ARE THERE CLINICAL CLUES TO AETIOLOGIC DIAGNOSIS?

It would be useful if the different patterns of clinical presentation were sensitive or specific for given aetiologies. This is the case in most instances, but not invariably so. As previously stated, in patients with acute pericarditis and features suggesting associated disease the likelihood of this being the cause for the pericardial syndrome is very high.⁷ Therefore, associated disease should be investigated accordingly.

In the immunologically competent patient presenting with classically acute pericarditis, the probability of acute idiopathic pericarditis is about 90%,^3,4 this probability being even greater when clinical features subside in the following few days. No further studies are needed in such cases. Subsequent relapses are also highly specific for idiopathic pericarditis.

Tamponade in the context of acute pericarditis is a more complex problem. In a study of our group,³ tamponade was significantly more common in patients with tuberculous or purulent pericarditis or neoplasia, but the high prevalence of acute idiopathic pericarditis made it the leading cause in absolute terms (table 1). However, in tamponade without findings of inflammation (chest pain or rub) we have demonstrated⁷ a likelihood ratio of 3.0 for neoplasia, although other aetiologies should be considered as well.

Other features such as a sustained clinical course (for example, over three weeks) also increase the likelihood of specific disease. However, in our part of the world,³ the predominance of idiopathic disease in any presentation of acute pericarditis and no associated conditions may be taken as a clinical rule. An exception is tamponade with massively haemorrhagic effusion, often but not always caused by malignancy.⁸ In addition, purulent pericarditis should always be considered in predisposing diseases (such as pleural empyema or mediastinal infection), even in the absence of definite features of pericarditis or tamponade.⁹

Thus, in only comparatively few patients with acute pericardial disease (mainly those with haemodynamic compromise or protracted course) will specific pericardial investigations be needed for aetiologic diagnosis.

DIAGNOSTIC YIELD OF PERICARDIOCENTESIS, PERICARDIAL BIOPSY, AND PERICARDIOSCOPY

Since 1977 our group has undertaken several cohort studies of patients with pericardial disease.^3,4,9,10 A key finding of these studies³ was that the diagnostic yield of either pericardiocentesis or subxiphoid pericardial drainage with biopsy were significantly much greater when performed on patients with cardiac tamponade than when indicated for purely diagnostic purposes (for example 35% v 5%³). In addition, we showed that persistent mild or moderate effusion was not by itself an indication for routine invasive diagnostic studies, as no late events associated with effusion were observed and effusion resolved in most cases.¹⁰ Suggested alternative diagnoses may not be clinically relevant.¹¹ Probably, the most appropriate approach represents a compromise between performing too many unnecessary invasive studies and missing too many specific diagnoses.

In recent years, greater sensitivity and specificity of pericardioscopy than pericardial drainage or biopsy has been claimed,^1,2 and it has been advocated as the preferred diagnostic technique. Although pericardioscopy may offer an advantage in selected patients, the available evidence does not seem to justify a widening of the indications for pericardial drainage. In addition to a likely selection bias, it is not clear whether a diagnosis relevant for management purposes could not be achieved by other means in most patients. Probably pericardioscopy should not modify the basic approach to the management of patients with pericardial effusion, although it may improve the diagnostic yield in selected patients or be the preferred method for drainage in experienced hands.

DIAGNOSIS OF TUBERCULOUS AND NEOPLASTIC PERICARDITIS

In our part of the world tuberculous pericarditis should not be diagnosed unless confirmed by microbiology or histology. Although the search for tubercle bacilli in pericardial fluid has a low yield, histology is more sensitive.¹² The value of identifying tubercle bacilli in other sources (sputum, gastric aspirate or biopsy of large lymph nodes) cannot be overemphasised.¹² For prognostic purposes a quick diagnosis is desirable, and thus one may ask whether high concentrations of adenosine deaminase in pleural or pericardial fluid should by themselves be sufficient diagnostic evidence.¹³ Our policy is to consider them just supportive rather than confirmative evidence of tuberculosis and to proceed with the search for mycobacteria or definite granulomas. On the other hand, low concentrations in pleural fluid should indeed not be taken as definitive evidence against pericardial tuberculosis. The diagnostic value of intermediate pericardial values (30–70 iu/l) is probably not well documented. Although it is likely that DNA amplification techniques in pericardial fluid may simplify the diagnosis,¹⁴ clinical experience is not sufficient for a definitive answer.

Primary or secondary neoplastic involvement of the pericardium may develop in a few patients without previously known malignancy, usually presenting with tamponade.³ Unlike in tuberculosis, here examination of the fluid has a larger yield than unguided biopsy techniques.^1,2

NEED FOR A COMPREHENSIVE APPROACH TO AETIOLOGIC DIAGNOSIS

There is not enough evidence in the literature to give hard and fast rules for the aetiologic diagnosis in all cases of acute pericardial disease, and a systematic approach such as our own, which is based on an experience of more than two decades, seems justified. Essentially, our guideline^7,15 consists of three stages.

Stage 1 includes basic laboratory studies, chest x ray, and Doppler echocardiogram. For patients without associated illness in whom clinical disease does not remit within one week or who have clinical features of tamponade, anti-DNA antibodies, rheumatoid factor, and three sputum or gastric aspirate cultures for mycobacteria are carried out. If definite pleural effusion is present, thoracocentesis is indicated. In addition to routine cytological and biochemical studies in pleural fluid, adenosine deaminase activity is measured and a search for mycobacteria is also done. Stage 1 also includes any investigation (such as computed tomographic scan, lymph node biopsy, serologic tests for toxoplasma, legionella, and mycoplasma, or assessment of the immunologic status) warranted by individual clinical findings.

Stage 2 involves pericardiocentesis, which is only indicated to treat cardiac tamponade and when purulent pericarditis is suspected. In pericardial fluid the same investigations as indicated for pleural fluid are performed. In appropriate laboratories, polymerase chain reaction (PCR) for tubercle bacilli can be measured.

Stage 3 involves subxiphoid pericardial drainage and biopsy, with histological examination (including stains for mycobacteria) and culture of the biopsy specimen. We indicate biopsy when pericardiocentesis is ineffective or tamponade relapses, and in selected patients with definite clinical disease of more than three weeks duration and no diagnosis. The decision for choosing pericardiocentesis rather than surgical drainage in the mentioned indications is based on local experience.

Our guideline showed good diagnostic performance in an initial study,³ and it has been validated in other settings.^7,9 Despite its possible limitations, this approach—which could be modified depending on the frequency of particular disorders in other populations—provides a rational and adequate way to manage most patients with acute pericardial disease. Chronic pericardial effusion requires a different approach.^7,16

A relevant question is whether our guideline, or the principles underlying it, should be modified in patients with HIV infection. It seems tempting to enlarge the indications of pericardial drainage in these patients, as a wider spectrum of specific aetiologies has been suggested for their pericardial effusions.¹⁷ However, although further studies are needed, in most cases no specific aetiology is identified,¹⁸ and when it is, it may be of questionable clinical relevance because of the patient’s condition.¹⁹ Therefore, while individualised attitudes may be warranted, for the time being there seems to be no reason to change a basically conservative approach to pericardial drainage in HIV patients, which should be indicated only to treat haemodynamic embarrassment or when there is a strong suspicion of treatable specific disease.