Quantitative SPECT/CT parameters of myocardial 99mTechnetium-3,3-diphosphono-1,2-propanodicarboxylic acid (DPD) uptake in suspected cardiac transthyretin amyloidosis

Background 99mTc-labelled bisphosphonates are used for imaging assessment of patients with transthyretin cardiac amyloidosis (ATTR). Present study evaluates whether quantitative SPECT/CT measurement of absolute myocardial 99mTc-labelled 3,3-diphosphono-1,2-propanodicarboxylic acid (Tc-DPD) uptake can diagnose patients with suspected ATTR. Methods Twenty-eight patients (25 male, age 80.03 ± 6.99 years) with suspected ATTR referred for Tc-DPD imaging had planar and SPECT/CT imaging of the chest. Three operators independently obtained Tc-DPD myocardial SUVmax and SUVmean above threshold (SMaT) (20, 40 and 60% of SUVmax), using a semi-automated threshold segmentation method. Results were compared to visual grading (0–3) of cardiac uptake. Results Twenty-two patients (78%) had cardiac uptake (2 grade 1, 15 grade 2, 5 grade 3). SUVmax and SMaT segmentation thresholds enabled separating grades 2/3 from 0/1 with excellent inter- and intra-reader correlation. Cut-off values 6.0, 2.5, 3 and 4 for SUVmax, SMaT20,40,60, respectively, separated between grades 2/3 and 0 /1 with PPV and NPV of 100%. SMaT20,40,60(cardiac)/SUVmean (liver) and SMaT20,40,60(cardiac)/SUVmean(liver/lung) separated grades 2 and 3. Conclusion Quantitative SPECT/CT parameters of cardiac Tc-DPD uptake are robust, enabling separation of patients with grades 2 and 3 cardiac uptake from grades 0 and 1. Larger patient cohorts will determine the incremental value of SPECT/CT quantification for ATTR management.


Introduction
Cardiac amyloidosis is characterized by protein misfolding and myocardial deposition mainly of monoclonal light chain (AL) or transthyretin (ATTR), resulting in restrictive cardiomyopathy and heart failure [1,2]. ATTR amyloidosis may be acquired, associated with wild-type transthyretin (TTR), or hereditary, associated with TTR gene variants. Cardiac ATTR amyloid deposits are present in up to 25% elder individuals [3], more common among patients with heart failure and preserved ejection fraction (HFpEF) [4].
Planar and SPECT images interpreted visually, by grading the intensity of myocardial uptake as compared to skeletal activity or by semi-quantitative techniques, measuring heart/whole body retention or heart/contralateral lung uptake on planar scans and have high sensitivity and specificity in cases with either intense or absent uptake, but are less accurate in equivocal radiotracer activity [2, 5-7, 10, 11].
New emerging therapies either prevent TTR amyloid formation [12,13] or inhibit TTR expression [14], with promising results. A sensitive and accurate method to quantify amyloid burden is needed to evaluate treatment response. Quantification of myocardial Tc-DPD uptake may overcome shortcomings of planar studies in borderline cases and may improve the diagnostic performance, monitor the amyloid burden over time and determine ATTR treatment response.

Patients
This retrospective two center international study includes 30 consecutive patients with cardiomyopathy and suspected cardiac amyloidosis based on symptoms, monoclonal protein studies, echocardiography and/or MRI. Biopsies were performed in 14 patients, including 6 endomyocardial, and genetic testing in 2 cases. Patients were referred for Tc-DPD SPECT/CT for routine clinical work-up between May 2013 and June 2017 at Chaim Sheba Medical Center, Israel and University College London Hospitals, UK. The study received IRB approval in both institutions and the need for informed consent was waived.

Tc-DPD studies
Planar whole body imaging was acquired at 5 and 180 min after i.v. injection of 743 ± 102 MBq of Tc-DPD using a SPECT/CT (Discovery 670, GE Healthcare) with low energy, high resolution collimators for a scan speed of 13 cm/min. SPECT/CT of the chest was performed with 180° L-mode SPECT, from 45° LAO to 45° LPO.
Step-and-shoot mode data were acquired every 3°, 30 s/step, with two 15-s frames in each step. CT parameters were 120 kV, 120 mA, slice thickness 2.5 mm. Data reconstruction used OSEM (4 iterations, 10 subsets) and were corrected for motion, attenuation, scatter and collimator blurring. Attenuation correction quality control ascertained good registration of SPECT and CT.
Visual assessment: Cardiac Tc-DPD uptake was assessed visually on planar scintigraphy and SPECT by Nuclear Medicine physicians according to the expert consensus recommendations [22], scored as grade 0absent cardiac, normal skeletal uptake; grade 1-mild cardiac uptake, inferior to rib; grade 2-moderate cardiac uptake equal to rib; grade 3-high cardiac uptake, mild or absent rib uptake.
Quantitative assessment: Quantification of Tc-DPD uptake was performed using Q.Volumetrix MI on a Xeleris workstation version 4 DR (GE Healthcare), given information of radioisotope, dose and time of injection. Tracer uptake (MBq/ml) or percent of injected dose (%ID) are calculated in a volume of interest (VOI), generated automatically or interactively by the user. Standardized uptake values (SUV) are calculated adding patient height and weight (Fig. 1).
Cardiac uptake: Segmentation of Tc-DPD uptake in the left ventricle (LV) wall was performed by drawing boundaries (a "Master VOI") based on the CT images of the co-registered SPECT-CT study (cyan lines, Fig. 1). The Master VOI was automatically projected to the co-registered SPECT images. The Master VOI of the left ventricle based on non-enhanced CT images includes also the lumen. This may lead to inclusion of blood pool uptake when projecting the master VOI on the SPECT images. Therefore, a threshold-based method was applied for segmenting Tc-DPD uptake in the corresponding SPECT images, as follows: within the Master VOI, SUVmax was automatically detected by the Q.volumetrix MI software. Then, 3 threshold value were applied to segment uptake at 20%, 40% and 60% of SUVmax. All voxels containing SUV values above the threshold inside the Master VOI were automatically segmented (red lines, Fig. 1). In addition to SUVmax, an average (mean) of SUV values in voxels selected by the threshold-based segmentation was calculated automatically by the Q Volumetrix MI software (SUV Mean above threshold-SMaT 20 , SMaT 40 and SMaT 60 for 20%, 40% and 60% of SUVmax, respectively) . Values obtained by applying SMaT 40 are depicted in line 2, table of Fig. 1). The exploration of 3 threshold values was driven by the need to find the most appropriate threshold approach (e.g. that would most accurately separate between grades of myocardial uptake).
Inter-observer variability was assessed by three operators who performed quantitative analysis for each study. One operator repeated analysis of all studies after four weeks, to assess intra-reader reproducibility.
Uptake in bone, lung, liver and soft tissue: Bone and lung SUVmean were calculated on SPECT inside segmented volumes obtained from an automatic CT-HU threshold-based segmentation algorithm (green and magenta lines respectively, Fig. 1).
For liver and soft tissue SUVmean 25 ml and 15 ml spheres were placed by operator in liver (blue lines, Fig. 1) and shoulder muscle (yellow lines, Fig. 1), respectively.
Cardiac SUV max , SMaT 20 , SMaT 40 and SMaT 60 were normalized to SUVmean of lungs, liver, bones and soft tissue, to SUVmean liver/lungs and to SUVmean bone/ soft tissue.

Statistical analysis
Continuous variables are presented as mean ± standard deviation (SD). Differences between datasets separated by grade of myocardial uptake were assessed with Mann-Whitney U test. p < 0.05 was statistically significant.
SUVmax and SMaT that could separate between grades of myocardial uptake were defined as cut-off value with highest true positive rate (TPR) and lowest false positive rate (FPR). Positive and negative predictive value (PPV, NPV) were also calculated.
Inter-reader correlation was calculated using Intraclass Correlation Coefficient (ICC, MedCalc Statistical Software version 18.11.3, Ostend, Belgium). Intra-reader reproducibility was calculated by the Bland-Altman method.

Results
The study included 30 patients. One patient with grade 2 uptake and one patient with grade 3 uptake and chronic renal failure who required dialysis and had incomplete data were excluded from the quantitative analysis. Of the remaining 28 patients (25 male, age 80.03 ± 6.99 years) 22 patients (78%, 20 male, age 80.04 ± 7.26 years) had Tc-DPD cardiac uptake. Two had grade 1, 15 grade 2 and five had grade 3 uptake. Six patients had no cardiac uptake (grade 0). ATTR was diagnosed in 21/22 patients with positive Tc-DPD studies and was excluded in one patient with known AL amyloidosis (Table 1).
Three readers' inter-reader correlation coefficients were all above 0.99 and intra-reader correlation coefficients ranged between 0.91 and 1.0 (Table 4).
With increasing grade of myocardial uptake the activity in bone and liver decreased significantly, (p = 0.01, p = 0.03, respectively). There was a tendency for higher soft tissue and lung activity with increasing grade of myocardial uptake, but not statistically significant (p = 0.60 and 0.98, respectively) (Figs. 2, 3).
SUVmax cut-off value of 6.0 separated between cardiac uptake grades 2 and 3 compared to 0 and 1 with FPR 0% and TPR, PPV and NPV 100%. Same results were obtained for SMaT 20  tissue) for all three thresholds were significantly different between grades 0 or 1 and 2 or 3 (p < 0.001), but not between grades 2 and 3 (Table 3).

Discussion
Endomyocardial biopsy is the gold standard for diagnosing ATTR, but does not provide information regarding disease extent, prognosis, progression, and response to therapy [1,[12][13][14].
Bone-seeking radiotracers are sensitive and specific for ATTR using a visual grading score or semi-quantitative assessment of myocardial uptake in cases with either intense or absent cardiac activity, but less accurate in equivocal cases [5-8, 10, 11, 22]. Grade 2 or 3 Tc-DPD/ PYP/HMDP uptake with no monoclonal antibodies in blood and urine has specificity and PPV above 98% for diagnosis of ATTR [10].
Novel therapeutic agents recently introduced may require precise assessment of disease burden at diagnosis and during follow up [1,[17][18][19].
Present results in 28 consecutive patients with clinically suspected ATTR indicate that quantitative SPECT/ CT measurements of Tc-DPD cardiac uptake are feasible and robust. We observed an increase in SUVmax and SMaT of cardiac Tc-DPD uptake with higher visual grade ( Table 2, Fig. 2). Normalization of the cardiac SUVmax or SMaT to that of the liver and to the liver/lungs ratio separated between grades 2 and 3 Tc-DPD uptake ( Table 3). The SUVmax or SMaT of cardiac/(bone/soft tissue) distinguished between grades 0/1 and 2/3 but not between grades 2 and 3. Higher cardiac SUVs correlating with increasing visual grade were previously described [15][16][17][18][19][20][21]. Large variations in SUVs were measured in current study for patients with both grade 2 and 3, whereas Scully et al. described similar findings only in patients with grade 2 cardiac uptake [18].
Results of quantitative assessment of extracardiac activity in present study differ from previous studies. The decrease in liver uptake with increasing cardiac activity reported here has not been previously demonstrated. The decrease in bone uptake is in agreement with some authors [18], but not all [20]. We have demonstrated that normalization of cardiac uptake to liver activity enables separation between grades 2 and 3, further enhanced by the use of the cardiac/(liver/lung) ratio. Cardiac/(bone/ soft tissue) ratio previously reported to distinguish between grades 2 and 3 [18] was not confirmed in our study group. We were the first to define quantitative cutoff values to separate between grades 2/3 compared to grades 0/1 and between grade 0 and grades 1/2/3 myocardial uptake with a FPR of 0% and TPR, PPV and NPV of 100%.
High reproducibility is crucial for robust and accurate quantification, enabling precise assessment and follow up in a clinical setting. All quantitative parameters of cardiac activity were highly reproducible with excellent interand intra-observer agreement. High inter- [18] and intraobserver [18,21] correlations were reported in certain previous patient subgroups.
Feasibility of quantitative SPECT/CT with Tc-DPD/ HDP/PYP in patients with suspected ATTR was recently assessed using CZT-based SPECT/CT scanners [19][20][21] with overall similar results to conventional SPECT/CT. Bone activity showed no significant differences between the scores, whereas lung and soft tissue activity was significantly higher in grade 2 and 3 [20]. Cardiac amyloid activity (CAA), defined as cardiac SUVmean*LV volume was proposed [21]. While quantitation results are similar, CZT-based SPECT/CT may be justified by reduced radiation exposure, improved throughput and patient comfort. With increasing availability in future, CZT-based devices may play an important role for repeat studies to monitor response to therapy. This also justifies the need for standardization of quantitative analysis.
Present study has several limitations. Although a dualcenter study the number of patients included in the analysis is limited, with only two grade 1 cases. Also, due to the retrospective design and reflecting clinical practice, patients did not have routinely endomyocardial biopsies. As mentioned, the Master VOI of the left ventricle based on non-enhanced CT images includes also the left ventricular lumen. This may lead to inclusion of blood pool uptake when projecting the master VOI on the SPECT images. SMaT enables to exclude voxels containing low SUV values, and may therefore overcome this limitation.
New knowledge gained: SPECT/CT quantitation of 99m Tc-DPD cardiac uptake is robust and separates grades 2 and 3 from grades 0 and 1 ATTR. Quantitative normalization to liver and lung, as well as specific cut-off values enables separation between all grades of cardiac uptake.
In patients with advanced ATTR visual and semi-quantitative heart-to-contralateral lung ratio on repeat Tc-PYP planar scans showed no significant change, despite clinical progression of ATTR [23]. Tc-DPD SPECT/CT quantification with reproducible measurements of amyloid burden may be of value for monitoring response to the newly available therapies and should be evaluated in future in large patient cohorts in order to establish normal values to be further used to define prognosis and to tailor and monitor response to treatment.

Conclusion
Quantitative SPECT/CT parameters of cardiac Tc-DPD uptake are robust, separating grades 2 and 3 myocardial uptake from grades 0 and 1. Normalization of cardiac SUVmax or SMaT to liver SUVmean and liver/ lung SUVmean ratio and specific cut-off values further separate between all grades. Evaluation in larger patient cohorts is needed to determine the added value of SPECT/CT quantification over visual assessment in diagnosis of borderline cases, in follow-up of patients with suspected ATTR, and in monitoring response to new available therapies.