Riina Kandolin, Division of Cardiology, Department of Medicine, Helsinki University Central Hospital, 00029 HUS, Finland. (fax: +358-9-471-74574; e-mail: email@example.com).
Abstract. Kandolin R, Lehtonen J, Graner M, Schildt J, Salmenkivi K, Kivistö SM, Kupari M (Helsinki University Central Hospital, Helsinki, Finland). Diagnosing isolated cardiac sarcoidosis. J Intern Med 2011; 270: 461–468.
Objectives. Cardiac sarcoidosis (CS) without clinically apparent extracardiac disease may escape detection because of the poor sensitivity of endomyocardial biopsy (EMB). We set out to analyse our experience of repeated and imaging-guided biopsies in clinically isolated CS.
Methods. We retrospectively reviewed the medical records, laboratory test results, imaging studies and pathological analyses of 74 patients with either histologically proven or clinically probable CS at our institution between January 2000 and December 2010.
Results. Fifty-two patients had histologically proven CS, of whom 33 (26 women) had disease that was clinically isolated to the heart. Sarcoidosis was detected in the first EMB in 10 of the 31 patients who underwent biopsy. CS was found by repeated EMBs, targeted by cardiac imaging, in seven additional patients, and 11 patients were diagnosed by sampling 18-F-fluorodeoxyglucose position emission tomography-positive mediastinal lymph nodes at mediastinoscopy. Together, the first biopsy (cardiac or mediastinal lymph node) provided the diagnosis in 34%, the second biopsy in 31% and the third in 22% of biopsied patients with isolated CS. Four (13%) of the remaining diagnosis were made after cardiac transplantation and one in a patient who did not undergo biopsy) at autopsy after sudden cardiac death.
Conclusions. Cardiac sarcoidosis may present without clinically apparent disease in other organs. At least two-thirds of patients remain undiagnosed after a single EMB session. The detection rate can be improved by repeated and imaging-guided cardiac or mediastinal lymph-node biopsies. Nevertheless, false-negative biopsy results remain a problem in CS patients with no apparent extracardiac disease.
Cardiac involvement is the most omnious manifestation of sarcoidosis. Myocardial inflammation, granuloma formation and fibrotic changes may result in conduction disturbances, sustained ventricular arrhythmias, segmental or diffuse left ventricular (LV) dysfunction, mitral regurgitation and heart failure [1–3]. These changes cause significant morbidity and may result in premature death due either to sudden arrhythmia or progressive heart failure [4–6]. Mortality from cardiac sarcoidosis (CS) is high, ranging from 10% to 40% over 5 years [4–7]. Although treatment of CS has not been evaluated in randomized trials, it is assumed that therapy with steroids and implantable defibrillators, and with transplantation in selected cases, may reduce symptoms and improve prognosis [1–4, 6]. There is also experience to suggest that starting treatment early, before deterioration of LV function, ensures the best prognosis .
Clinically, there are two main scenarios of CS presentation. The first encompasses patients with known or easily apparent sarcoidosis elsewhere in the body who develop symptoms and/or signs of cardiac disease. Given the histopathology of extracardiac sarcoidosis, typical abnormalities in 12-lead electrocardiogram (ECG) supported by findings at echocardiography, magnetic resonance imaging (MRI), myocardial perfusion scanning or positron emission tomography (PET) constitute sufficient evidence for the diagnosis of cardiac involvement . The second scenario, clinically isolated CS, includes patients admitted primarily for cardiac symptoms, usually for conduction disturbances, ventricular arrhythmias or LV dysfunction, in whom there is no clinical evidence of sarcoid involvement in other organs. This has been considered a rare presentation, but in a recent autopsy study of 25 cases up to 40% of patients dying suddenly from CS had no obvious extracardiac disease . Positive proof of CS requires diagnostic histopathology; however, the sensitivity of endomyocardial biopsy (EMB) is no better than 19–30% [6, 7, 10–12]. As the mere clinical suspicion of CS hardly justifies the long-term immunosuppression of proven disease, every effort should be made to confirm the diagnosis. Here, we report our recent experience with isolated CS focusing on the use of modern cardiac imaging to guide the diagnostic procedure.
From the year 2000 until 2010, CS with histopathological verification was diagnosed in 52 patients at our institution. Nineteen of them, nine men and 10 women aged 51.7 ± 11.4 years (mean ± SD), had known or apparent extracardiac sarcoidosis. The remaining 33 patients, seven men and 26 women aged 49.9 ± 9.3 years, had no history or evident manifestations of sarcoidosis outside the heart according to clinical examination, chest X-rays and laboratory tests. These patients were classified as having CS clinically isolated to the heart; this group included more female patients than the group with extracardiac sarcoidosis (P = 0.049). In addition, there were 22 patients with a strong clinical suspicion but no histological evidence of CS. Of the 74 patients, we focused on the 55 with either confirmed or suspected isolated CS. Their medical records, laboratory test and imaging results and pathological analyses of biopsy materials were retrospectively reviewed and analysed for the present study.
Changes in diagnostic strategy from 2000 to 2010
From 2000 until 2005, clinical examination, 12-lead ECG, laboratory tests, echocardiography and MRI without contrast agent were used to explore the aetiology of unknown myocardial disease. Selective coronary angiography and LV cineangiography were performed whenever coronary artery disease needed to be excluded. EMBs were taken if the presence of a chronic infiltrative or inflammatory myocardial disease, such as sarcoidosis, or a rapidly advancing myocarditis was considered possible. Biopsies were not usually performed for dilated cardiomyopathy if no arrhythmias were detected. They were routinely taken from the right ventricular septum.
In 2006, we revised our diagnostic strategy and have since been actively using gadolinium-enhanced cardiac MRI (Gd-MRI) and 18F-fluorodeoxyglucose positron emission tomography (18FDG-PET) combined with resting myocardial perfusion imaging to identify and localize possible infiltrative and/or inflammatory myocardial processes. Initially, 18FDG-PET scanning was focused on the heart but whole-body images have been routinely acquired since the end of 2008. Concomitant with the active use of Gd-MRI and 18FDG-PET, we changed our biopsy policy by trying to specifically sample the myocardial areas in which there were signs of damage and/or inflammation. Since then, more LV EMBs have been acquired and the number of EMB samples per session has increased. In patients with a high degree of clinical suspicion of CS despite one or two negative EMBs, we have acquired 18FDG-PET-positive mediastinal lymph nodes by mediastinoscopy for histological evaluation.
The imaging methods
All imaging studies were performed as part of routine care. For cardiac ultrasound studies, we used a Vivid 7 echocardiograph (GE Healthcare, Helsinki, Finland) Kuortaneenkatu 2, Helsinki, PL 300, 00031 GE, including two-dimensional, Doppler and M-mode imaging. Since 2005, the studies have been digitally archived and available for retrospective analyses. For the present study, we noted all abnormalities of LV end-diastolic diameter, ejection fraction and wall thickness.
MRI was performed with a 1.5 T imager (Avanto; Siemens, Erlangen, Germany) using a multichannel body-array coil as the receiver. Breath-hold cine MRI was performed using ECG-gated segmented true fast imaging with steady-state precession (true FISP). Images were obtained in vertical and horizontal long-axis planes and in the contiguous short-axis planes covering the whole left ventricle. Between 5 and 15 min after injection of a contrast agent (gadoterate meglumine, Dotarem® 0.1 mmol kg−1), late-enhancement images were acquired in the same views as for cine images, using inversion recovery turbo fast-low angle shot (FLASH) sequence.
For PET studies, patients fasted for 12 h, and blood glucose had to be <7 mmol L−1 prior to imaging. After an intravenous injection of 18FDG-PET (303 ± 57 MBq), patients rested for 60 min in a semidarkened room. Images were acquired with a Gemini PET/CT scanner (Philips, Andover, MA, USA). First, a CT surview (30 mA, 120 kVp) was obtained. Then, patients were scanned from the base of the skull to mid-thigh (50 mA, 120 kVp). PET emission imaging of the whole body was supplemented by a separate scan (35 mA, 140 kVp) of the cardiac area 1 h (16 cm frame, 10 min) and 3 h (16 cm frame, 15 min) after the injection.
Within 1 week after the PET/CT scan, 99mTc-tetrofosmin myocardial perfusion imaging at rest was performed. The data were acquired 1 h after administration of 99mTc-tetrofosmin (524 ± 56 MBq) using a double-head ADAC Forte single-photon emission computed tomography (SPECT) gamma camera (Philips) equipped with low-energy, high-resolution collimators. Myocardial perfusion was visually compared with LV 18F-FDG uptake during PET. An area with reduced perfusion and enhanced 18F-FDG accumulation was interpreted as suggestive of an active inflammatory process in the absence of coronary artery disease [13–15].
Analysis of biopsies
Between 2000 and 2010, a total of 576 patients underwent diagnostic EMBs (transplant surveillance biopsies were excluded) at our institution. In 53 patients (9.2%), the indication was a suspicion of CS. EMB was performed between 2000 and 2005 in six of these patients and between 2006 and 2010 in the remaining 47. Right ventricular biopsies were taken from the ventricular septum under fluoroscopic and echocardiographic guidance via the right internal jugular vein or, less often, via the right femoral vein. LV biopsies, performed only after 2006, were acquired through a long biopsy sheath or a suitable coronary guiding catheter introduced into the left ventricle via the femoral artery. The mean number of right ventricular samples per biopsy session increased from 5.5 during the early period (range, 4–7) to 6.2 (range, 3–10) over the later period of our study. The mean number of LV samples was 5.3 (range, 2–19). Samples of mediastinal lymph nodes were taken by thoracic surgeons at mediastinoscopy under general anaesthesia.
The histological criteria for sarcoidosis were the presence of well-formed noncaseating granulomas and multinucleated giant cells without myocardial necrosis and with only few eosinophils. Special stains, such as Ziehl-Neelsen and Gomorri, were used to rule out other causes of granulomatous inflammation.
Group differences were analysed using chi-square test or Fisher’s exact test. P-values of <0.05 were considered statistically significant. Continuous variables are presented as mean value ± SD or range. Categorical variables are presented as absolute number and percentages. All analyses were performed using SPSS-17 for Windows (SPSS Inc., Chicago, IL, USA).
Clinical presentation and cardiac imaging
Of the 33 patients with clinically isolated CS, three were diagnosed during the 6-year period from 2000 to 2005 and the remaining 30 over the next 5 years, from 2006 to 2010 (P < 0.000). Table 1 shows the modes of clinical presentation and a summary of the findings of the imaging studies. Thirty of the 33 patients (91%) were admitted acutely for either a symptomatic high-grade atrioventricular (AV) conduction block, heart failure or a ventricular tachyarrhythmia. All but two patients had echocardiographic signs of myocardial disease either at presentation or during follow-up (Table 1). Coronary angiography was performed in 23 patients. The findings were normal in all except one patient with a diagnosis of ischaemic cardiomyopathy who had significant coronary stenosis and underwent cardiac transplantation. Examination of the explanted heart, however, revealed extensive CS.
Table 1. Clinical presentation and summary of the results of imaging studies
aIn addition, ventricular extrasystole (n = 1) or unexplained syncope (n = 1).
Mode of presentation
High-grade atrioventricular conduction block
Sustained ventricular tachycardia
Resuscitation from VF
Reduced ejection fraction (<50%)
LV dilatation (EDD >55 mm in women, >60 mm in men)
(n = 16)
(n = 13)
Late myocardial Gd-enhancement
Local wall thinning or thickening
Myocardial perfusion (99m-tetrofosmin SPECT)
(n = 25)
(n = 16)
(n = 25)
(n = 16)
Focally increased myocardial uptake
Mismatch with perfusion
(n = 15)
(n = 6)
Uptake in mediastinal lymph nodes
None of the 33 patients with clinically isolated CS had signs of extracardiac sarcoidosis at physical examination or from conventional chest X-rays. Liver enzymes were also normal in all patients. However, the serum concentrations of angiotensin-converting enzyme (ACE) and lysozyme (LZM) were elevated in five of 28 patients (18%) and 10 of 23 patients (44%), respectively.
Sixteen of the 33 patients (49%) underwent Gd-MRI during the early diagnostic evaluation. MRI was omitted in the remaining 17 patients mainly because of early implantation of a pacemaker. Myocardial areas of late Gd-enhancement suggestive of infiltration or inflammation were found in 15 of 16 patients (94%) (Table 1). The areas varied in location and extension but involved the ventricular septum in all patients. Myocardial perfusion defects were found in 24 of 25 patients (96%) undergoing 99m-tetrofosmin SPECT. The 18FDG-PET studies revealed focally enhanced myocardial glucose uptake superimposable upon a perfusion defect in 20 of 25 patients (80%). The ‘hot spot’ suggestive of inflammation involved the ventricular septum in 15 of 20 patients (75%). Thirteen patients had prominent 18F-FDG uptake in the mediastinal lymph nodes, and seven patients had uptake outside the chest (lymph nodes, liver, spleen and bone marrow). An example of 18FDG-PET images is shown in Fig. 1.
Over the study period, isolated CS was strongly suspected in 22 additional patients (age 48.4 ± 14 years, 14 women) in whom diagnostic biopsies remained negative. Their mode of clinical presentation and findings at cardiac imaging are also summarized in Table 1. There were no statistical differences in age and sex distribution, mode of clinical presentation, findings at MRI/PET or serum concentrations of ACE and LZM in these 22 patients compared with those with histologically verified isolated CS. However, as Table 1 shows, LV dilatation, systolic dysfunction and septal thinning at echocardiography were less common in these 22 patients, suggesting less extensive LV myocardial disease.
Biopsies and histological confirmation of CS
Table 2 shows a summary of the diagnostic biopsies in all 54 patients in whom the possibility of CS was initially strongly considered. The cumulative yield of repeated biopsies in patients who ultimately received histological confirmation of CS is also shown. In our 54 patients, 80 EMBs were carried out without any major complications. Forty-seven EMBs were performed in 31 of the 33 patients with ultimately confirmed CS. The first EMB revealed CS in 10 of 31 patients (32%). Repeated EMBs identified seven additional cases of CS so that the cumulative yield of EMBs was 17 in 31 patients (55%). Retrieval and histological examination of 18FDG-PET-positive mediastinal lymph nodes identified sarcoidosis in 11 additional patients. An example of lymph-node histology is shown in Fig. 2. One patient had undergone first right ventricular and then LV biopsy but CS was found only at autopsy after sudden cardiac death. The remaining four patients were diagnosed during examination of the explanted heart after transplantation. One of these patients had not received any pretransplant diagnostic biopsies.
Table 2. Summary of diagnostic biopsies in 54 patients with a strong initial clinical suspicion of CS
Order, site and number of biopsies
Histopathological confirmation of CS (n)
Proportion of proven CS cases (n = 32)a
CS, cardiac sarcoidosis.
aAdditionally, one patient with CS, who had not undergone diagnostic biopsies, was diagnosed during post-transplant study of the native heart.
bAll three patients had negative right ventricular biopsies.
cThis patient had negative right and left ventricular biopsies.
1st biopsy, n = 54
Right ventricle, n = 50
Left ventricle, n = 3
Mediastinal lymph node, n = 1
2nd biopsy, n = 29
Right ventricle, n = 14
Left ventricle, n = 9
Mediastinal lymph node, n = 6
3rd biopsy, n = 9
Left ventricle, n = 4
Mediastinal lymph node, n = 5
Post-transplant study of native heart
In total, 33 EMBs (eight from the left ventricle) and one mediastinal lymph-node biopsy were carried out with negative results in the 22 patients in whom CS was strongly clinically suspected but remained only a probable diagnosis.
Treatment and outcome
The treatment included steroids in all 28 patients with a pretransplant and premortem diagnosis of sarcoidosis, azathioprine in 15 patients and either an implantable cardioverter-defibrillator (n = 16), a physiological pacemaker (n = 13) or a biventricular pacemaker (n = 3) in a total of 29 patients (88%). During a mean follow-up time of 41 (2–102) months, six of the 33 patients (18%) underwent cardiac transplantation because of terminal heart failure or recurrent uncontrollable ventricular tachycardias. Two of them died early (within 3 weeks) after transplantation, but the remaining four were doing well without a recurrence of CS at post-transplant follow-up after 23–55 months. Furthermore, one patient died of sudden cardiac death, one of terminal heart failure (with premortem diagnosis of dilated cardiomyopathy) and one of colon cancer (this patient experienced a defibrillator discharge because of ventricular tachycardia before death). Amongst the remaining 24 patients, eight (33%) experienced a defibrillator discharge because of ventricular tachycardia (n = 6) or fibrillation (n = 2).
Of the 22 patients with clinically suspected but not proven CS, eight (36%) were treated with steroids as the likelihood of CS was considered to be high enough to warrant treatment. During follow-up, major adverse cardiac events (cardiac death, cardiac transplantation, defibrillator discharge because of ventricular fibrillation or ventricular tachycardia) were less common in patients with suspected compared to biopsy-verified CS (4/22 vs. 17/33, P = 0.013).
The present observational series shows that sarcoidosis not uncommonly involves the heart without clinically apparent disease in other organs. The diagnosis of isolated cardiac involvement is difficult, however, as at least two-thirds of first EMBs are negative and laboratory abnormalities are nonspecific and practically unavailing. Yet, if the likelihood of an inflammatory cardiomyopathy remains high, pursuing the diagnosis with repeated and imaging-guided biopsies of the myocardium or mediastinal lymph nodes is worthwhile and will markedly improve the detection rate of CS. Thus, in our hands, the diagnostic yield of the second biopsy (31%) was close to the yield of the first one (34%), and even the third biopsy contributed significantly (22%) to the diagnoses of CS.
Over the study period, altogether 17 of 576 diagnostic EMBs (3%) contributed to the diagnosis of CS. This is a higher proportion than found by Felker et al. in their study of patients with initially unexplained cardiomyopathy, but the criteria for EMB sampling were different between the two studies . Notably, the number of patients identified with clinically isolated CS at our institution increased 10-fold from the first to the second half of the last decade. We believe this change resulted from better awareness of isolated cardiac involvement and from improved cardiac imaging that led, respectively, to diagnostic perseverance and better targeting of the biopsies. Because of the seriousness of CS, continuing the effort to diagnose the condition is particularly important in young and middle-aged patients presenting with unexplained ventricular tachycardia and/or high-degree AV block combined with evidence of myocardial disease.
The sensitivity of the first EMB session in exposing sarcoidosis was 32% in our patient population. This figure is likely to be an overestimate, however, because some of the 22 patients with negative EMBs despite a strong suspicion of CS probably had the disease. The observed differences in LV structure and function between our patients with proven versus suspected disease (see above) could be attributed to a more advanced disease in the former group. Earlier reports of 19–30% sensitivity for EMB [6, 7, 9–12] were based on small patient populations, and most of these studies used the diagnostic criteria of the Japanese Ministry of Health and Welfare  as a standard for the diagnosis of CS. Repeated biopsies were not performed, and cardiac MRI or 18FDG-PET was either not available or not systematically used to map the best target areas for biopsy [6, 7, 10–12].
Although focal myocardial late enhancement at Gd-MRI and increased glucose uptake at cardiac 18FDG-PET are nonspecific signs of myocardial damage or inflammation, in histologically proven extracardiac sarcoidosis they provide sensitive signs of sarcoid involvement of the heart [2, 12–15, 18–21]. In our patients with biopsy-proven CS, abnormal Gd-MRI and 18FDG-PET findings were observed in 94% and 80%, respectively. In addition to diagnostic utility, these imaging methods may have wider applications in CS. Thus, the extent of myocardial late enhancement at cardiac Gd-MRI is directly related to the impairment of LV function and bears prognostic significance [12, 20, 21]. Both of these imaging methods, as well as the longer-established cardiac gallium-67 scintigraphy , may also demonstrate the response to steroid therapy. However, without histopathological verification, even typical abnormalities at cardiac MRI or PET hardly justify the diagnosis of or the long-term immunosuppression for CS. We think that the prognostic notoriety of CS also underscores the importance of an unequivocal diagnosis and helps to make the risks of diagnostic biopsies more acceptable.
It is noteworthy that we found increased 18F-FDG uptake in mediastinal lymph nodes, suggestive of active inflammation, in 13 of 15 patients (87%) despite lack of hilar lymphadenopathy according to chest X-rays. In accordance with our experience, Otsuka et al. recently reported that the finding of mediastinal lymphadenopathy evaluated by computed tomography is not uncommon in CS despite normal-appearing conventional chest X-rays . Their explanation, based on animal experiments, was that the upper mediastinal lymph nodes receive drainage from the heart whereas the lung and bronchial regions drain into the hilar nodes. There have been reports of solitary cases of CS diagnosed by mediastinal lymph-node biopsy [24, 25] but our study is the first to include several patients and to suggest a high success rate for this approach in suspected CS.
In addition to its retrospective design, some other limitations of our analyses need to be recognized. As our patients presented with purely cardiac symptoms and did not develop clinical evidence of systemic sarcoidosis on follow-up, wider investigations for extracardiac sarcoidosis such as high-resolution computed tomography of the chest or ophthalmological and dermatological consultations were not routinely conducted. Thus, although our patients had isolated CS from a cardiology perspective, some of them may have had clinically silent involvement as suggested by results of whole-body PET scanning and elevated serum levels of ACE and LZM. Yet, a recent autopsy study showed that truly isolated cardiac involvement is not uncommon in fatal sarcoidosis . Another important limitation of our approach is that the number of false-negative biopsy diagnoses remains unknown. This relates to the 22 patients with suggestive clinical and imaging features of isolated CS who ultimately had negative EMBs. Therefore, although we found that repeated and imaging-guided biopsies of the myocardium or mediastinal lymph nodes clearly diminished the proportion of undiagnosed CS cases, the true sensitivity of this approach remains uncertain.
In conclusion, our report shows that sarcoidosis commonly involves the heart without clinically apparent extracardiac disease. Patients with CS are frequently young or middle-aged women presenting with serious cardiac arrhythmias and/or conduction disturbances combined with signs of myocardial disease. Abnormalities at cardiac imaging, in particular Gd-MRI and 18FDG-PET, are frequent and suggest localized areas of myocardial damage and/or inflammation. Repeated imaging-guided EMBs are commonly needed for correct histopathological diagnosis. Furthermore, in a clinical setting suggestive of CS, ‘hot’ mediastinal lymph nodes at 18FDG-PET provide a biopsy target with a high success rate of exposing sarcoid histopathology.
We acknowledge support for this research from the Finnish Foundation for Cardiovascular Research and from a Finnish government grant for medical research (EVO).