Cardiovascular Magnetic Resonance and prognosis in cardiac amyloidosis

Amyloidosis is an uncommon condition caused by the deposition of misfolded, insoluble aggregated protein with a characteristic β-sheet structure in tissues throughout the body [1]. Cardiac involvement is frequent in systemic amyloidosis of immunoglobulin light chain (AL) and transthyretin (TTR) types, is associated with a poor prognosis [2], and can have therapeutic implications. Accumulation of amyloid in the myocardial interstitium [3] leads to diastolic dysfunction and restrictive cardiomyopathy that progresses to overt heart failure and death [4]. Cardiac involvement is the cause of death in approximately half of patients with AL amyloidosis [5].

Cardiac amyloidosis is characterized histologically by infiltration and expansion of the interstitial space with amyloid protein, along with some associated endomyocardial fibrosis [6]. We have previously reported that cardiovascular magnetic resonance (CMR) frequently shows a characteristic pattern of global subendocardial late gadolinium enhancement (LGE) in cardiac amyloidosis that accords with the transmural histological distribution of amyloid [6]. However, there are also abnormal myocardial and blood pool gadolinium kinetics which are likely to reflect cardiac amyloid load, and therefore might relate to prognosis in these patients. We examine here the hypothesis that LGE and gadolinium kinetics might be of prognostic value in cardiac amyloidosis.

We prospectively followed our previously reported cohort of 29 patients with proven cardiac amyloidosis from the National Amyloidosis Centre of the United Kingdom [6], who were recruited and scanned between August 2002 and April 2003. The study was approved by the local Ethical Committee and written, informed consent was obtained from all patients. The amyloidosis was of TTR type in 4 patients, and monoclonal light chain (AL) type in 25 cases. The CMR methodology and baseline characteristics of these subjects have been published previously [6]. Briefly, all patients underwent biopsy (2 cardiac, 27 non-cardiac sites), 2D-echocardiography and Doppler studies, ¹²³I-SAP scintigraphy to measure extra-cardiac amyloid burden, serum NT pro BNP assay, and CMR. CMR was performed on a 1.5T scanner (Siemens Sonata, Erlangen, Germany) with acquisition of fast imaging with steady-state free precession 2 chamber, 4 chamber and contiguous short-axis breath-hold cines (7 mm slice thickness, repetition time/echo time of 3.2/1.6 ms; temporal resolution 25 ms; retrospective gating; pixel size 2.4 × 1.5 mm; flip angle 60°; acquisition time 18 heartbeats) for mass and volumes measurements. For LGE, a peripheral bolus injection (0.1 mmol/kg) of gadolinium-DTPA (Schering, Berlin, Germany) was given. A T₁ mapping method was developed consisting of a magnetization prepared segmented FISP cine with a 30 ms increment in inversion time for each frame. This was run every 2 minutes after the gadolinium bolus starting immediately after gadolinium injection and prior to late enhancement imaging, which was only started after 10 minutes. Segmentation was 13–25 lines, with triggering every 2–3 heart beats. Subsequent to this gadolinium kinetics imaging, late gadolinium enhancement images were also acquired using a segmented inversion recovery sequence (segmentation was 13 to 25 lines, with triggering every 2 to 3 heartbeats).

Echocardiography was performed on a GE Vingmed System with the use of standard techniques, as previously described [6]. In brief, left ventricular (LV) wall thickness was measured from the M-mode at the level of the chordae. Diastolic dysfunction at the time of study entry was assessed by PW Doppler of transmitral and pulmonary venous inflow velocities. These were measured in at least 3 consecutive beats and averaged for each measurement of the following: transmitral flow – peak velocity of early (E) and late (A) filling waves; E/A ratio; E-wave deceleration time; pulmonary venous flow: peak velocity of systolic, diastolic and A reversal waves. Abnormal diastolic function was classified according to standard criteria [7, 8], in 3 dysfunctional filling patterns: slow relaxation, pseudonormal, and restrictive.

For patients with AL amyloidosis, the haematological response to chemotherapy was categorized into no response, partial response or complete response, according to whether aberrant amyloidogenic serum free light chain (FLC) concentration had fallen <50%, ≥50% or completely following chemotherapy [9].

Accumulation of amyloid in the myocardial interstitium results in late gadolinium enhancement (LGE), often with a predominant diffuse, global and subendocardial distribution that matches the distribution of amyloid on histology [6, 11, 12], although other more focal patterns have also been reported [13]. This is associated with substantial alterations in gadolinium kinetics, with faster washout of gadolinium from blood and myocardium than normal [6]. This study extends the value of these diagnostic findings, and shows for the first time that CMR may yield useful prognostic information in patients with cardiac amyloidosis. The presence alone of LGE did not correlate significantly with survival, although the sample size of this study was relatively small, but gadolinium kinetics were associated with significant survival differences according to several derived parameters. A likely explanation for the discrepancy between LGE and the kinetics with outcome is the superior discrimination by gadolinium kinetics for the severity and transmurality of the myocardial amyloid burden. This is illustrated by the results of the intramyocardial T1 difference parameter. Previous work showed that the usual pattern of amyloid protein deposition in the myocardium is predominantly subendocardial with variable transmural extension, [6] under which circumstances the subepicardial (low amyloid, low gadolinium, high T1) minus subendocardial T1 (higher amyloid, higher gadolinium, lower T1) difference would be expected to be high. With greater myocardial amyloid deposition, including greater deposition in the subepicardium, the transmural difference in amyloid diminishes, leading to a reduction in the intramural T1 gradient. This is supported by the other significantly predictive derived parameter in which myocardial T1 is compared with blood T1; the subepicardial minus blood T1 difference also predicted survival, with lower values associated with death. Usually blood contains more gadolinium (lower T1) because of a substantially greater volume of distribution, and therefore a large positive T1 gradient between subepicardium and blood exists. This gradient would be reduced if greater subepicardial amyloid (lower T1) is present. Of interest is that the subendocardial minus blood T1 difference did not show differences in survival. It therefore appears that deposition in the subepicardium may be the principal statistical driver for prediction of mortality. The explanation for this finding can be that subendocardial amyloid deposition is usual, and it is not until the burden of amyloid deposition increases that the subepicardium has substantial involvement. Thus it is possible that subepicardial involvement partially reflects the total myocardial amyloid burden. However, the direct predictive value for death of subepicardial T1 was less than the derived measures already mentioned (P = 0.039; table 2; Kaplan Meier curves not significant), and therefore the derived measures are more important.

Cardiac involvement in amyloidosis is a major risk factor for adverse outcome [14]. Previously proposed associations with poor prognosis in cardiac amyloidosis include reduced ejection fraction, low ECG voltages, increased left ventricular wall thickness on echo, and the type of amyloidosis (with worse prognosis in AL compared with TTR type) [15]. In addition, the degree of diastolic dysfunction [16] and suppression of amyloidogenic serum light chains by chemotherapy, and lower baseline values and greater reductions in NT-proBNP have been associated with improved outcome [9, 17]. Our data now suggest that gadolinium kinetics may be even more predictive than these measures. The value of the CMR measurements may in part be due to the fact that cardiac amyloid burden cannot be measured satisfactorily by other techniques, and therefore CMR may offer a fundamental new window into the cardiac pathology in this disease.

Treatment options in amyloidosis have been expanding [18], and some novel pharmaceuticals have lately shown considerable promise [19, 20]. Patients with a particular poor prognosis may benefit from sequential heart and autologous stem cell transplantation [21‐23]. Recognition that T1 mapping in cardiac amyloidosis may be significantly more predictive of poor prognosis than the other currently used measures may be beneficial in the clinical management of AL patients by identifying those in whom early use of more intensive chemotherapy might be justified with the aim of achieving more rapid and complete remission of their underling clonal plasma cell disease. It is of interest that there is precedent for the burden of myocardial disease as assessed by CMR being predictive of outcome in hypertrophic, dilated and siderotic cardiomyopathies [24‐26], with the severity of fibrosis [24, 25, 27] and iron loading [28].

A number of the gadolinium kinetics parameters in this study were significantly associated with mortality, but the one with greatest discriminatory value was the intra-myocardial T1 gradient after gadolinium injection, with 95% accuracy at a threshold value of 23 ms (Kaplan Meier analysis P = 0.002). Although further experience and reproduction of these results by other centres is necessary, the technique is in principle straightforward and could be implemented on most 1.5T scanners. Optimisation of the acquisition sequence and analysis software would also be valuable.

This study shows that gadolinium kinetics assessed by T1 relaxation CMR is useful in the prognostic assessment of cardiac amyloidosis and this could have utility in assessing patients for treatments, and in their follow-up.