Noninvasive Diagnosis of Biopsy-Proven Cardiac Amyloidosis

Transcription

1 Journal of the American College of Cardiology Vol. 43, No. 3, by the American College of Cardiology Foundation ISSN /04/$30.00 Published by Elsevier Inc. doi: /j.jacc Noninvasive Diagnosis of Biopsy-Proven Cardiac Amyloidosis Infiltrative Cardiomyopathy Joseph E. Rahman, MD,* Emelie F. Helou, MD,* Ramona Gelzer-Bell, MD,* Richard E. Thompson, PHD, Chih Kuo, MD, E. Rene Rodriguez, MD, Joshua M. Hare, MD,* Kenneth L. Baughman, MD, FACC,* Edward K. Kasper, MD, FACC* Baltimore, Maryland OBJECTIVES BACKGROUND METHODS RESULTS CONCLUSIONS This study analyzed the utility of electrocardiographic (ECG) and echocardiographic findings in the diagnosis of amyloidosis proven by endomyocardial biopsy. Cardiac amyloidosis is associated with characteristic ECG and echocardiographic changes, yet each finding alone is relatively nonspecific. A combination of noninvasive prognostic parameters would be desirable for this tissue-based diagnosis. We performed an analysis of 196 consecutive patients referred for endomyocardial biopsy because of clinical suspicion of cardiac amyloidosis. The diagnosis was confirmed in 58 patients (29%). The ECGs, echocardiograms, and right heart hemodynamic data were reviewed to determine which findings strongly correlate with the diagnosis. These findings were then used to build multivariate logistic regression models that predict the log-odds of having cardiac amyloidosis. The univariate analysis showed that low-voltage and pseudo-infarction patterns on the ECG and increased myocardial thickness and speckled-appearing myocardium on the echocardiogram were associated with biopsy-proven cardiac amyloidosis (each p 0.01). In multivariate logistic regression models, a combination of a low voltage and measures of myocardial thickness produced the most statistically useful models. For instance, one model showed that if a low voltage was present and interventricular septal thickness is 1.98 cm, the diagnosis of cardiac amyloidosis could be made with a sensitivity of 72% and a specificity of 91%. In this model, the positive predictive and negative predictive values were 79% and 88%, respectively. In patients with suspected cardiac amyloidosis, a combination of noninvasive parameters namely, a low voltage and increased intraventricular septal thickness is a useful diagnostic tool. (J Am Coll Cardiol 2004;43:410 5) 2004 by the American College of Cardiology Foundation Amyloidosis frequently involves the heart. Deposition of amyloid fibrils in myocardial tissue results in reduced ventricular compliance with impairment of relaxation and eventually contraction. Amyloid fibrils may also deposit in myocardial vessels and cause local ischemia. Amyloid deposition is associated with fibrosis of conduction tissue, resulting in conduction abnormalities and arrhythmias. Valvular dysfunction from amyloid deposition may also occur. Cardiac amyloidosis is associated with a variable but generally poor prognosis (1), and early recognition may improve the outcome. The diagnosis of cardiac amyloidosis can be difficult because the clinical presentation is similar to that of other cardiomyopathies. Cardiac amyloidosis is suggested by characteristic echocardiographic findings namely, a granular myocardial appearance, thickened ventricular walls, atrial dilation, and diastolic dysfunction progressing to From the *Division of Cardiology, Department of Medicine, and Department of Pathology, Johns Hopkins University School of Medicine; and Department of Biostatistics, Johns Hopkins School of Public Health and Hygiene, Baltimore, Maryland. This study was funded by the Charity Mae Foundation, Fargo, North Dakota. Manuscript received April 30, 2003; revised manuscript received August 21, 2003, accepted August 25, systolic dysfunction (2). The ECG frequently demonstrates low-voltage and pseudo-infarct patterns, conduction abnormalities, and arrhythmias. However, each finding alone is nonspecific. In this study, we aimed to determine which parameters or combination of parameters correlate with endomyocardial biopsy-proven cardiac amyloidosis and thus can be used to predict disease status. METHODS Study patients. This is a retrospective case-control study of 196 patients who were referred to the Johns Hopkins Medical Institutions between January 1984 and May 2000 for endomyocardial biopsy because of a clinical suspicion of cardiac amyloidosis. Patients were referred for evaluation of congestive heart failure (CHF) in the setting of established systemic amyloidosis, multiple myeloma, connective tissue disease, and chronic inflammatory disorders, or clinical features suggestive of these systemic disorders. There were no other specific criteria for inclusion or exclusion. Approval for this study was obtained from the Johns Hopkins Medicine Institutional Review Board. We performed a detailed examination of the medical records of these patients. In addition, we obtained approval to conduct phone

2 JACC Vol. 43, No. 3, 2004 February 4, 2004:410 5 Rahman et al. Noninvasive Diagnosis of Cardiac Amyloidosis 411 Abbreviations and Acronyms CHF congestive heart failure ECG electrocardiogram/electrocardiographic IVS interventricular septal LV left ventricular NYHA New York Heart Association OR odds ratio calls to patients or their surviving family members to determine accurate mortality information. Informed consent was obtained during the phone conversations. Seventytwo percent of patients or their surviving family members were reached by phone. The National Death Registry was used to confirm the status of the remainder. A review of the medical records was performed by a team of physicians blinded to the endomyocardial biopsy results. Biopsy results were separated from other patient information and introduced into the data base after collection of clinical data was completed. The following information was reviewed and recorded: demographic data, New York Heart Association (NYHA) heart failure class at the time of biopsy, ECG findings, echocardiographic results, and hemodynamic parameters derived from right heart catheterization performed at the time of biopsy. The ECGs were analyzed for the following characteristics: rhythm, conduction abnormalities (i.e., right or left bundle branch block), a low-voltage pattern (defined by total height of the QRS complex in the limb leads 5 mm and 10 mm in the precordial leads), a pseudo-infarction pattern (pathologic Q waves on the ECG, but no coronary artery disease by angiography), and a left or right atrial abnormality. The echocardiograms were analyzed for the following characteristics: interventricular septal (IVS) thickness, posterior wall thickness, left ventricular (LV) diastolic diameter, LV systolic diameter, atrial size, pericardial effusion size (if present), overall ejection fraction, and granular/ sparking appearance of the myocardium by visual inspection. Standard hemodynamic data obtained from right heart catheterizations were reviewed. Endomyocardial biopsy. The endomyocardial biopsies were obtained in the standard fashion. A minimum of four biopsies were taken from each patient. All biopsies were examined by a single, experienced pathologist who was blinded to all other study data. The biopsies were analyzed with a standard hematoxylin-eosin preparation. All biopsies were also stained with congo red to identify the presence of amyloidosis. If the congo red staining was equivocal, the biopsy was reviewed under electron microscopy to confirm the diagnosis. Finally, all biopsies with amyloidosis underwent immunohistochemical staining to help ascertain the exact etiology of the amyloid. Staining was performed for amyloid AL and transthyretin. They were also stained for thioflavin T, a more sensitive stain for amyloidosis (compared with congo red staining), but not specific for the Table 1. Summary of Biopsy and Survival Results source of the amyloid. Stains for lambda and kappa light chains were available in a limited number of the patients. Statistical analysis. All statistical analyses were performed using the statistical software package STATA version 7.0. As an initial step, univariate analyses were performed on all demographic and cardiac variables of interest to determine which set of variables was statistically associated with cardiac amyloidosis. Continuous variables were assessed using the parametric t test for independent samples (e.g., mean arterial pressure, right ventricular systolic pressure), and all categorical variables were assessed using the chisquare goodness-of-fit test (e.g., square root sign). Those variables that were found to be associated with disease status were then used to build multivariate logistic regression models that predict the log-odds of being positive for amyloidosis on endomyocardial biopsy. We first considered models with only noninvasive cardiac variables and then added invasive measures to see whether these variables would increase our ability to predict disease status. RESULTS Biopsy Results Patients (n 196) Mean Duration of Survival After Biopsy (days) Amyloidosis Amyloidosis and myocarditis 8 66 Myocyte hypertrophy with/ 66/35 1,114 without interstitial fibrosis Myocarditis 22 1,025 Nonspecific findings Biopsy and survival results. A total of 196 patients were evaluated. Fifty-eight of these patients (29.1%) had an endomyocardial biopsy diagnostic of cardiac amyloid. Eight of these patients also had lymphocytic infiltrates, which met the Dallas criteria for myocarditis. Of the remaining 138 patients (70.9%), the most common diagnosis was hypertrophy with evidence of interstitial fibrosis (which occurred in 66 patients). Thirty-five patients had hypertrophy without interstitial fibrosis, and 22 patients had myocarditis (Table 1). The mean post-biopsy survival varied with the biopsy result. For all 58 patients with amyloidosis, the mean survival duration was significantly shorter than that of the rest of the patients (364 vs. 1,053 days, p 0.001). Moreover, the subset of eight amyloidosis patients who had evidence of myocardial inflammation had an even poorer survival (66 days, p 0.001). All other patients with diagnoses had a survival duration of 1,000 days, except for patients with interstitial fibrosis or an indeterminate biopsy. Patient characteristics. The demographic, clinical, and hemodynamic characteristics of the patients, according to their biopsy results, are shown in Tables 2 and 3. Patients in the amyloidosis group were slightly older and more often male. The duration of symptoms before obtaining the

3 412 Rahman et al. JACC Vol. 43, No. 3, 2004 Noninvasive Diagnosis of Cardiac Amyloidosis February 4, 2004:410 5 Table 2. Summary of Demographic and Background Medical Data Amyloid (n 58) Negative Biopsy (n 138) p Value Mean age (yrs) Gender (%) Male 72% 57% Female 28% 43% Race (%) Caucasian 69% 70% NS African American 31% 25% NS CHF class at presentation I II 19% 37% III 47% 43% IV 33% 20% Mean duration of symptoms NS (months) Elevated creatinine (% of 21% 11% patients with 2 mg/dl) Thrombocytopenia (% of patients with 150 K) 23% 18% NS CHF congestive heart failure; NS not significant. biopsy was no different between the groups. However, the amyloidosis group had a larger percentage of patients with more advanced heart failure (i.e., NYHA class III/IV). The hemodynamic data are also consistent with a greater degree of hemodynamic compromise in the amyloidosis group. Electrocardiography. Table 4 summarizes the ECG findings. Both low-voltage and pseudo-infarction patterns were statistically more prevalent in the patients with cardiac amyloidosis. Left bundle branch block was found to be less prevalent in the amyloidosis group (4% vs. 14%), but this did not quite achieve statistical significance (p 0.065). Echocardiography. Table 5 summarizes the echocardiographic findings. The mean IVS thickness and posterior Table 3. Clinical and Hemodynamic Characteristics According to Biopsy Result Amyloid (n 58) Negative Biopsy (n 138) p Value HR (beats/min) NS MAP (mm Hg) RVSP (mm Hg) NS RVDP (mm Hg) NS MPAP (mm Hg) PCWP (mm Hg) CI (l/min per m 2 ) SV (ml) SVRI (dynes s m 2 /cm 5 ) 3, , LVSWI (g/m per m 2 ) Square root sign (%) Equalization of diastolic pressure (%) CI cardiac index; HR heart rate; LVSWI left ventricular stroke work index; MAP mean arterial pressure; MPAP mean pulmonary artery pressure; NS nonsignificant; PCWP pulmonary capillary wedge pressure; RVDP and RVSP right ventricular diastolic and systolic pressure, respectively; SV stroke volume; SVRI systemic vascular resistance index. Table 4. Electrocardiographic Findings Amyloid (n 48) Negative Biopsy (n 128) Chi-Square Result Atrial flutter 0% 3% Atrial fibrillation 19% 30% Left atrial abnormality 15% 20% Right atrial abnormality 0% 4% First-degree AV block 13% 5% Interventricular conduction 8% 10% delay LBBB 4% 14% RBBB 17% 12% Paced rhythm 8% 20% Low voltage 56% 13% Pseudo-infarction 60% 36% AV atrioventricular; LBBB and RBBB left and right bundle branch block; respectively. wall thickness were significantly larger in patients with amyloidosis (1.6 vs. 1.2 cm, p ). Also, the LV systolic and diastolic diameters were both significantly smaller in patients with amyloidosis. Patients with amyloidosis were also more likely to have restrictive physiology (increased E wave and decreased A wave) and a sparkling/ granular appearance to the myocardium. Immunohistochemical staining. Immunohistochemical staining with congo red was performed on all patients to ascertain the etiology of amyloidosis (Table 6). There were seven patients who stained positive for amyloid AL, suggesting that light-chain deposition from multiple myeloma was the etiology of their amyloidosis. Seven patients stained positive for transthyretin amyloidosis. Two patients stained positive for both amyloid AL and transthyretin. No other specific immunohistochemical stains were performed (e.g., for amyloid AA). However, staining for thioflavin T (a more specific marker for amyloidosis than congo red) was performed. A total of 34 patients (including the 13 patients with amyloid AL and transthyretin) stained positive. Therefore, 24 patients were congo red positive but did not have any positive immunohistochemical stains. Logistic regression models. Table 7 summarizes the logistic regression models for the probability of biopsy-proven amyloidosis, using the independent variables of IVS thickness, low voltage, patient age, gender, and race (white vs. nonwhite). With respect to the odds ratios (OR), the categorical variable base levels are female, white, and the absence of a low voltage on the ECG. For the continuous variables of IVS thickness and age, the ORs are given for each unit increase in the variable. Thus, in model 1, the OR of amyloidosis is 1.97 times greater for males than for females, 1.75 times greater for whites than nonwhites, and 8.38 times greater for patients with a low voltage than a normal voltage. In addition, each year increase in age gives an OR of 1.02 for a positive result, and each unit increase in IVS thickness produces a 7.7-fold increased risk of having cardiac amyloidosis. However, the ORs are statistically different from 1.0 for IVS thickness and low voltage only.

4 JACC Vol. 43, No. 3, 2004 February 4, 2004:410 5 Rahman et al. Noninvasive Diagnosis of Cardiac Amyloidosis 413 Table 5. Echocardiographic Findings Amyloid (n 47) Negative Biopsy (n 115) p Value LV diastolic diameter (cm) LV systolic diameter (cm) Left atrial size (cm) IVS thickness (cm) Posterior wall thickness (cm) Restrictive physiology (increased E/A) 14 (30%) 13 (11%) Sparkling/granular appearance of myocardium 12 (26%) 5 (4%) Biatrial enlargement 13 (28%) 44 (38%) E/A reversal 7 (15%) 26 (23%) Wall motion abnormality 8 (17%) 25 (22%) LV ejection fraction 44% 42% Data are presented as the mean value or number (%) of patients. E/A ratio of early transmitral flow velocity to atrial flow velocity; IVS interventricular septal; LV left ventricular. Therefore, a more parsimonious model (model 2) that includes only IVS thickness and low voltage may be considered. In this model, the OR of cardiac amyloidosis is 7.34 times greater for patients with a low voltage than a normal voltage, and for every 1-cm increase in IVS thickness, there is 10.4-fold higher likelihood of a positive biopsy. This model can also be used to formulate a simple yet powerful test for the likelihood of amyloidosis. By utilizing the mathematic equations that define this model, one is able to define a cut-off point for IVS thickness, which can serve as a testing parameter (Appendix). We have demonstrated that for patients with a low voltage, a cut-off point of 1.98 cm for IVS thickness optimizes both the sensitivity and specificity of our model to predict the presence of amyloidosis. Similarly, a cut-off point of 1.13 cm for IVS thickness can be defined as the testing point that maximizes the sensitivity and specificity of the test for patients without a low voltage. Thus, in our test, a patient with an ECG that shows a low voltage and an echocardiogram with an IVS thickness of 1.98 cm will be classified as having amyloidosis. Likewise, a patient with a low voltage but IVS thickness 1.98 cm would be classified as not having amyloidosis. Our test yields a positive predictive value of 79% and a negative predictive value of 88%. Table 8 shows the sensitivity, specificity, and positive and negative predictive values of our test. DISCUSSION The diagnosis of cardiac amyloidosis is important for prognosis and therapy. The clinical presentation varies and Table 6. Immunohistochemical Staining Results Stain Number of Patients Amyloid AL 7 Transthyretin (TTR) 4 Amyloid AA and TTR 2 Thioflavin T 21 Congo red only 24 may mimic other infiltrative cardiomyopathies or storage disorders, as well as hypertrophic cardiomyopathy. Previous investigations have shown that multiple ECG and echocardiographic findings suggest amyloid but are nonspecific in isolation. Typically, there is echocardiographic evidence of thickened ventricular and septal walls, a normal or small ventricular size, atrial enlargement, and refractile myocardium (3), as well as a low voltage on the ECG. The finding of increased myocardial echogenicity is 87% sensitive and 81% specific for cardiac amyloid when compared with control subjects with LV hypertrophy (4). The use of a combination of echocardiographic and ECG features increases specificity. In the same study, increased echogenicity plus increased atrial septal thickness is 67% sensitive and 100% specific for cardiac amyloid. However, prospective identification of refractile echocardiograms is not precise due to observer dependence and technical manipulations during imaging (1). A distinctive characteristic of amyloid cardiomyopathy is the inverse relationship between ECG voltage and LV mass (5). A low voltage on the ECG and increased septal and posterior LV wall thickness on the echocardiogram are highly specific for cardiac amyloidosis in the setting of biopsy-proven systemic amyloidosis (6). To the best of our knowledge, the diagnostic parameters of the aforementioned echocardiographic and ECG features have not been determined in a cohort of patients with endomyocardial biopsy-proven amyloidosis. In this study, we aimed to establish the sensitivity and specificity of the typical echocardiographic and ECG features of cardiac amyloidosis, as well as a model derived from noninvasive parameters that is predictive of disease status. Logistic regression models that predict the log-odds of cardiac amyloidosis were developed, based on noninvasive parameters. Multiple models were used to determine the diagnostic parameters of sensitivity, specificity, and predictive value. Models using myocardial echogenicity paired with wall thickness or low voltage were highly predictive of cardiac amyloidosis, but not more so than the selected model utilizing septal thickness and low voltage. Use of

5 414 Rahman et al. JACC Vol. 43, No. 3, 2004 Noninvasive Diagnosis of Cardiac Amyloidosis February 4, 2004:410 5 Table 7. Summary of Logistic Regression Models Variable Model 1 Model 2 OR (CI) p Value OR (CI) p Value Age ( ) Gender ( ) Race ( ) IVS thickness ( ) ( ) Low voltage ( ) ( ) n Deviance H-L p value In addition to presenting the odds ratios (OR) and corresponding confidence interval (CI) for each independent variable, this table also gives the total number of patients included in the model, the model deviance, the p values from the Hosmer-Lemeshow (H-L) goodness-of-fit test, and the area under the curve obtained from receiver-operating characteristics analysis. The deviance of the regression model is a measure of the model s fidelity to the data. Ideally, we would like to seek models that have a high fidelity (e.g., low deviance) but that are also parsimonious (e.g., with as few a number of independent variables as possible). The H-L test is a goodness-of-fit test that compares the model s predictions with the data. The null hypothesis is that the model fits the data, so a nonsignificant p value for this test suggests that we cannot reject this hypothesis. IVS interventricular septal. these variables is simpler and lessens the variability associated with identifying granular myocardium. We propose that in patients with clinical suspicion of cardiac amyloid, the combination of septal thickness and low voltage can be used for diagnosis. Additionally, serial echocardiography can be used to follow the patient s clinical status and response to treatment. Others have demonstrated that increasing LV wall thickness and mass/voltage ratio are associated with shorter survival (7). Study limitations. This study has several limitations. Patients were referred for suspicion of cardiac amyloidosis and therefore likely had an increased incidence of ECG and echocardiographic evidence of myocardial involvement. However, this would lessen any observed difference between the groups. A chart review did not provide sufficient information to determine the NYHA heart failure class at the time of echocardiography and endomyocardial biopsy. It has been demonstrated that clinical CHF strongly correlates with the degree of echocardiographic abnormalities, in particular, increased wall thickness (7). In this study, a higher incidence of clinical CHF in those with histologic evidence of cardiac amyloid would falsely create or amplify a difference in echocardiographic abnormalities between groups. However, it has been demonstrated that ECG and echocardiographic changes consistent with amyloid are detectable before the development of CHF (8). Another confounding factor may be the combined analysis using primary and secondary amyloidosis. It is known that the type and amount of cardiac dysfunction is related to the type and extent of amyloid infiltration. At autopsy, it has been demonstrated that septal thickness is higher in primary Table 8. Predictive Ability of Model 2 Sensitivity 72.22% Specificity 91.25% Positive predictive value 78.79% Negative predictive value 87.95% and familial amyloidosis than in secondary amyloidosis (9). Studies comparing echocardiographic findings between types of amyloidosis have yielded conflicting results. One study demonstrated an increased IVS thickness in primary amyloidosis compared with familial and secondary amyloidosis (10), but another study did not show any significant difference in septal thickness between primary and familial amyloidosis (11). It is also known that amyloid AA rarely involves the heart (12), so the potential effect on interpretation of this data is unclear. Cardiac amyloidosis is associated with a variable prognosis. In primary amyloidosis, the median survival is six months once heart failure develops. For senile amyloidosis, the median survival is five years once heart failure develops (13). Because of therapeutic and prognostic implications, it is important that a tissue diagnosis is made. We propose that if a histopathologic diagnosis is available from other tissue, that the findings of increased septal thickness and low voltage are sufficient for the diagnosis of cardiac amyloidosis, as the specificity is high for patients with suspected cardiac involvement. Recent research from Dispenzieri et al. (14) suggests that elevated cardiac troponins may also be useful in the detection and prognosis of cardiac amyloidosis in patients with known systemic amyloidosis. Recommendations for further study. It is noted that this study identified a subset of patients having evidence of both amyloidosis and myocarditis. This histopathologic finding was strikingly associated with a worse prognosis. To the best of our knowledge, this has not been previously reported. In this subset of patients, the ECG and echocardiographic findings were not significantly different from those with other forms of amyloidosis. This finding would favor the use of endomyocardial biopsy over noninvasive measures in the diagnosis of cardiac amyloid, as this histologic finding has important prognostic implications. We recommend further study to define this pathologic entity and its clinical features.

6 JACC Vol. 43, No. 3, 2004 February 4, 2004:410 5 Rahman et al. Noninvasive Diagnosis of Cardiac Amyloidosis 415 Reprint requests and correspondence: Dr. Edward K. Kasper, Chief of Cardiology, Johns Hopkins Bayview Medical Center, 4940 Eastern Avenue, Baltimore, Maryland ekasper@jhmi.edu. REFERENCES 1. Picano E, Pinamonthi B, Ferdeghini EM, et al. Two-dimensional echocardiography in myocardial amyloidosis. Echocardiography 1991; 8: Falk RH, Comenzo RL, Skinner M. The systemic amyloidoses. N Engl J Med 1997;337: Siqueira-Filho AG, Cunha CL, Tajik AJ, Seward JB, Schattenberg TT, Giuliani ER. M-mode and two-dimensional echocardiographic features in cardiac amyloidosis. Circulation 1981;63: Falk RH, Plehn JF, Deering T, et al. Sensitivity and specificity of the echocardiographic features of cardiac amyloidosis. Am J Cardiol 1987;59: Carroll JD, Gaasch WH, McAdam KP. Amyloid cardiomyopathy: characterization by a distinctive voltage/mass relation. Am J Cardiol 1982;49: Hamer JP, Janssen S, van Rijswijk MH, Lie KI. Amyloid cardiomyopathy in systemic and non-hereditary amyloidosis: clinical, echocardiographic and electrocardiographic findings in 30 patients with AA and 24 patients with AL amyloidosis. Eur Heart J 1992;13: Cueto-Garcia L, Reeder GS, Kyle RA, et al. Echocardiographic findings in systemic amyloidosis: spectrum of cardiac involvement and relation to survival. J Am Coll Cardiol 1985;6: Backman C, Olofsson BO. Echocardiographic features in familial amyloidosis with polyneuropathy. Acta Med Scand 1983;214: Roberts WC, Waller BF. Cardiac amyloidosis causing cardiac dysfunction: analysis of 54 necropsy patients. Am J Cardiol 1983;53: Moyssakis I, Triposkiadis F, Rallidis L, Hawkins P, Kyriakidis M, Nihoyannopoulos P. Echocardiographic features of primary, seconday, and familial amyloidosis. Eur J Clin Invest 1999;29: Dubrey SW, Ha K, Skinnee M, La Valley M, Falk RH. Familial and primary cardiac amyloidosis: echocardiographically similar diseases with distinctly different clinical outcomes. Heart 1997;78: McCarthy RE, Kasper EK. A review of the amyloidoses that infiltrate the heart. Clin Cardiol 1998;21: Cacoub P, Axler O, De Zuttere D, et al. Amyloidosis and cardiac involvement. Ann Med Interne (Paris) 2000;151: Dispenzieri A, Kyle RA, Gertz MA, et al. Survival in patients with primary systemic amyloidosis and raised serum cardiac troponins. Lancet 2003;361: APPENDIX Based on the following logit coefficients for model 2: logit *X 2.341*IVS where X 0 if low voltage is false and1if true, and the relationship between the probability of a positive response and the logistic coefficients: prob cardiac amyloidosis exp *X 2.341*IVS 1 exp *X 2.341*IVS we can solve for the values of IVS thickness and low voltage that give the cut-off probability of 0.5. For patients with a low voltage, the cut-off value of IVS thickness is equal to 1.98 cm, and the cut-off value of IVS thickness for a normal voltage is equal to 1.13 cm.