, with a sensitivity and specificity of 0.57 and 0.95, respectively [47]. It is noteworthy that the diagnostic criterion based on the visual score of planar images alone is flawed. In the study of Poterucha, 32% of patients with positive visual assessment results were due to excessive tracer accumulation in the cardiac blood pool [52]. This increased uptake of cardiac blood pool leading to decreased visual diagnostic accuracy is widespread. Therefore, in most studies, patients with a visual score of 1 or 2 on planar imaging underwent further SPECT imaging [19, 21, 22, 29, 38, 42, 51]. In addition, it has been shown that the visual judgment method of SPECT images alone has higher diagnostic sensitivity (1.00 vs. 0.93) and specificity (0.99 vs. 0.91) than the visual grading score of planar images [50]. Therefore, routinely adding SPECT or SPECT/CT imaging to improve the diagnostic accuracy of ATTR-CA should become a routine procedure [47, 60]. Another noteworthy aspect is that some researchers have concluded that the diagnostic performance of absolute quantitative measurements of myocardial radiotracer uptake values and visual grading score are in good agreement. In all of these studies, the sensitivity of myocardial SUVmax/peak for diagnosing ATTR-CA reached 100%. Moreover, the quantitative myocardial uptake values from SPECT/CT also more accurately reflect the load of myocardial amyloid deposition [31, 34, 35, 41, 61]. This finding may also herald the potential value of SPECT/CT quantitative myocardial radiotracer uptake values in diagnosis, efficacy assessment of treatment, and prognosis prediction that can be explored.

Furthermore, the results of our meta-analysis showed that the diagnostic threshold of the visual score of 1 was more sensitive but less specific than that of 2. The studies of Gillmore and Cappelli compared the diagnostic performance of a visual score of 1 and a visual score of 2 and found that score 1 was more sensitive and less specific than score 2 [26, 27]. AL-CA was a frequent cause of false positives and was predominantly distributed in the group with a visual score of 1. Quarta et al. mentioned that 39% of AL-CA patients had varying degrees of 99mTc-DPD uptake. Moreover, patients with AL-CA presenting with 99mTc-DPD uptake tend to have a poorer cardiac function and a worse prognosis, so care should be taken to distinguish ATTR-CA from AL-CA at the time of diagnosis carefully [62]. In addition to AL-CA, the presence of the following disorders can also lead to the false-positive diagnosis, including extensive myocardial infarction, unstable angina, cardiotoxicity due to adriamycin, pericarditis, alcoholic cardiomyopathy, pericardial tumors, and hypercalcemia [63].

It has been shown that 99mTc-MDP, the radiotracer most commonly used for bone scintigraphy, is less concentrated in the myocardium of ATTR-CA patients and is less suitable as a specific cardiac imaging agent for the diagnosis of ATTR-CA [18, 20]. The imaging agents routinely used for the diagnosis of ATTR-CA are 99mTc-DPD, 99mTc-PYP, and 99mTc-HMDP. All three tracers are effective for the diagnosis of ATTR-CA. Studies have demonstrated differences in pharmacokinetics, plasma protein binding, renal excretion, and degree of bone binding between 99mTc-DPD and 99mTc-HMDP, but the differences in uptake and distribution in patients with ATTR-CA have not been conclusively established [64,65,66]. The results of our meta-analysis showed differences in pooled sensitivity and specificity between the different imaging agents, which may also suggest slight differences in the affinity of bone-seeking tracers for ATTR amyloid-deposited myocardium. In a comparative study of dual nuclide imaging in six patients, the author described a mild difference in the degree of concentration of 99mTc-HMDP and 99mTc-DPD in the myocardium of ATTR-CA patients, but the difference was not statistically significant [67]. Unfortunately, few studies have directly compared the differences of these three imaging agents in diagnosing ATTR-CA. In clinical practice, most hospitals choose bone-seeking radiopharmaceuticals for cardiac use not based on better diagnostic performance, no doubt, but rather on which imaging agent is more readily available.

ATTR-CA is divided into ATTRwt-CA (senile systemic amyloidosis) and ATTRm-CA (familial amyloid cardiomyopathy). Studies on the diagnosis of ATTRwt-CA have shown that bone scintigraphy has superior diagnostic sensitivity and specificity [5, 51]. Similarly, bone scintigraphy has excellent diagnostic accuracy in patients with amyloid cardiomyopathy with common mutant TTR genotypes. However, bone scintigraphy is not ideal for the diagnosis of amyloid cardiomyopathy with rare TTR genotypes. In Musumeci’s study, the sensitivity (10.5%) and diagnostic accuracy (37%) of bone scintigraphy were very low for ATTR-CA with Phe64Leu TTR gene mutation [39]. Other researchers have similarly found low myocardial uptake of the bone radiotracer in ATTR-CA patients with Phe64Leu mutation, leading to false-negative results [52]. Thus, the original mechanisms of bone tracer binding to ATTR amyloid-deposited myocardium, including high calcium loading in ATTR amyloid tissue leading to the high uptake of bone radiotracer and high affinity of ATTR amyloid fibrous tissue for bone radiotracer leading to the high uptake of the bone radiotracer, remain to be further investigated.

Another aspect that deserves our attention is the current development of SPECT instruments and application software, especially the highly sensitive 360-degree rotating cadmium telluride (CZT) detector, which improves the diagnostic accuracy of nuclear medicine. Compared to the conventional sodium iodide (NaI) detector, the CZT detector offers higher detection efficiency and better detection sensitivity, i.e., reducing the dose of tracer used and reducing the imaging duration while ensuring better image clarity. The efficacy of the SPECT gamma camera equipped with the CTZ detector for diagnosing ATTR-CA performed better in all the studies we included, with a diagnostic sensitivity of almost 100% [34, 36, 44, 54].

Finally, we need to mention some limitations in this meta-analysis. First, our inclusion criteria were not very strict, and the most important of which needs to be discussed is that not all the reference standards of our included literature were endocardial biopsies. Some studies used tissue biopsies of carpal tunnel ligaments and spinal ligaments as the reference standard for diagnosis. Eldhagen et al. found that the presence of ATTR amyloid deposition in the ligamentum flavum of the patient with lumbar spinal stenosis was not associated with ATTR amyloid deposits in the myocardium [68]. By analyzing surgical resection specimens, Sueyoshi found that patients with bilateral carpal tunnel syndrome had a 33.3% incidence of ATTR amyloid deposition in carpal tunnel ligaments or tendons, and patients with lumbar spinal stenosis had a 44.4% incidence of ATTR amyloid deposition in spinal ligaments [69]. It has also been shown that patients with ATTR mutations have an 87.5% positive rate of amyloid detection by Congo-red staining of the carpal tunnel ligament [70]. Therefore, the diagnostic accuracy calculated based on reference standards of biopsies of non-cardiac tissue for diagnosing ATTR-CA by bone scan may deviate to some extent from the actual accuracy. There are also studies in which the reference standard included biopsies of abdominal adipose tissue. Studies that included a large number of amyloidosis cases have found that abdominal fat pad aspiration biopsy has a low diagnostic sensitivity for ATTR amyloidosis (12–27.3%) but a relatively high diagnostic sensitivity for AL amyloidosis (73.2–84%) [71,72,73]. Thus, abdominal fat biopsy as a diagnostic method to exclude AL is safe, applicable, and relatively accurate. Further, we included several studies for specific patient populations, including Musumeci’s study of patients with the Phe64Leu genotype, a rare type of TTR mutation [39], Nitsche’s study of patients with cardiomyopathy with aortic stenosis [51], and Lindmark’s study of patients with ATTRwt-CA only [5]. It could be partly responsible for the heterogeneity between studies. Finally, we included many studies with small samples, which may indicate the low stability of the results of our meta-analysis.

Conclusions

Bone-seeking tracers (99mTc-DPD, 99mTc-PYP, and 99mTc-HMDP) play an essential role in diagnosing ATTR-CA. One hour and 3-h imaging images show differences in the degree of radiotracer concentration in the ventricular blood pool, resulting in slight differences in diagnostic sensitivity and specificity. The visual evaluation of planar cardiac imaging is sufficient to make an accurate diagnosis of ATTR-CA, but the combination of SPECT imaging significantly improves the specificity and sensitivity of the diagnosis. Both quantitative ratios (H/CL, H/WB, H/M) from planar imaging and quantitative cardiac bone tracer uptake values from quantitative SPECT imaging provide accurate diagnostic information. However, bone scans are not very effective in diagnosing ATTR-CA patients with rare mutation types.