Utility of FDG PET and Cardiac MRI in Diagnosis and Monitoring of Immunosuppressive Treatment in Cardiac Sarcoidosis

Abstract

Purpose

To compare the contributions of cardiac MRI and PET in the diagnosis and management of cardiac sarcoidosis (CS), with particular reference to quantitative measures.

Materials and Methods

This is a retrospective, observational study of 31 patients (mean age, 45.7 years) with proven extracardiac sarcoidosis and possible CS who were investigated with fluorine 18 fluorodeoxyglucose (FDG) PET/CT and cardiac MRI. Patients were treated at physicians’ discretion with repeat combined imaging after an interval of 102–770 days (median, 228 days).

Results

Significant myocardial FDG uptake was shown on visit 1 (myocardial maximum standardized uptake value [SUV_max] > 3.6) in 17 of 22 patients who were subsequently treated. Myocardial SUV_max decreased at follow-up (6.5 to 4.0; P < .01) and was matched by significant decreases in FDG-avid lung and mediastinal node disease. A volumetric measure of myocardium above a threshold SUV (cardiac metabolic volume) decreased from a mean of 42.5 to a mean of 4.1 (P < .001). This was associated with significant improvement in the left ventricular ejection fraction (LVEF) (45.8 increasing to 50.9; P < .031). There was no change in volume of late gadolinium enhancement at treatment. Patients who were untreated showed no change in any FDG PET or cardiac MRI parameter.

Conclusion

Myocardial FDG uptake in patients suspected of having CS is presumed to represent active inflammation. When treated with corticosteroids, this resolved or regressed at follow-up, with an improvement in LVEF and FDG-avid thoracic disease. Patients who were untreated showed no change in any parameter. Quantification of FDG-avid myocardium using cardiac metabolic volume is proposed as a useful objective measure for assessing response to therapy.

© RSNA, 2020

See also commentary by Gutberlet in this issue.

Summary

Diagnosis of cardiac sarcoidosis benefits from the combined use of cardiac MRI and fluorine 18 fluorodeoxyglucose (FDG) PET; at follow-up imaging, response to treatment was best assessed using quantitative FDG PET, and late enhancement at cardiac MRI did not regress.

Key Points

• ■ Fluorine 18 fluorodeoxyglucose (FDG)–avid myocardial volume decreased significantly in patients treated for cardiac sarcoidosis (CS) (44 mL to 4 mL; P < .001); treatment was associated with improved left ventricular ejection fraction at cardiac MRI (45.8 increasing to 50.9; P < .031) but was associated with no change in volume of myocardial late enhancement.

• ■ There was no change in any FDG PET or cardiac MRI parameter in the untreated group of patients presumed to have CS.

Introduction

Symptomatic cardiac sarcoidosis (CS) is suspected in only 3%–5% of patients with sarcoidosis but is found at autopsy in 20%–30% of patients with sarcoidosis (1). Deaths due to CS represent 13%–25% of all sarcoidosis-related deaths in the United States (2,3) and represent up to 85% of all sarcoidosis-related deaths in Japan (3,4). As endomyocardial biopsy has a poor sensitivity for confirming CS (5), diagnosis is typically made using clinical criteria. The most referenced criteria come from the Japanese Ministry of Health and Welfare (JMHW) (6,7) and the Heart Rhythm Society (HRS) (8). Both include imaging findings from fluorine 18 fluorodeoxyglucose (FDG) PET and cardiac MRI; however, each examines a different aspect of the disease process. Late gadolinium enhancement (LGE) on cardiac MR images is believed to represent myocardial fibrosis or scarring (9), whereas abnormal myocardial FDG uptake on PET images represents active inflammation (10). The diagnosis of CS might therefore be enhanced when these techniques are combined. Additionally, authors generally describe the presence or absence of myocardial FDG uptake at PET without formal quantification of extent. The goal of this study was to compare the contributions of cardiac MRI and PET in the diagnosis and management of CS with particular reference to quantitative measures and clinical diagnostic criteria. Although not a primary objective, we also compared the impact of combined imaging on the diagnosis of CS using clinical criteria.

Materials and Methods

Study Population

This was a retrospective, observational study of patients suspected of having CS referred for imaging between August 2012 and May 2018. Patients were identified from our institutional CS imaging registry, which has approval from the institutional review board. The need for informed consent was waived, and a full chart review was not required. All patients had evidence of extracardiac sarcoidosis, either from biopsy results or from classic lung CT appearances. No patient had been shown to have CS at endomyocardial biopsy.

Eligibility Criteria

Patients were included if they had attended for combined FDG PET/CT and cardiac MRI on two occasions and had not been treated with oral steroids or other immunosuppressants for at least 6 months before their first attendance. FDG PET/CT and cardiac MRI were performed on the same day wherever possible. Patients with a delay between PET/CT and cardiac MRI greater than 8 weeks on either visit were excluded. Patients in whom there was a change in immunosuppressive treatment status between PET/CT and cardiac MRI examinations (at either time point) were also excluded. Finally, patients who had or were suspected of having ischemic heart disease, valvular heart disease, insulin-dependent diabetes, or other inflammatory myocardial diseases were excluded to prevent false-positive myocardial FDG uptake. Of 90 patients in the registry eligible on the basis of two combined FDG PET and cardiac MRI examinations and a definite diagnosis of extracardiac sarcoidosis, 58 were excluded on the basis of the above criteria, with one being excluded for noncompliance with diet (Fig 1).

Figure 1:

Flowchart details selection criteria for the study cohort and fluorine 18 fluorodeoxyglucose (FDG) PET and cardiac MRI late gadolinium enhancement (LGE) results in treated and untreated groups at initial and follow-up studies. CMR = cardiac MRI.

All patients were required to follow a strict low-carbohydrate diet the day before the examination (<3 g of carbohydrate over 24 hours) in combination with an overnight fast (11). Those who failed to adhere to either the diet or the fast had the PET component canceled and rebooked within 8 weeks. Patients with type 2 diabetes taking oral hypoglycemics stopped their hypoglycemic medication on the day of the low-carbohydrate diet and on the day of the examination and had to be diet and overnight-fast compliant.

All cases were reviewed independently by a cardiothoracic radiologist (R.A.C.) and an imaging cardiologist (A.M.C.) with expertise in PET and MRI (R.A.C., 30 years of experience in cardiac MRI and 18 years of experience in cardiac PET; A.M.C., 18 years of experience in cardiac MRI and 17 years of experience in cardiac PET).The clinical diagnosis of CS was based on criteria from the JMHW (7) and the HRS (8).

FDG PET/CT Imaging

Patient blood sugar levels were measured before injection of FDG, and a serum glucose level below 7 mmol/L was required. Radiotracer was given according to body weight (5.18 MBq/kg), with all PET/CT examinations being performed an hour later (Gemini TF; Philips, Eindhoven, the Netherlands). CT images were acquired from the neck to mid-abdomen during quiet breathing (0.5 second per rotation, 100-mAs tube current, and 120-kVp tube voltage). PET images were acquired with three to four bed positions (50% overlap) covering the same range, with 1 minute per bed position. Both data sets were reconstructed in 3-mm-thick slices with 50% overlap. PET data were acquired in three-dimensional mode and were iteratively reconstructed using CT-based attenuation correction.

Cardiac MRI

All examinations were performed with a 1.5-T scanner (Aera; Siemens Medical Systems, Erlangen, Germany) with a 32-channel phased-array coil. Steady-state free precession (SSFP) cine images were acquired in three long-axis planes (8-mm slice thickness; 25 phases; retrospective gating). T2-weighted short-tau inversion recovery images (8-mm slice thickness) were acquired at three short-axis locations (base apical, mid-apical, and periapical).

Intravenous gadolinium-based contrast material was then administered (0.2 mmol per kilogram of body weight gadopentetate dimeglumine), and short-axis cine SSFP imaging (8-mm thickness; 25% gap) was performed through the ventricles for assessment of biventricular ventricular function. At 10 minutes following contrast material administration, LGE imaging was performed in short and long axes using a phase-sensitive inversion recovery sequence, matching the cine SSFP slice thickness and positions. Inversion time was determined on an individual patient basis to obtain optimal nulling of the unenhanced myocardial signal (typically 270–330 msec).

FDG PET/CT Analysis

All PET/CT data were transferred to an offline workstation for analysis using three-dimensional fusion software (Oasis; Segami, Columbia, Md).

Active extracardiac sarcoid involvement was considered present if there was abnormal intra- or extrathoracic nodal-tracer uptake, abnormal pulmonary parenchymal-tracer uptake, or abnormal tracer activity in the liver or spleen. Maximum standardized uptake value (SUV_max) thresholds were used to define abnormality as per Wang et al (12) (for nodes, an SUV_max > 2.5 and/or >1.0 higher than blood-pool SUV_max; for pulmonary nodules or nodular consolidation, an SUV_max > 1.5 and/or >0.5 higher than background normal lung; and for liver or spleen, an SUV_max > 5.0).

The SUV_max of the mediastinal blood pool was estimated using a 30-mm spherical region of interest in the right atrium. The threshold for identifying abnormal myocardial FDG uptake was based on a published historical population of oncology patients with no evidence of cardiac disease, who had undergone the same imaging protocol on the same imaging system with the same dietary and fasting preparation. Myocardial SUV_max in this diet-prepared group was 1.8 ± 0.9 standard deviation (SD), giving a threshold for abnormal myocardium in this study population of SUV_max greater than 3.6 (mean + 2 SDs) (11). This was further characterized as intense if SUV_max was greater than 4.5 (mean +3 SDs). Any patient found to have diffuse intense myocardial uptake was excluded because of presumed failure of suppression of normal physiologic glucose metabolism (13).

The volume of abnormal myocardial FDG uptake was quantified using dedicated software (Metavol; Hokkaido University, Sapporo, Japan). Cardiac metabolic volume was calculated using a threshold SUV_max greater than 3.6. Having identified sites of abnormal cardiac activity, the distribution of FDG-avid myocardium uptake was mapped to a modified American Heart Association model for left ventricular (LV) segmentation designed to include the right ventricle (14). All PET measurements for both examinations were performed by the same observer (A.M.C.). To assess intraobserver variability in the measurement of cardiac metabolic volume, measurements for the initial FDG PET/CT examination were repeated.

Cardiac MRI Analysis

All cardiac MRI data were transferred to an offline workstation for visual assessment and analysis (cvi42; Circle Cardiovascular Imaging, Calgary, Canada). LV and right ventricular volumes and ejection fractions were derived using manual planimetry of short-axis SSFP slices (15). The presence of myocardial edema was assessed visually from T2-weighted short-tau inversion recovery images and postcontrast cine SSFP images, and myocardial LGE was quantified from the magnitude images of the short-axis phase-sensitive inversion recovery stack (16,17). A region of interest drawn in remote normal myocardium on each magnitude slice was used to define the signal threshold for calculation of LGE volume (mean of normal myocardium + 5 SDs) (18). As LGE in the right ventricle was also included in the volume calculation, LGE volume is expressed in milliliters rather than as a percentage of the LV myocardium, as in other publications (19). As with abnormal myocardial FDG uptake, the distribution of LGE was mapped to the model of ventricular segmentation. All measurements for both examinations were performed by the same observer (R.A.C.). To assess intraobserver variability in the assessment of LGE volume, this component of the analysis was repeated for all initial cardiac MRI examinations.

Statistical Analysis

Categorical data are reported as frequencies and percentages. Continuous data were assessed for normal distribution using the Shapiro-Wilk test and expressed as mean ± SD. Paired Student t tests were used for paired comparisons of normally distributed data (ventricular function and volumes). SUV_max values for blood pool, myocardium, and so forth, and LGE and cardiac metabolic volumes were not normally distributed. The Wilcoxon signed rank test was used to compare these paired data sets. Values, however, are given as mean ± SD, as median values on visit 2 were often zero. Unpaired data (SUV_max values, LGE, and cardiac metabolic volumes between treated and untreated groups at initial PET/CT and cardiac MRI) were not normally distributed and were compared using the Mann-Whitney U test. Intraobserver variability in measurement of cardiac metabolic volume and LGE volumes were assessed by Bland-Altman analysis and are expressed as the bias ± 2 SDs for limits of agreement. A P value less than .05 was considered to indicate a significant difference. Statistical analyses were performed using MedCalc, version 12.1.3 (MedCalc Software, Mariakerke, Belgium).

Results

Thirty-one patients (24 men, seven women; mean age of men, 46 years ± 14.5; mean age of women, 46 years ± 15.4) underwent cardiac MRI and FDG PET on two occasions separated by a median of 228 days. FDG PET was performed in the morning, and cardiac MRI was performed in the afternoon. Both examinations were completed the same day in 50 of 62 combined studies. Intervals between PET and cardiac MRI for the remaining 12 attendances were less than 1 week (four patients), 1–4 weeks (two patients), and 5–8 weeks (six patients). None of these patients had a change in immunosuppressive therapy between undergoing PET and cardiac MRI. Patient demographics at first attendance are summarized in Table 1.

Table 1:

Patient Characteristics at Baseline

Cardiac Findings at FDG PET and Cardiac MRI: Visit 1

Abnormal myocardial FDG uptake (SUV_max > 3.6) was seen in 21 patients; 14 of these were referred for imaging on the basis of electrocardiographic (ECG) and echocardiographic criteria, and seven were referred for LGE results suspicious for sarcoidosis at a historical cardiac MRI. In 18 of these patients, myocardial SUV_max was greater than 4.5 (ie, focal intense), and in the three in whom this threshold was not met (ie, focal), LV function was impaired. In one of these, the site of myocardial FDG PET uptake was also matched by LGE.

At T2-weighted imaging and postcontrast cine SSFP, myocardial edema and/or hyperemia was seen in four of 31 cases (13%). In all four, this was matched by LGE and focal intense myocardial FDG uptake. Seventeen patients suspected of having active CS by myocardial FDG uptake had no edema at MRI. Nineteen patients showed a nonischemic pattern of myocardial enhancement (61%). Seven of these had basal septal thinning, and 15 showed LV impairment (LV ejection fraction [LVEF] < 50% using JMWH criteria). Twelve showed more severe LV impairment (LVEF < 40% using HRS criteria).

Intraobserver variability for measurement of LGE volume was low (bias, 0.7 mL ± 3.2) and was similar to previous published values (20). Intraobserver variability for measurement of cardiac metabolic volume was extremely low (bias, 0.0 mL ± 0.7).

Extracardiac Findings at FDG PET and Cardiac MRI: Visit 1

Abnormal extracardiac FDG avidity was seen in 29 of 31 patients (94%); it was seen in the lung parenchyma in 25 (81%), in thoracic nodal groups in 25 (81%), and in an extrathoracic location (cervical, axillary, or abdominal nodes, the spleen, the liver, or a combination) in 18 (58%). Two patients showed abnormal myocardial FDG uptake in the absence of sarcoid-related tracer uptake elsewhere. Both were associated with LGE on cardiac MR images and both had a history of remote treatment with corticosteroids for suspected CS on the basis of nonischemic LGE at a prior cardiac MRI.

Concordance between Cardiac MRI LGE and Myocardial FDG for Presence or Absence of Cardiac Involvement: Visit 1

There were 15 of 31 cases in which both LGE and abnormal myocardial FDG uptake were present and six in which there was neither (concordance 68%). In six patients, there was abnormal myocardial FDG uptake with no detectable LGE, and in four there was LGE without abnormal myocardial FDG (Figs 1, 2). Cardiac MRI and PET results were discordant in 32%. When FDG PET/CT and cardiac MRI findings were combined with ECG abnormalities, 14 patients (45%) met modified JMHW criteria for the diagnosis of CS and 20 (64%) met HRS criteria. Eight patients met neither (Table 2). Three JMHW-positive patients were HRS-negative because of the requirement for biopsy confirmation of extracardiac sarcoidosis by HRS criteria. JMHW criteria require either biopsy or clinical imaging, in keeping with extracardiac disease. Had cardiac MRI been used alone, only eight cases would have met JMHW criteria, and 20 would have met HRS criteria. Using FDG PET alone, only one case would have been JMHW-positive, and 17 cases would have been HRS-positive.

Figure 2a:

Images in a 50-year-old man with stage 2 sarcoidosis for 12 months develops resting tachycardia (100+ beats per minute). There is no myocardial tracer uptake on the (a) two-chamber fused PET/CT image, despite (b) multifocal nonischemic late gadolinium enhancement (LGE) in the anterior and inferior wall on the cardiac MR image (black arrowheads). The patient is not treated with corticosteroids and returns 6 months later. (c) There is still no myocardial fluorine 18 fluorodeoxyglucose uptake, although LGE is a little more prominent. There was no change in biventricular function or volumes (patient 7, Table 2).

Table 2:

FDG PET/CT and Cardiac MRI Findings on Visit 1 and Correlation with JMHW and HRS Criteria

Figure 2b:

Images in a 50-year-old man with stage 2 sarcoidosis for 12 months develops resting tachycardia (100+ beats per minute). There is no myocardial tracer uptake on the (a) two-chamber fused PET/CT image, despite (b) multifocal nonischemic late gadolinium enhancement (LGE) in the anterior and inferior wall on the cardiac MR image (black arrowheads). The patient is not treated with corticosteroids and returns 6 months later. (c) There is still no myocardial fluorine 18 fluorodeoxyglucose uptake, although LGE is a little more prominent. There was no change in biventricular function or volumes (patient 7, Table 2).

Figure 2c:

Images in a 50-year-old man with stage 2 sarcoidosis for 12 months develops resting tachycardia (100+ beats per minute). There is no myocardial tracer uptake on the (a) two-chamber fused PET/CT image, despite (b) multifocal nonischemic late gadolinium enhancement (LGE) in the anterior and inferior wall on the cardiac MR image (black arrowheads). The patient is not treated with corticosteroids and returns 6 months later. (c) There is still no myocardial fluorine 18 fluorodeoxyglucose uptake, although LGE is a little more prominent. There was no change in biventricular function or volumes (patient 7, Table 2).

Findings at FDG PET and Cardiac MRI: Visit 2

Thirty-one follow-up cardiac MRI and PET studies were performed 102–770 days (median, 228 days) after the initial combined study. Twenty-two of 31 patients (68%) were treated with oral prednisone between visits, seven having concomitant treatment with methotrexate or mycophenolate. Although there was no standard regimen for immunosuppression, all patients were initiated with a moderate-to-high dose of prednisone (30–50 mg per day), tapering to a maintenance dose of 5–10 mg per day. Treatment duration varied from 3 to 15 months, with all patients being re-examined while still taking immunosuppressants or within 3 months of stopping. In the two instances in which the interval between visits exceeded 12 months, one was for a patient not receiving treatment (770 days) and the other was for a patient who had been taking steroids for 15 months (497 days).

As an observational study, the decision to treat was made by the referring physician and the patient. Despite five patients meeting JMHW and/or HRS clinical criteria for CS, they were not treated. Three of these had active cardiac CS as determined with FDG PET. The reason for not treating is unknown. Four patients who met neither set of criteria were treated.

There were significant reductions in myocardial SUV_max (6.5 ± 3.1, decreasing to 4.0 ± 2.5; P < .01) and cardiac metabolic volume (42.5 mL ± 65.1, falling to 4.1 mL ± 10.5; P < .001). The decrease in abnormal cardiac metabolic activity was matched by an improvement in LVEF (45.8 ± 12.7, increasing to 50.9 ± 8.7; P < .031) and a reduction in LV end-systolic volume. Right ventricular end-systolic volume also decreased significantly with a trend toward improved right ventricular ejection fraction. The fall in myocardial SUV_max was accompanied by a significant reduction in FDG-avid lung and mediastinal node disease (Fig 3). Despite the improvement in LVEF in the treated group, there was no change in the volume of LGE (Table 3). Unfortunately, neither initial myocardial SUV_max nor the size of the decrease in cardiac metabolic volume predicted improvement in LVEF. Whether this lack of correlation is genuine or has been masked by relatively short follow-up interval remains unclear.

Figure 3a:

Images in a 35-year-old man with third-degree heart block and stage 1 sarcoidosis. (a) Fluorine 18 fluorodeoxyglucose (FDG)–avid myocardium (maximum standardized uptake value [SUV_max] = 7.6) in the inferior wall of the left ventricle on a two-chamber fused PET/CT image (black arrowheads). (b) This matches late gadolinium enhancement (LGE) on cardiac MR image but is more extensive (white arrowheads). (a) A large FDG-avid subcarinal lymph node (SUV_max = 6.1) (black arrow). (a, c) At follow-up after 6 months of oral corticosteroids, myocardial FDG uptake has resolved, as has uptake in the large subcarinal node (black arrow). (b, d) There is little change in LGE. Left ventricular function is mildly impaired on both studies. (b, d) A large subcarinal lymph node (white arrow) has resolved at MRI (patient 14, Table 2).

Table 3:

Changes between Visits for Ventricular Parameters, SUV_max, and LGE for Treated and Untreated Groups

Figure 3b:

Images in a 35-year-old man with third-degree heart block and stage 1 sarcoidosis. (a) Fluorine 18 fluorodeoxyglucose (FDG)–avid myocardium (maximum standardized uptake value [SUV_max] = 7.6) in the inferior wall of the left ventricle on a two-chamber fused PET/CT image (black arrowheads). (b) This matches late gadolinium enhancement (LGE) on cardiac MR image but is more extensive (white arrowheads). (a) A large FDG-avid subcarinal lymph node (SUV_max = 6.1) (black arrow). (a, c) At follow-up after 6 months of oral corticosteroids, myocardial FDG uptake has resolved, as has uptake in the large subcarinal node (black arrow). (b, d) There is little change in LGE. Left ventricular function is mildly impaired on both studies. (b, d) A large subcarinal lymph node (white arrow) has resolved at MRI (patient 14, Table 2).

Figure 3c:

Images in a 35-year-old man with third-degree heart block and stage 1 sarcoidosis. (a) Fluorine 18 fluorodeoxyglucose (FDG)–avid myocardium (maximum standardized uptake value [SUV_max] = 7.6) in the inferior wall of the left ventricle on a two-chamber fused PET/CT image (black arrowheads). (b) This matches late gadolinium enhancement (LGE) on cardiac MR image but is more extensive (white arrowheads). (a) A large FDG-avid subcarinal lymph node (SUV_max = 6.1) (black arrow). (a, c) At follow-up after 6 months of oral corticosteroids, myocardial FDG uptake has resolved, as has uptake in the large subcarinal node (black arrow). (b, d) There is little change in LGE. Left ventricular function is mildly impaired on both studies. (b, d) A large subcarinal lymph node (white arrow) has resolved at MRI (patient 14, Table 2).

Figure 3d:

Images in a 35-year-old man with third-degree heart block and stage 1 sarcoidosis. (a) Fluorine 18 fluorodeoxyglucose (FDG)–avid myocardium (maximum standardized uptake value [SUV_max] = 7.6) in the inferior wall of the left ventricle on a two-chamber fused PET/CT image (black arrowheads). (b) This matches late gadolinium enhancement (LGE) on cardiac MR image but is more extensive (white arrowheads). (a) A large FDG-avid subcarinal lymph node (SUV_max = 6.1) (black arrow). (a, c) At follow-up after 6 months of oral corticosteroids, myocardial FDG uptake has resolved, as has uptake in the large subcarinal node (black arrow). (b, d) There is little change in LGE. Left ventricular function is mildly impaired on both studies. (b, d) A large subcarinal lymph node (white arrow) has resolved at MRI (patient 14, Table 2).

Comparing treated and nontreated groups at visit 1, there was no statistical difference in SUV_max values in mediastinal node or lung disease. Myocardial SUV_max, however, was significantly lower in the nontreated group (P = .03), presumably reflecting less severe cardiac involvement. Cardiac metabolic volume and LGE volume in the nontreated group were also lower, but this did not reach statistical significance (P = .095 and P = .217, respectively). In the untreated group, there was no reduction in myocardial SUV_max, cardiac metabolic volume, or volume of LGE. There was also no change in ventricular function or volumes.

Figure 4 shows the distribution of abnormal myocardial FDG uptake and LGE on initial and follow-up visits. This confirms previous reports that CS has a predilection for the proximal anterior wall and proximal septum and the proximal and mid-inferior and mid-inferolateral walls (21). The LV apex and periapical region were virtually spared. Although sites of abnormal myocardial FDG uptake were frequently matched by LGE on the initial study, FDG uptake was considerably more extensive. Unmatched FDG uptake was particularly common in the inferior and inferolateral wall of the LV, sites often believed to be nonsuppressed physiologic myocardial uptake. Regression or resolution of these sites with treatment, matched by improvement in extrathoracic disease and increased LVEF, suggests this genuinely represents disease (Fig 5). In the untreated group, inferior and inferolateral FDG uptake was unchanged or increased between visits. When present, LGE was frequently patchy and at the mid wall as described in CS, but there was no definitive pattern. During the follow-up period of this study, LGE did not resolve.

Figure 4a:

Plots of distribution of abnormal myocardial fluorine 18 fluorodeoxyglucose (FDG) uptake and late gadolinium enhancement (LGE) on initial and follow-up studies in the (a) treatment and (b) nontreatment groups. The color of the dots represents FDG uptake alone, matched FDG and LGE, and LGE alone in each segment. The size of the dot indicates the number of patients with that abnormality. In the treatment group, matched FDG and LGE on visit 1 becomes unmatched LGE on visit 2 in the majority of cases. There is little change in the untreated group.

Figure 5a:

Images in a 54-year-old man with stage 2 sarcoidosis for 2 years develops new right-bundle-branch block shown on an electrocardiogram. Myocardial fluorine 18 fluorodeoxyglucose (FDG) uptake (maximum standardized uptake value [SUV_max] = 4.6) without matching late gadolinium enhancement (LGE) is shown on the (a) four-chamber fused PET/CT image (black arrowheads) and (b) cardiac MR image. With normal left ventricular (LV) function, this patient does not meet Japanese Ministry of Health and Welfare criteria but does meet Heart Rhythm Society criteria. He is treated with corticosteroids for 6 months. Follow-up PET/CT and cardiac MR images show resolution of cardiac and extracardiac FDG uptake at all sites. LGE does not develop, and LV function is unchanged. Pulmonary parenchymal disease also partly regresses at MRI (patient 24, Table 2).

Figure 4b:

Plots of distribution of abnormal myocardial fluorine 18 fluorodeoxyglucose (FDG) uptake and late gadolinium enhancement (LGE) on initial and follow-up studies in the (a) treatment and (b) nontreatment groups. The color of the dots represents FDG uptake alone, matched FDG and LGE, and LGE alone in each segment. The size of the dot indicates the number of patients with that abnormality. In the treatment group, matched FDG and LGE on visit 1 becomes unmatched LGE on visit 2 in the majority of cases. There is little change in the untreated group.

Figure 5b:

Images in a 54-year-old man with stage 2 sarcoidosis for 2 years develops new right-bundle-branch block shown on an electrocardiogram. Myocardial fluorine 18 fluorodeoxyglucose (FDG) uptake (maximum standardized uptake value [SUV_max] = 4.6) without matching late gadolinium enhancement (LGE) is shown on the (a) four-chamber fused PET/CT image (black arrowheads) and (b) cardiac MR image. With normal left ventricular (LV) function, this patient does not meet Japanese Ministry of Health and Welfare criteria but does meet Heart Rhythm Society criteria. He is treated with corticosteroids for 6 months. Follow-up PET/CT and cardiac MR images show resolution of cardiac and extracardiac FDG uptake at all sites. LGE does not develop, and LV function is unchanged. Pulmonary parenchymal disease also partly regresses at MRI (patient 24, Table 2).

Discussion

Our study, using both modalities nearly simultaneously, has shown myocardial FDG to be a more sensitive marker for active myocardial disease than the presence of either edema at T2-weighted imaging or LGE. When present, myocardial FDG uptake virtually always regressed or resolved with treatment, but neither the intensity of uptake nor the metabolic volume were good predictors of improvement in LVEF. Coincident regions of FDG uptake and LGE were common but were not the rule; choosing to perform either PET or cardiac MRI alone could underestimate the extent of disease and the need for treatment. In the treatment group, LVEF improved by more than 5% in eight patients. Although five of these demonstrated a combined pattern of LGE and myocardial FDG uptake, one had myocardial FDG uptake alone, and two had LGE alone. In both LGE cases, there was focal myocardial FDG uptake just below our threshold for abnormality (SUV_max > 3.6).

The use of JMHW or HRS criteria is equally problematic in guiding treatment. Five patients meeting one or other criteria were not treated, with only one of these deteriorating at follow-up. Eighteen of 22 patients in the treatment group met clinical criteria, but only seven of these showed improvement in LVEF. This mirrors the experience of others using the same criteria and underlines the need for better reference standards (22).

In this study, response to therapy was more readily assessed with PET than with cardiac MRI. Both myocardial SUV_max values and cardiac metabolic volume decreased with treatment, and this was frequently accompanied by matched decreases in metabolic activity in nodal and lung disease. The most sensitive indicator of the myocardial response to treatment was cardiac metabolic volume. Although previous investigators have suggested that cardiac MRI can be used for monitoring the treatment response in CS (23), we have found it unhelpful. Myocardial edema on T2-weighted images was uncommon and, where present, was always matched by delayed enhancement (four of 19 patients with LGE [21%]). This is similar to previously published data in which five of 34 patients with LGE showed edema on T2-weighted images (24). LGE in CS is generally considered to be a marker of scarring or fibrosis, and, as anticipated, the volume of LGE did not change at follow-up.

Three patterns of myocardial involvement were seen in this cohort: myocardial FDG uptake alone, matched myocardial FDG uptake and LGE, and LGE alone. The pathologic basis for these patterns remains to be confirmed, but if CS in the myocardium is similar to that described in the lung, these imaging patterns can be explained (25). The earliest phase of CS is believed to be an inflammatory myocarditis, indistinguishable from a lymphocytic myocarditis (21). This might correspond to unmatched myocardial FDG uptake. In the lung, this can resolve spontaneously, but it is not known whether this also occurs in the heart. If so, it may not always need to be treated. As others have found (26), the most common appearance in this study was matched myocardial FDG uptake and LGE, representing the active granulomatous phase. The late fibrotic phase, well recognized in sarcoid lung disease, contains few inflammatory cells and is presumed responsible for unmatched myocardial LGE. In the absence of FDG-avid myocardium, response to immunosuppression is theoretically less likely.

For FDG PET to be of value in identifying CS, physiologic myocardial FDG uptake must be suppressed. There is no consensus as to the best technique for doing this, but most investigators advocate a preparatory low-carbohydrate diet and overnight fast (27,28). There is no agreement that failure to comply with the diet should lead to cancellation of the study. Published failure rates for myocardial suppression are high (29). In this study, patients found to be noncompliant with strict dietary instructions (as judged by food questionnaire responses on the day of imaging) had their examinations canceled and rebooked. We canceled and rebooked six of 31 patient examinations (19%) and had one failed suppression that was excluded as per Figure 1. Had we not canceled and rebooked, our rate of failed myocardial suppression may have been similar to those of published series.

Limitations

This is a single-center, retrospective observational study based on registry-collected data. To identify a coherent patient group with stringent inclusion criteria, a large number of cases had to be excluded. Despite resulting small numbers, our cohort is similar in size to those of other published series in which only FDG PET or cardiac MRI were used (23,30,31), and the impact of treatment on cardiac and extracardiac FDG uptake and LV remodeling and function was significant.

We did not perform a resting nuclear perfusion examination as part of our study protocol. The rationale was that cardiac MRI can be used to accurately identify abnormalities that cause resting defects on nuclear perfusion (ie, hibernating myocardium, myocardial infiltration, and scarring [LGE]) (32–34) and, as such, could replace the resting nuclear perfusion scan. In a previous study comparing LGE with resting nuclear perfusion in CS (35), a third of segments showing nonischemic LGE had normal resting perfusion. In all segments without LGE, resting perfusion was also normal. More recently, the same has been shown in inflammatory cardiomyopathies in which nuclear perfusion was compared with PET/CT or PET/MRI (33). Although there remains a risk that focal FDG uptake in resting ischemia might be misinterpreted as CS, in this cohort, patients who had or were suspected of having ischemia were excluded.

Although using an SUV_max threshold discriminates between abnormal and physiologic myocardial tracer uptake in the majority of cases, there is a risk of missing low-grade active disease. The overlap between physiologic and low-grade pathologic uptake means there is no threshold that can be used to identify all patients with active CS. Whether missing low-grade CS is important or not remains unclear.

Our cardiac MRI study protocol did not include native T1 mapping. This has recently been shown as a way to discriminate between patients with CS and control patients without CS and has potential for assessing disease activity (36). This sequence, however, was not available to us in the early years of data collection.

This study combined concurrent cardiac MRI and PET/CT for the diagnosis and management of CS. Logistically, this was complex, and, at times, one component or the other had to be canceled and rebooked. These difficulties are eliminated using PET/MRI, and there are theoretical advantages in inflammatory cardiomyopathies (26,33,37). Access to PET/MRI, however, is even more restricted than having PET/CT and cardiac MRI in the same institution. It has also yet to be shown that this technique has significant advantages over same-day examinations performed as we describe here.

Conclusion

CS has important prognostic implications for patients with established extracardiac sarcoidosis, and yet it may be difficult to detect using any single imaging modality. We have found same-day FDG PET/CT and cardiac MRI to be both practical and effective in the diagnosis and monitoring of CS. Using either technique alone creates a risk of missing treatable cardiac disease or treating disease that is inactive. For monitoring patients with CS, we have found cardiac metabolic volume at FDG PET to be the most useful parameter for guiding therapy.