Introduction

Viral myocarditis is a common cardiac disease that is identified in up to 9 % of post-mortem examinations [1, 2]. It appears to be a major cause of sudden, unexpected death, and may progress to dilated cardiomyopathy [3].

Several groups recently described possible predictors of clinical outcome in viral myocarditis and dilated cardiomyopathy, including clinical parameters [4], cardiovascular MR [5], myocardial inflammation [4], and the presence of viral genomes in the myocardium [6]. Nevertheless, risk stratification in these patients remains a difficult business [7], and especially the role of PVB19 genomes in the myocardium continues to be controversial [6, 8].

Consequently, the primary objective of this study was to establish the prognostic value of the myocardial load of PVB19 genomes in patients presenting for endomyocardial biopsy work-up of myocarditis and/or unclear non-ischemic cardiomyopathy in comparison to clinical and cardiovascular MR parameters. Specifically, we sought to identify the best predictors for adverse events in this patient group during a long-term follow-up.

Methods

Patient population

One-hundred-eight consecutive patients presenting at any of the participating institutions between January 2007 and June 2011 for endomyocardial biopsy work-up of suspected myocarditis and/or unclear non-ischemic cardiomyopathy (all comers) were enrolled in the long-term follow-up if they fulfilled the following criteria: (1) presence of PVB19 genome in the myocardium, AND (2) coronary artery disease ruled out by invasive angiography, AND (3) successfully underwent CMR or Echo imaging for assessment of ventricular size and function. Patients with valvular or congenital heart disease were not included. The study has been approved by the local ethics committee and the study protocol conformed to the ethical guidelines of the 1964 Declaration of Helsinki and its later amendments. All patients gave informed consent prior to their inclusion in the study. Some of the patients (n = 36) were part of a previous report [5].

Endomyocardial biopsy protocol

At least five biopsies were preferentially taken from the ventricle demonstrating LGE. Patients demonstrating LGE exclusively in the LV lateral wall underwent selective LV biopsies (21 %), while those demonstrating LGE in the septum or having no LGE at all underwent either biventricular (59 %), or selective RV biopsies (20 %).

Histopathological analysis

Endomyocardial biopsies were stained with Masson’s trichrome as well as Giemsa and examined by light microscopy. For immunohistology, tissue sections were treated with an avidin–biotin-immunoperoxidase method (Vectastain Elite Kit, Vector, Burlingame, CA), applying the following monoclonal antibodies: CD3 (T-cells; Novocastra Laboratories, Newcastle, UK), CD68 (macrophages; DAKO, Hamburg, Germany), and HLA-DR (DAKO, Hamburg, Germany) as described previously. The presence of focal or diffuse mononuclear infiltrates with >14 leukocytes per mm2 (CD3+ T lymphocytes and/or CD68+ macrophages) in the myocardium, in addition to enhanced expression of HLA class molecules, was used for the diagnosis of inflammation [5, 9].

Fulfilling the above criteria, acute myocarditis was diagnosed by extended interstitial infiltration of inflammatory cells (see above) with necrosis of adjacent myocytes.

Chronic myocarditis was characterized by ongoing degeneration of myocytes, persistent inflammation, and presence of myocardial remodeling/fibrosis.

DCM was characterized by the presence of remodeling/fibrosis with no signs of inflammation.

“PVB19 genome only” was diagnosed in the absence of myocyte necrosis and the absence of inflammation, e.g., consistent with healed myocarditis.

Detection of viral genomes

DNA and RNA were extracted with the use of proteinase-K digestion followed by extraction with phenol/chloroform. Nested polymerase chain reaction/reverse transcriptase–polymerase chain reaction was performed for the detection of viral genomes as described [6]. As control for successful extraction of myocardial nucleic acids, the housekeeping gene (GAPDH) was detected by PCR [4]. Specificity of all viral amplification products was confirmed by automatic DNA sequencing [9].

CMR protocol

ECG gated CMR imaging was performed in breath-hold using a 1.5T Magnetom Sonata, or Magnetom Aera (Siemens-Healthcare, Germany) in line with SCMR/EuroCMR recommendations [10]. Both cine and LGE short-axis CMR images were prescribed every 10 mm (slice thickness 6 mm) from base to apex. In-plane resolution was typically 1.2 × 1.8 mm. Cine CMR was performed using a steady-state-free-precession-sequence. LGE images were acquired on average 5–10 min after contrast administration using segmented IR-GRE [11] constantly adjusting inversion time [12]. The contrast dose (Gadodiamide or Gadopentetat-Dimeglumin) was 0.15 mmol/kg.

CMR analysis

Cine and contrast images were evaluated by two experienced observers as described elsewhere [5, 9]. In brief, endocardial and epicardial borders were outlined on the short-axis cine images. Volumes and LV-EF were derived by summation of epicardial and endocardial contours. The LV-mass was calculated by subtracting endocardial from epicardial volume at end-diastole and multiplying by 1.05 g/cm3 [13]. LGE was assessed using the Siemens Argus analysis software package.

Clinical follow-up

Clinical follow-up was performed using a standardized questionnaire at least 2 years after initial presentation. In case of a suspected event, all necessary medical records were reviewed by some of the authors (S.G., I.K., J.S., A.P., H.M.) acting as end-point committee.

Variables, endpoints, and definitions

All variables were collected directly from patients, and/or medical records except CMR parameters, which were evaluated as described above. Variables include general characteristics and follow-up results. Most variables are self-explanatory; all others are defined below.

There were two primary combined endpoints named endpoint 1 and endpoint 2. Endpoint 1 “all cause death” was defined as SCD, or cardiac death, or non-cardiac death. Thus, endpoint 1 could only be reached by suffering a lethal event. Endpoint 2 was defined as either cardiac death, or aborted SCD, or hospitalization for heart failure. Consequently, endpoint 2 could be reached by suffering either a lethal or a non-lethal event such as hospitalization for heart failure. The explicit meaning of the events is described in the following paragraphs:

Death: death from any cause.

Cardiac death: death from all cardiac causes.

SCD: unexpected arrest of presumed cardiac origin within 1 h after onset of any symptoms that could be interpreted as being cardiac in origin.

Aborted SCD: resuscitation after cardiac arrest defined as performance of the physical act of cardioversion and/or CPR in a patient who remains alive 28 days later.

Hospitalization for heart failure: Hospitalization as an in-patient >24 h, and heart failure as primary diagnosis according to the hospitals final report.

Statistical analysis

Absolute numbers, percentages, and medians (with quartiles) were computed to describe the patient population. Categorical variables were compared by Chi-square test or Fisher exact test as appropriate; continuous parameters by using Wilcoxon rank-sum test. The distribution of viral copy loads by histologic groups is presented as Box-plot. Kaplan–Meier curves were calculated for visualizing the cumulative event-free survival of patients for both endpoints. A log-rank test was performed to compare both survival curves. A multivariable Cox proportional hazard model was used for analyzing independent associations with mortality and other endpoints. p values <0.05 were considered significant. All p values are results of two-tailed tests. Statistical analyses were performed using the SAS© statistical package, version 9.2 (SAS, Cary, North Carolina).

Results

Patient characteristics

Chronic myocarditis according to the definition described above was the most frequent histopathologic diagnosis in this patient cohort, followed by end stage DCM, and the presence of PVB19 genome only. Acute myocarditis was observed in one patient. Patients classified as “others” include hypertensive heart disease, hypertrophic cardiomyopathy after remodeling, Tako-Tsubo cardiomyopathy, and myocardial amyloidosis. Dyspnea and angina were the primary reasons to seek medical attention, followed by fatigue and various combinations of heart failure and other symptoms (Table 1).

Table 1 Baseline patient characteristics
Full size table

As per inclusion criteria, the genome of PVB19 was present in the myocardium of all patients. In nine patients, the viral load was <100 GE of PVB19. Seventy-two patients had between 100 and 499 GE of PVB19 in their myocardium, and more than 500 GE could be detected in 27 patients (IQR 559, 846, max. 2450 GE). Note that there seems to be a trend toward a lower viral load in the patient cohorts with DCM and the presence of PVB19 genome only (e.g., healed myocarditis) as final diagnosis (Fig. 1).

Fig. 1
figure 1

Distribution of myocardial viral load. Final diagnosis “Chronic Myocarditis,” “DCM,” “PVB19 genome only,” and “Other” was made on the basis of histological evaluation of all myocardial biopsy samples (see text for definitions). The only patient diagnosed with acute myocarditis is not displayed in this figure. Note that there seems to be a trend toward a lower viral load in the patient cohort with DCM and PVB19 genome only as final diagnosis made by histopathology

Full size image

No patient was treated with anti-viral, or anti-inflammatory medication prior to inclusion or during follow-up, but all patients with heart failure received state of the art heart failure medication. If clinically indicated, ICD implantation was performed (n = 15 patients).

Imaging findings

The mean LV-EF was 40 %, and the mean LV-EDV was 175 ml. CMR imaging was performed in 90/108 patients (Table 1); in all others, ventricular function and size were evaluated by echo (18/108). LGE was present in 50/90 patients, most commonly occurring in a non-ischemic pattern located in the subepicardial or intramural areas of the LV. Typical patient examples are displayed in Fig. 2.

Fig. 2
figure 2

Histological findings in patients with different entities. Three patients examples presenting with reduced LV-EF but different histological findings: Top row normal histology, but up to 720 GE of PVB19, LV-EF 15 %. Middle row Histology revealed chronic myocarditis with up to 1200 GE of PVB19, LV-EF 29 %. Bottom row diagnosed with DCM by histology (no inflammation but fibrosis), and 213 GE of PVB19, LV-EF 40 %

Full size image

Dividing our patient population in subgroups with LV-EF above and below 45 % revealed that patients with LV-EF <45 % had larger ventricles, were more often diagnosed with DCM, had a higher prevalence of LGE, higher NYHA classes, and were more likely to suffer from dyspnea than from chest pain (Table 2). Comparing patients with LGE to patients without LGE demonstrates larger ventricles, poorer ventricular function, and a higher prevalence of DCM in the group with LGE (Table 3). When looking at different viral loads in the myocardium (Table 4), our data reveal the best ventricular function in the group with the highest viral load (>500 GE) and the highest incidence of DCM in the group with the lowest viral load. In addition, patients with myocardial inflammation by histology (Table 5) had more angina and a trend toward a higher viral load compared to patients without inflammation, who had larger ventricles and poorer LV-EF (mostly due to end stage DCM).

Table 2 Characteristics of patients with LVEF <45 % and LVEF >45 %
Full size table
Table 3 Characteristics of patients with LGE and no LGE
Full size table
Table 4 Characteristics of patients with different myocardial viral loads (<99 GE, 100–499 GE, >500 GE)
Full size table
Table 5 Characteristics of patients with inflammation and no inflammation
Full size table

Follow-up results

During follow-up 11 of 108 patients died, two patients suffered SCD but were successfully shocked by their ICD, and 21 patients were hospitalized for heart failure (Table 1). Thus, eleven patients reached endpoint 1 “all cause death” as described above (Table 6). Most of the lethal events (n = 8) occurred for cardiac reasons. Of the remaining three patients, one died from severe cerebral hemorrhage, one from lung cancer, and one from lymphoma.

Table 6 Univariate analysis endpoint 1—all cause death
Full size table

In addition, 27 patients reached endpoint 2 including cardiac death, aborted SCD, and hospitalization for heart failure (Table 7). Note that there is no relation between reaching endpoint 1 or endpoint 2 and the viral load in the myocardium, but 90 % of patients reaching endpoint 1 demonstrated myocardial LGE.

Table 7 Univariate analysis endpoint 2—cardiac death or aborted SCD or hospitalization for heart failure
Full size table

Predictors of events

For evaluation of predictors for adverse events, we looked at (1) all patients who reached endpoint 1 (Table 6) and (2) all patients who reached endpoint 2 (Table 7). There was no significant correlation between clinical presentation and endpoint 1 (Table 6). However, ventricular extrasystoles, which may be a surrogate parameter for undetected arrhythmias, symptoms of heart failure, which are a surrogate parameter of impaired LV-EF, and LV-EF itself were related to endpoint 2.

In addition, functional parameters such as LV-EF, LV-EDV (for endpoint 2), as well as the histologic diagnosis of DCM and the presence of LGE (for all endpoints) reached statistical significance in the univariate analysis. In fact, the presence of LGE yielded an odds-ratio for a lethal event of 8.56 (endpoint 1), and of 5.52 for endpoint 2.

Kaplan–Meier survival curves for endpoint 1 comparing LV-EF, the presence of LGE, and the myocardial load of PVB19 genomes are displayed in Fig. 3a–c. Note that only one patient without LGE and one with normal LV-EF died during follow-up (both from cancer as described above). Figure 4a–c displays the Kaplan–Meier survival curves for endpoint 2.

Fig. 3
figure 3

Kaplan–Meier survival curves—Endpoint 1. Kaplan–Meier Survival Curves with regard to endpoint 1 (all cause death), displayed for LV-EF, presence of LGE, and myocardial PVB19 load. The number of patients at risk is shown at the bottom of the figure. Note that in the group without any LGE only a single patient suffered an event (death due to malignant disease, see text for details) during follow-up

Full size image
Fig. 4
figure 4

Kaplan–Meier survival curves—Endpoint 2. Kaplan–Meier survival curves with regard to endpoint 2 (including cardiac death, aborted SCD and hospitalization for heart failure), displayed for LV-EF, presence of LGE, and myocardial PVB19 load. The number of patients at risk is shown at the bottom of the figure

Full size image

Multivariable Cox regression analysis including the presence of LGE, the initial LV-EF, the initial LV-EDV, the viral load, and the histologic diagnosis of DCM revealed a trend for LV-EDV at the initial presentation (p = 0.07, hazard ratio 1.02 per ml for endpoint 1) as a possible independent predictor of lethal events. In this model, the presence of LGE (p = 0.17), the viral load (p = 0.16), the histological diagnosis of DCM (p = 0.16), and the LV-EF upon presentation did not reach significance for endpoint 1. Looking at patients suffering, endpoint 2 revealed a trend for LV-EF at the initial presentation (p = 0.07, hazard ratio 1.04 per  % EF for endpoint 2), and the presence of LGE (p = 0.07, hazard-ratio 8.9 for endpoint 2) as possible independent predictors of endpoint 2. All other parameters, including viral load (p = 0.79) were not significant. Typical patient examples are viewed in Fig. 5.

Fig. 5
figure 5

Typical Example of Patients With and Without Major Adverse Event during Follow-up. The left panel (Patient A) shows CMR images of a 29-year-old female patient with normal LV-EF (67 %) and no LGE. Due to chest pain, prior febrile infection, frequent ventricular extrasystoles, and positive troponine, she underwent coronary angiography in which CAD could be ruled out. Myocardial biopsies revealed chronic myocarditis and a PVB19 load of 846 GE. Despite this relatively high viral load, she did not suffer any event during follow up. Patient B displays CMR results of a 40-year-old female presenting with palpitations, fatigue, chest pain, and dyspnea. One month ago she had a severe febrile infection. CMR showed a reduced LV-EF (40 %) and LGE in the posterolateral wall (white arrows), suggesting myocarditis. CAD was ruled out by coronary angiography. Myocardial biopsies revealed chronic myocarditis and 330 GE of PVB19. Despite this relatively low viral load, the patient had reduced LV-EF, positive LGE, and suffered SCD during follow-up

Full size image

Discussion

This study is of clinical importance since we clearly demonstrate that the viral load of PVB19 genomes in the myocardium is not related to the long-term clinical outcome. Furthermore, our data suggest a growing role of non-invasive imaging parameters such as ventricular size and function, as well as LGE for risk stratification in patients with non-ischemic myocardial disease. Note that no patient with normal LVEF or the absence of LGE suffered cardiac death during long-term follow-up.

Patient characteristics

Dyspnea and angina were the primary reasons to seek medical attention, followed by fatigue and various combinations of heart failure and other symptoms, which is similar to other published patient cohorts presenting with PVB19-related myocardial disease [8, 14]. Furthermore, the mean myocardial viral load in the present study is in line with other recent German reports [6, 15].

Our data reveal a trend toward a lower viral load in the patient cohort with DCM as final diagnosis made by histopathology. This supports the results of other groups [6, 16], but challenges findings of Stewart et al. [8], who found the highest viral copy numbers in patients with DCM. A possible explanation could be the different patient populations studied, and the fact that PVB19 genome persistence in human tissues can be life-long [17], independent of active inflammation, representing a source of information about past and not necessarily of recent events.

CMR findings

Despite a median LV-EF of 40 %, there was a broad spectrum of LV impairment ranging from severely impaired to completely normal ventricular function (IQR 30–58 %). In the 90 patients undergoing CMR imaging, LGE was present in 55.6 % and was usually located in a non-CAD–pattern in the subepicardial or intramural areas of the LV, as described previously [9, 14, 18]. Patients with scar indicated by LGE had larger ventricles and poorer LV-EF compared to those without scar (Table 3). This finding also matches the results from other inflammatory [5] or non-ischemic cardiomyopathy populations [19].

Interestingly, the best ventricular function was seen in the group with the highest myocardial viral load (>500 GE), or in the presence of myocardial inflammation (Table 5). The highest incidence of the final diagnosis of end stage DCM was in the group with the lowest viral load, or in the group without myocardial inflammation. Whereas the high incidence of DCM in the group with the lowest viral load and without inflammation conceptually makes sense, reflecting some form of partial virus elimination over time [6], and the fact that DCM patients do not have myocardial inflammation by our definition above, the high load of viral genomes and the presence of inflammation in the group with the best LV-EF are more difficult to understand. However, the most likely explanation for the better LV-EF is that in this group with high viral load and/or active inflammation the end stage of post inflammatory heart disease with fibrous myocardial remodeling resulting in poor function has not yet been reached (as it has been reached in the DCM group). Stewart et al. [8] also described a better ventricular function for the PVB19 positive group compared to patients without myocardial PVB19 genome presence, concluding that the detection of PVB19 genome by PCR alone may not be sufficient to explain a pathologic effect [20].

Follow-up results and predictors of events

In our population with symptoms ranging from mild to severe, all-cause mortality was 10.2 %, and cardiac mortality was 7.4 %. SCD (including aborted SCD) occurred in 6.5 % of patients during follow-up. Thus, our event rate was much lower than in Mason’s Myocarditis Trial [21], most likely due to different inclusion criteria and disease severity, but almost as high as in the non-ischemic cardiomyopathy group of the SCD-HEFT trial [22], although LV function was better in our cohort underscoring the importance of risk stratification and optimal clinical management in these patients.

Importantly, the present data clearly indicate that the viral load of PVB19 genomes in the myocardium is not related to the clinical outcome, as suggested by other studies [16]. The most likely explanation for this finding is based on the fact that PVB19 genome persistence in human tissues can be life-long without relevant activity and replication [17]. Thus, the detection of PVB19 genome by PCR alone is not sufficient to explain a pathologic effect. This is underscored by recent findings from Bock et al. [15] identifying PVB19 RNA replication intermediates demonstrated by RT-PCR amplification of the NS1 and VP1 regions of the PVB19 genome as a good surrogate parameter for active virus replication, which is (1) related to the inflammatory activity in the myocardium and (2) the clinical course (myocarditis vs. DCM). Furthermore, the authors discuss co-infection with other cardiotropic viruses like HHV6 in combination with host specific determinants as factors reactivating PVB19 replication from long-term persistent or latent infection. This idea is also supported by earlier data from our group describing a co-infection of PVB19 and HHV6 as a predictor for a poor clinical outcome in myocarditis patients [14]. However, additional data are needed to clarify these issues in the future.

Interestingly, despite a trend toward more cardiac deaths in the group with active inflammation (Table 5), we could not confirm myocardial inflammation as a predictor of poor outcome (Table 6 and 7), as other groups have suggested [4]. However, we believe that this is explained by the high incidence of end stage DCM in our population, since end stage DCM patients by definition do not have myocardial inflammation, but are well known for a poor prognosis, which is underscored by our finding that the histologic diagnosis of DCM is a predictor of adverse events.

When looking at long-term predictors for adverse events, we found functional and morphological parameters determined by non-invasive imaging to be most promising. In fact, ventricular size and function upon initial presentation (assessed by echo or CMR) and the presence of LGE were potential predictors for adverse events, whereas the absence of LGE was a predictor for a favorable outcome without suffering any major adverse event. Importantly, no patient with normal LV-EF or the absence of LGE suffered cardiac death during long-term follow-up. This finding underscores the value of cardiac imaging in management and risk stratification of patients with non-ischemic myocardial disease, matching the results of earlier studies [5, 23, 24].

Clinical implications

Based on our data and the results from other groups discussed above, we believe that the detection of PVB19 genome by PCR alone, as well as the viral load in the myocardium determined by this technique does not allow risk stratification in patients suffering from non-ischemic myocardial disease. Whether additional parameters such as PVB19 RNA replication intermediates serving as a surrogate parameter for active virus replication or co-infections with other viruses may play a role in the clinical routine some time in the future needs to be determined by additional studies.

Non-invasive cardiac imaging, however, including ventricular morphology, function, and LGE in particular, appears to be a valuable tool for risk stratification of patients with myocardial disease, which is ready for the clinical routine and—if normal—can give suffering patients and worrying physicians some peace of mind. Note that 90 % of patients reaching endpoint 1 demonstrated LGE in the myocardium, and that no patient with normal LV-EF or the absence of LGE suffered cardiac death during long-term follow-up in the present study, matching earlier results [5, 23, 24].

As described above, we again identified impaired LV-EF and signs of heart failure as important predictors of adverse cardiac events. This reproducible finding [5, 23, 24] once more suggests that one should carefully optimize heart failure therapy in all patients with non-ischemic myocardial disease and any signs of heart failure.

Conclusion

Our data demonstrate that the viral load of PVB19 genomes in the myocardium is not related to the long-term clinical outcome. Furthermore, this study suggests a growing role of imaging parameters such as ventricular size and function, and LGE for risk stratification in patients with non-ischemic myocardial disease.