A 70-year-old man with a history of well-controlled hypertension and angina was admitted with a ruptured abdominal aortic aneurysm (De Bakey type IIIb) which presented with sudden ripping abdominal pain that became unbearable and had the patient writhing in agony. On examination, he was a well-built man in acute distress, with a heart rate of 57 min−1, a blood pressure of 110/70 mmHg and no pulse deficit.
In the operating theatre, a radial artery cannula, an introducer for a pulmonary arterial catheter and a 12-French gauge central venous catheter were set in place - the last being connected to a rapid infusion system. Anaesthesia was induced with fentanyl, etomidate and vecuronium and the trachea was intubated, at which time the arterial blood pressure decreased to about 50 mmHg for about 30 s. As surgery started, a Hewlett Packard 5 Mhz transoesophageal omniplane probe (Palo Alto, CA) was inserted into the oesophagus and a pulmonary arterial catheter was inserted via the introducer. A dopamine infusion of up to 10 μg kg−1 min−1 was used to keep the mean arterial pressure above 60 mmHg. The heart rate ranged between 95 and 105 min−1 and the first available reading of mean pulmonary artery pressure was 16 mmHg, with a wedge of 8 mmHg. Opening the abdomen revealed half a litre of free blood, a retroperitoneal haematoma and a ruptured infrarenal aortic aneurysm. Control of the proximal aorta was achieved within a few minutes.
Echocardiographic monitoring was carried out using Hewlett Packard (Palo Alto, CA) equipment, principally with the transverse plane, four-chamber view. About 20 min after aortic cross-clamping, the patient showed more pronounced two-dimensional and Doppler echocardiographic signs of acute myocardial ischaemia, and the haemodynamic status became unstable with an increased requirement for catecholamine support. A tube graft was successfully inserted into the aorta, but while the abdomen was being closed, the patient suffered a cardiac arrest in ventricular fibrillation. This responded to a first period CPR which was carried out in accordance with the recommendations of the American Heart Association (JAMA 1992; 268: 2199-2241) with an optimal chest compression rate of about 100 min−1. However, a second cardiac arrest followed 5 min later which was unresponsive to 40 min of resuscitative effort. The autopsy showed a posterior myocardial infarction, which was consistent with echocardiographic findings before and during CPR.
The findings during the earlier period of CPR were as follows. During the downstroke of chest compression, the mitral valve closed as the aortic valve opened simultaneously (Fig. 1) and the left ventricular cross-sectional area was reduced by 21% (thus making this an artificial systole). During the release of chest compression, the aortic valve closed, the mitral valve opened (Fig. 2), and transmitral Doppler inflow reached its maximum to refill the ventricle in an artificial diastole. However, after about 5 min of the second attempt at CPR, there was a diminution in both the valve movements and the 'diastolic' transmitral inflow: the ultimate failure of resuscitation was foreshadowed by the failure of the mitral valve to close at all during chest compression.
The exact mechanism underlying CPR remains unclear. Kouwenhoven et al. hypothesised that forward flow during chest compression occurred primarily because of compression of the ventricle between the sternum and vertebrae combined with the normal competence of the heart valves ('cardiac pump' hypothesis). MacKenzie et al. measured cardiac output during CPR and demonstrated that the pressures measured by appropriately placed catheters at the aortic root and in the right atrium were raised equally during chest compression - as a consequence, they argued, of the general increase in intrathoracic pressure. Rudikof et al. demonstrated further that the rise in intrathoracic pressure collapsed the great veins (but not the arteries) at the thoracic outlets so that the raised intrathoracic pressure was transmitted more effectively to the extrathoracic arteries than to the extrathoracic veins, which produced the arteriovenous pressure gradient for forward flow in the systemic circulation. This was the basis for the 'thoracic pump' hypothesis.
Clearly, an appreciation of valve movement and chamber size during CPR is fundamental to understanding the exact mechanism of blood flow in this pathophysiological circumstance. Transthoracic two-dimensional echocardiography has been used to demonstrate mitral valve position during closed-chest compression in animals [4,5,6] and humans [7,8], but the technique is limited by motion artefacts and does not provide adequate resolution for precise analysis of valve movement. Werner et al. and Rich  used the technique in early studies, which supported the thoracic pump hypothesis. They observed that the mitral and aortic valves remained open during both the application and release phase of chest compression, and that there were no changes in left ventricular cross-sectional area. The thoracic pump hypothesis assumes that blood flow occurs simply because of intermittent compression of the central blood volume by increase in intrathoracic pressure and that the heart and pulmonary vessels are merely a passive conduit. By contrast, Deshmukh et al. reported normal mitral valve function during CPR in pigs  and humans , although valve closure became less complete after 5 min in those for whom resuscitation proved unsuccessful. They also observed a 25% reduction in left ventricular cross-sectional area during the compression phase. Feneley et al. carried out an extensive echocardiographic study of how different ways of compressing the chest can affect mitral valve function. The sharp, high velocity sternal descent that is a feature of compressions given at rates above 60 min−1 creates a left ventriculo-atrial pressure gradient that closes the mitral valve and opens the aortic valve so that the left ventricle empties and reduces in size. With the more gradual sternal descent that is associated with slower compression rates of between 40 and 60−1, the mitral valve fails to close. Hackl et al1  reported similar results with pigs.
The failure of mitral valve closure reported by Werner et al. during CPR in humans may be because of the following reasons. The patients they studied had been pronounced brain-dead before cardiac arrest occurred and had not received adrenaline or any other cardiotonic agent. The time between cardiac arrest and echocardiographic recording was not stated, and cardiac viability and mitral valve function may already have deterioriated before imaging began. Furthermore no changes in ventricular size were observed, which might indicate that the depth of sternal descent had not been enough to compress the heart. The report of CPR in two patients by Clements et al. is more difficult to interpret because the underlying rhythm was not ventricular fibrillation but ventricular tachycardia in one case and sinus tachycardia in the other. Finally, because of differences between the thoracic anatomy in the pig, the dog and man, it might not be appropriate to compare the mechanisms of CPR between species.
Transoesophageal echocardiography provides an excellent window, in dogs  as well as in humans [12-14], for visualizing the mitral valve, the left ventricle and its outlet, the aortic valve, during CPR. Halpern et al., using transoesophageal echocardiography in dogs, tested whether mechanical compression of the heart or the general rise in intrathoracic pressure was the mechanism of mitral valve closure during CPR. They changed intrathoracic pressure by inflating and deflating the chest either through a cannula through the chest wall while the airway was open (so that the lungs were deflated and inflated reciprocally with the chest), or by inflating and deflating the lungs directly via the airway with the transthoracic cannula clamped (so that the lungs were inflated synchronously with the chest). Rises in intrathoracic pressure accompanied by lung deflation produced higher left ventricular than left atrial pressures and closed the mitral valve, whereas rises in intrathoracic pressure associated with lung inflation through the airway left the mitral valve open. Different human studies using transoesophageal echocardiography have presented different observations of the behaviour of the mitral valve during chest compression. Higano et al. reported that the mitral valve closed during manual chest compression and opened during the release phase, whereas Wright  showed that it stayed open throughout the phase of chest compression. However, Wright's paper fails to state the underlying cardiac rhythm or the time intervals between the onset of the rhythm disturbance, the start of CPR and the start of transoesophageal echocardiographic imaging: nor does it state the medications that were used during CPR. Recently, Porter et al. reported controversial results from transoesophageal echocardiography, namely that the mitral valve can close or stay open during the compression phase of CPR so that there was no consistent relation between Doppler transmitral flow during the release phase and left ventricular fractional shortening during the compression phase of CTR. They assumed that factors such as non-uniform increases in intrathoracic pressure determined the position of the mitral valve during chest compression but failed to recognize that, in any severe myocardial ischaemia, the papillary muscle is ultimately weakened to the extent that it allows regurgitant flow, which is the only thing that can explain a difference between 'diastolic transmitral inflow' and 'systolic' fractional shortening.
We suspect that the absence of valve motion observed in previously reported failures of CPR with echocardiographic recording is because of the same loss of valvular function that we observed after 5 min in this case in which resuscitation ultimately failed. Our observations of the importance of normal valve function support the cardiac pump theory, according to which the greatest cardiac output during CPR should be obtained by optimizing the velocity, the depth and the number of cardiac compressions in unit time rather than simply the duration for which the thorax is compressed.
1 Kouwenhoven WB, Jude JR, Knickerbocker CG. Closed chest cardiac massage. JAMA
2 MacKenzie GJ, Taylor SH, McDonald A, Donald K. Hemodynamic effects of external cardiac compression. Lancet
3 Rudikoff MT, Maughan WL, Effron M, Freund P, Weisfeldt ML. Mechanisms of blood flow during cardiopulmonary resuscitation. Circulation
4 Feneley MP, Maier GW, Gaynor JW, Gall SA, Kisslo JA, Davis JW. Sequence of mitral valve motion and transmitral blood flow during manual cardiopulmonary resuscitation in dog. Circulation
5 Hackl W, Simon P, Mauritz W, Steinbereithner K. Echocardiographic assessment of mitral valve function during mechanical cardiopulmonary resuscitation in pigs. Anesth Analg
6 Deshmukh HG, Weil MN, Rackow EC, Trevino R, Bisera J. Echocardiographic observations during cardiopulmonary resuscitation: a preliminary report. Crit Care Med
7 Werner JA, Greene HL, Janko CL, Cobb LA. Visualization of cardiac valve motion in man during external chest compression using two-dimensional echocardiography. Implications regarding blood flow. Circulation
8 Deshmukh HG, Weil MH, Gudipati CV, Trevino R, Bissera J. Mechanism of blood flow generated by precordial compression during CPR. I Studies on closed chest precordial compression. Chest
9 Rich S, Wix HL, Shapiro EP. Clinical assessment of heart chamber size and valve motion during cardiopulmonary resuscitation by two-dimensional echocardiography. Am Heart J
10 Clements FM, Bruijn NP, Kissol JA. Transesophageal echocardiographic observation in patients undergoing closed chest massage. Anesthesiology
11 Halpern HR, Weiss JL, Guerci AD et al.
Cyclic elevation of intrathoracic pressure can close the mitral valve during cardiac arrest in dog. Circulation
12 Higano ST, Oh JK, Ewy GA, Seward JB. The mechanism of blood flow during closed chest cardiac massage in humans: transesophageal echocardiographic observations. Mayo Clin Proc
13 Wright RF. Transesophageal echocardiography during cardiopulmonary resuscitation in humans. Circulation
14 Porter TR, Ornato JP, Guards CS, Roy VG, Burns CA, Nixon JV. Transesophageal echocardiography to assess mitral valve function and flow during cardiopulmonary resuscitation. Am J Cardiol