Complications of endomyocardial biopsy in children

Statistical methods

Statistical analysis was performed to identify patient and procedural variables that were predictive of a biopsy complication. Outcome variables included all complications, perforations and arrhythmias. Univariate and multivariate analysis was performed using S-Plus version 3.3 for Hewlett-Packard (Hewlett Packard Inc., Palo Alto, California) UNIX machines. Mann-Whitney Utest was used for comparison of continuous variables that did not conform to a normal distribution. Student ttest was used for continuous variables that were normally distributed. Categorical variables were compared using chi-square analysis. A p value <0.05 was considered significant. Backward stepwise logistic regression was used for the multivariate analysis. All significant variables in the univariate analysis were entered into the multivariate analysis. Significant variables (p < 0.05) in the multivariate analysis were entered into a decision-making program. This program (8)maximizes differences of each significant variable and generates a flow diagram model that displays subgroups at highest risk of perforation.

Indication

Eighty-five percent (846) of the procedures were performed for transplant rejection surveillance. Fifteen percent (154) of the procedures were performed to rule out myocarditis. See Table 1for a breakdown of the number of patients and the number of procedures for each subgroup. Of those with suspected myocarditis, 115 had a dilated cardiomyopathy with depressed systolic function. Thirty-eight patients had an unexplained arrhythmia with normal cardiac function. Additionally, one patient with a multisystem glycogen storage disease underwent a biopsy to assess possible cardiac involvement.

Operator experience

As part of a teaching institution, pediatric cardiology fellows participate in a large percentage of cardiac catheterizations. A biopsy performed by a fellow was differentiated from a biopsy performed by an attending. A fellow performed the biopsy under direct supervision of an attending physician in 457 cases. An attending physician performed the case without the aid of a fellow in 543 cases. Each attending physician had significant biopsy experience prior to this study date. There were 11 complications, including 3 perforations, when a fellow performed the biopsy procedure. Twelve complications, including six perforations, occurred when an attending physician performed the procedure. There was no statistical difference in the complication rates between biopsies performed by fellows compared with biopsies performed by attending physicians. There was no difference in attending physician complication rates, and there was no difference in complication rates in the first half of the study compared with the latter half.

Biopsy site

The biopsies were performed in the right ventricle in 986 cases and in the left ventricle in 14 cases. Biopsies of the left ventricle were done to evaluate isolated left ventricular dysfunction in 10 cases. One patient with a heterotopic heart transplant, in whom the allograft right ventricle was inaccessible, underwent four left ventricular biopsies from the right femoral artery approach.

Vascular access

Vascular access was obtained through the right internal jugular vein in 533 (53%), right femoral vein in 327 (33%), left femoral vein in 92 (9%), left internal jugular vein in 44 (4%) and right femoral artery in 4 (1%). Of the 14 left ventricular biopsies, 4 were obtained from the femoral artery and 10 were obtained through the patent foramen ovale from the femoral vein.

Patient status

The procedure was considered elective in 965 patients (96%). An elective procedure was defined as a case in which the patient was hemodynamically stable and not reliant upon inotropic, ventilatory or mechanical support. The remaining 35 procedures were performed on ill patients receiving intravenous inotropic support alone (9 cases, 1%), ventilation alone (8 cases, 1%) or both intravenous inotropes and ventilation (18 cases, 2%). Patients that were intubated only for airway protection; usually small infants were considered elective cases.

Biopsy results

Histology results were divided into five categories. Severe allograft rejection (grades 3b–4, classified according to the standardized criteria developed by The International Society for Heart and Lung Transplantation [9]) was found in 22 cases (2.6% of biopsies performed for rejection surveillance). Moderate rejection (grade 2 to 3a) was found in 193 cases (22.8% of biopsies performed for transplant rejection surveillance). Acute inflammatory changes consistent with active myocarditis were seen in 40 cases (26% of biopsies performed for suspected myocarditis). Nonspecific abnormalities including fibrosis, myocyte hypertrophy and resolving or borderline myocarditis were seen in 44 cases (28.6% of biopsies performed for suspected myocarditis). No significant abnormality, including grade 0 to 1 rejection, was seen in 696 cases (69.6% of all biopsies). This included 70 cases (45.4%) of biopsies performed for suspected myocarditis and 626 cases (74%) of biopsies for transplant rejection surveillance.

Cardiac perforation

There was one biopsy-related death, secondary to perforation of the right ventricular outflow tract in a two-week-old, 3.6 kg infant with dilated cardiomyopathy. Resuscitation with immediate pericardiocentesis and sternotomy was performed but was unsuccessful. There were nine total perforations for an overall incidence of 0.9% (Table 2). Eight, including the fatal case, occurred among the 154 patients being evaluated for possible myocarditis for a 5.2% incidence. Among infants or young children less than 10 kg on intravenous inotropic support biopsied for undiagnosed cardiomyopathy, the incidence of perforation was 33.3% (5/15 cases). The final perforation occurred in a patient undergoing heart transplant rejection surveillance (1 of 846) for a 0.1% incidence. This patient did not decompensate or require immediate therapeutic intervention. Histological exam of the biopsy specimen revealed epicardium and an epicardial coronary vessel. The patient developed a loud continuous murmur and small pericardial effusion. Subsequent angiography (Fig. 1)confirmed the presence of a coronary artery to right ventricular fistula and aneurysm of the left anterior descending coronary artery. This patient remains asymptomatic and has subsequently undergone further biopsies without complications. Of the other seven surviving patients with perforation, six required immediate interventions for hemodynamic compromise associated with cardiac tamponade. Emergent echocardiography, pericardiocentesis and surgery were performed in the catheterization laboratory in five cases and in the operating room in one case. Tamponade was relieved and perforations were localized and successfully closed. In the remaining case, hemodynamic compromise did not develop until the patient was stabilized in the intensive care unit. Pericardiocentesis without surgery was successful for stabilization. There were no neurological deficits associated with perforation in the eight surviving patients.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Selective left coronary angiogram in the right anterior oblique view (A)and four chamber view (B), showing diffusely dilated left anterior descending branch with an aneurysmal dilation (open arrows)and multiple fistulous communications to the right ventricle (arrow heads).

Significant risk factors for perforation in the univariate analysis included: younger patient age (p = 0.001), smaller size (p < 0.001), nonelective status (p < 0.001), cardiomyopathy as an indication for biopsy (p < 0.001) and femoral approach (p < 0.01). Operator experience, bioptome type and biopsy histology were not significant variables in the univariate analysis. Multivariate analysis demonstrated that the greatest risk of perforation was in children biopsied as part of the evaluation for myocarditis (p = 0.01) and in those requiring inotropic support (p < 0.01). The multivariate analysis is graphically displayed using a decision tree software package (Fig. 2). Maximizing differences in the significant variables from the multivariate analysis generated each level of this graph.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Graphic display of statistically significant risk factors for myocardial perforation with endomyocardial biopsy. The cardiomyopathy group includes patients with dilated cardiomyopathy and arrhythmia who were biopsied for suspicion of myocarditis.

Although 577 (58%) of the procedures were performed from the internal jugular vein, all nine perforations were from the femoral venous approach (p < 0.01). Perforation occurred more commonly when the procedure was performed from the left femoral vein (6 of 92) compared with the right femoral vein approach (4 of 327) (p < 0.01). Site of vascular access, however, was not found to be a significant variable in the multivariate analysis.

Arrhythmia

Significant arrhythmia, requiring intervention, occurred in five patients for an incidence of 0.5% (Table 2). Atrial fibrillation occurred in two patients, bradycardia with hypotension in one patient and ventricular fibrillation in two patients. Each arrhythmia was controlled in the catheterization laboratory with antiarrythmics or by cardioversion with no adverse outcome. All five of these arrhythmias occurred during biopsy from the femoral venous approach and four of the five occurred during manipulation of the catheter or sheath. Also, four of five arrhythmias occurred in patients with dilated cardiomyopathy, none of whom had biopsy proven myocarditis. Multivariate analysis demonstrated the greatest risk of arrhythmia occurred in children requiring intravenous inotropic support (p < 0.01).

Other complications

Two patients undergoing EMB from the right internal jugular approach required chest tube placement for pneumothorax and hemothorax. One patient had worsening tricuspid regurgitation due to an avulsed chordae tendinae. Endocardial stripping was noted on echocardiogram after left ventricular biopsy. A mobile pedunculated echodense strip of endomyocardium was noted along the left ventricular free wall. It was postulated that the bioptome stripped the endomyocardium during an unsuccessful attempt to obtain a biopsy specimen. A generalized seizure probably related to the use of chlorpromazine hydrochloride occurred in one patient with a previous history of seizures. Two patients had significant deterioration of their heart failure. One patient had worsening pulmonary edema and required aggressive diuresis. Another patient with dilated cardiomyopathy became hemodynamically unstable and required an intraaortic balloon pump for stabilization.

Summary of morbidity and mortality

The overall incidence of a serious complication from endomyocardial biopsy was 19 of 1,000 or 1.9%. Overall perforation rate was 0.9% and overall mortality was 0.1%. In the evaluation of possible myocarditis in dilated cardiomyopathy patients or patients with new onset arrhythmia, the incidence of complication was 14 of 154 or 9.1%, perforation was 5.2% and mortality was 0.6%. In patients receiving endomyocardial biopsy for transplant rejection surveillance, the incidence of serious complication was 5 of 846 or 0.6%, perforation was 0.1% and there was no mortality.

Incidence of complications

Endomyocardial biopsy remains the standard tool for diagnosis of myocarditis and allograft rejection. Since Lurie et al. (10)reported their initial EMB experience in the pediatric age group, there have been several small series describing complications such as pneumothorax, hemothorax, dysrhythmia, heart block, perforation and death. Complication rates vary from none (3–5,11)to 10.6% (4). This study reflects comparable overall pediatric EMB complication rates while demonstrating the relative safety of rejection surveillance EMB compared with the risk of EMB in small sick patients with suspected myocarditis.

Risk factors for complications

Technical difficulties described in adults are magnified in children. The internal jugular venous approach has lessened some of these difficulties (12), but percutaneous access from the neck is technically more demanding in infants. Endomyocardial biopsy from the groin (13)necessitates a long preformed outer sheath to properly position the bioptome in the right ventricle. Proper sheath curvature and placement is crucial in avoiding trauma to the thin right ventricular outflow tract. Partial removal of the sheath prior to obtaining each biopsy will minimize the possibilities of the sheath contributing to perforation. Biopsy from the internal jugular venous approach does not require an additional sheath within the heart.

Statistically significant risk factors for perforation in this series included inotropic dependency and cardiomyopathy. It is likely that this reflects the thinner right ventricular wall in patients with dilated cardiomyopathy. Intrinsic disease of the myocardium, such as myocarditis, may further weaken the structural integrity of the ventricular wall, thus increasing the risk of perforation.

In this series, all perforations occurred using the femoral venous approach with an increased incidence from the left groin versus the right groin. This finding, however, did not reach statistical significance in the multivariate analysis and might merely reflect other factors, such as an increased failure rate of percutaneous access to the right femoral vein in the smaller, sicker patients. It might reflect preexisting or previous lines with thrombosis of the right femoral venous system. Similarly, the femoral venous approach as compared with the jugular approach was not found to be a statistically significant risk factor for perforation in the multivariate analysis. The higher incidence of perforation in this group might reflect that the femoral venous approach was often the operator’s first choice of access in the sickest patients.

Utility of EMB

The clinical utility of EMB for evaluating patients with possible myocarditis has been debated (11,13,15). Nevertheless, EMB continues to be widely used in many centers for the evaluation of possible myocarditis. It has been shown to be a useful diagnostic as well as prognostic tool for unexplained cardiac dysfunction (15–18). While we recognize that the use of corticosteroids, intravenous gamma globulin and other more potent immunosuppressants have not been studied in randomized controlled trials in the pediatric patient with biopsy proven myocarditis. Histological confirmation of an inflammatory cardiomyopathy may alter management of the patient in other ways. Cardiac dysfunction secondary to active inflammation is frequently reversible with restoration of normal cardiac function in many pediatric patients (15). As a cardiac transplant center with extracorporeal membrane oxgenation support and cardiac assist device capabilities, an endomyocardial biopsy has proven to be a valuable tool in the evaluation of the critically ill patient referred for transplantation for new onset cardiomyopathy (18). When inflammatory cardiomyopathy is confirmed, transplantation is deferred until the patient has been given a full chance to undergo myocardial recovery. This would include, if necessary, a trial of extracorporeal membrane oxgenation support or left ventricular (or biventricular) assistance. If there was no evidence of an inflammatory process, then the critically sick patient with ventricular failure would be considered for transplantation at the earliest possible time.

Small, sick children with possible myocarditis pose the most difficult diagnostic dilemmas. While the described benefits of knowing if a child with acute cardiomyopathy has myocarditis also apply to the infant population, the risk of biopsy in the critically sick infant or small child (<10 kg) appears excessive. Indeed, in this study there was a 33% incidence of cardiac perforation in this subgroup. In light of this, we would no longer recommend that this group of patients undergo endomyocardial biopsy as a routine part of their evaluation for possible myocarditis.

Summary

Compared with the established risk of EMB in adults, there is an increased risk in children, especially in children dependent on inotropes undergoing EMB for the evaluation of myocarditis. In contrast to patients with dilated cardiomyopathy, we, like others (6), have found that EMB in the evaluation of allograft rejection carries a very low risk and can be performed with acceptable morbidity and no mortality (14).