Introduction

The oxygen-diffusion method of apnea testing is a technique commonly used in the evaluation of brain death [1, 2]. The safety of this procedure has been recently evaluated and only rarely had to be aborted [3]. Acute lung injury is common in critically ill brain-injured patients and after brain death, which often results from trauma, neurogenic pulmonary edema, or aspiration. Abnormal chest X-ray and respiratory measurement findings are typically reasons to pause before performing the apnea test. Adequate pre-test oxygenation removes alveolar nitrogenous stores and facilitates oxygen exchange [4]. We postulated that poor pretest pulmonary gas exchange would compromise a carbon dioxide challenge with direct oxygen flow close to the alveoli.

Patients and Methods

The A–a gradient and \( {{P_{{{\text{a}}_{{{\text{O}}_{ 2} }} }} } \mathord{/ {\vphantom {{P_{{{\text{a}}_{{{\text{O}}_{ 2} }} }} } {F_{{{\text{i}}_{{{\text{O}}_{ 2} }} }} }}} \kern-\nulldelimiterspace} {F_{{{\text{i}}_{{{\text{O}}_{ 2} }} }} }} \) were calculated for patients in the previously studied cohort [3]. Arterial blood gas (ABG) values were obtained immediately preceding the start of apnea testing. Patients that completed the procedure during the declaration of brain death were compared to those whose studies were aborted. The standard A–a gradient formula [5] used was:

  • A–a gradient = \( P_{{{\text{A}}_{{{\text{O}}_{ 2} }} }} - P_{{{\text{a}}_{{{\text{O}}_{{_{ 2} }} }} }} \)

  • \( P_{{{\text{A}}_{{{\text{O}}_{{_{ 2} }} }} }} = \left( {F_{{{\text{i}}_{{{\text{O}}_{ 2} }} }} \times \left[ {P_{\text{atm}} - P_{{{\text{H}}_{ 2} {\text{O}}}} } \right]} \right) - \left( {P_{{{\text{a}}_{{{\text{CO}}_{ 2} }} }} \div 0.8} \right) \)

\( P_{{{\text{A}}_{{{\text{O}}_{ 2} }} }} \) represents alveolar oxygen partial pressure, \( P_{{{\text{a}}_{{{\text{O}}_{ 2} }} }} \) represents arterial oxygen partial pressure, \( P_{{{\text{a}}_{{{\text{CO}}_{ 2} }} }} \) represents arterial carbon dioxide partial pressure, and \( F_{{{\text{i}}_{{{\text{O}}_{ 2} }} }} \) represents inspired oxygen fraction. Patm represents the atmospheric pressure (760 mmHg) and \( P_{{{\text{H}}_{ 2} {\text{O}}}} \), water partial pressure (47 mmHg at 37°C). Systolic blood pressure (SBP) preceding apnea testing was recorded.

Statistical analysis was performed using the Wilcoxon rank-sum tests for continuous variables due to non-Gaussian distribution of the data. Fisher’s exact tests were used to compare categorical variables due to limited sample size for patients who aborted apnea test. Associations between the apnea test completion status and potential risk factors were assessed using univariate logistic regression. Odds ratios along with 95% confidence intervals were reported. All the tests were 2 sided and p-values less than 0.05 were considered statistically significant. Analysis was performed using SAS version 9.0 software (SAS Inc., Cary, NC, USA).

Results

The A–a gradient was calculated for 207 of the 228 patients with a brain death evaluation. Twenty patients from the original cohort did not undergo a carbon dioxide challenge: nine patients were unsafe for testing because of refractory hypotension despite hemodynamic support; nine patients had pulmonary compromise [refractory hypoxemia and positive end expiratory pressure requirements ≥ 15 cm H2O; two patients had diffuse pulmonary edema (one with associated pneumothorax) and one diagnosed with pulmonary contusion]. Two additional patients were not tested for other reasons (e.g. cerebral silence on EEG before apnea testing attempted; barbiturate intoxication as confounder, nuclear scan performed as confirmatory study). One patient was excluded from the current analysis due to erroneous ABG values.

Apnea testing was aborted in 10 of the 207 patients because of progressive hypotension or hypoxemia that developed shortly after disconnecting ventilatory support. Table 1 summarizes the clinical characteristics and reasons for test failure. Sixty percent of those with apnea test failure were male versus 51% in the comparative group (P = 0.75). Younger patients were more likely to have aborted apnea tests (median age: 23 years, 25–75% quartile: 19–42 years) when compared to those who completed the procedure (median age: 47 years, 25–75% quartile: 25–62 years) (P = 0.02). Seventy percent of the patients in the aborted group died of complications related polytrauma versus only 25% in the comparative group.

Table 1 Clinical characteristics of patients who failed the apnea test
Full size table

A higher A–a gradient was associated with apnea test failure (P = 0.003). The median A–a gradient in this group was 376 mmHg (25–75% quartile: 255–476 mmHg) compared to 175 mmHg (25–75% quartile: 104–282 mmHg) in those who successfully completed apnea testing. Interestingly, the presence of acidemia just prior to initiation of a carbon dioxide challenge was independently associated with test completion failure (P = 0.028). The median values of those who had aborted procedures was 7.29 (25–75% quartile: 7.23–7.41) versus 7.39 (25–75% quartile: 7.33–7.44) in the comparative group. However, neither the \( {{P_{{{\text{a}}_{{{\text{O}}_{ 2} }} }} } \mathord{/ {\vphantom {{P_{{{\text{a}}_{{{\text{O}}_{ 2} }} }} } {F_{{{\text{i}}_{{{\text{O}}_{ 2} }} }} }}} \kern-\nulldelimiterspace} {F_{{{\text{i}}_{{{\text{O}}_{ 2} }} }} }} \) ratio (P = 0.14) nor other independent ABG measurements (e.g., pre-\( P_{{{\text{a}}_{{{\text{CO}}_{ 2} }} }} \) or pre-\( P_{{{\text{a}}_{{{\text{O}}_{ 2} }} }} \)) yielded similar associative power. Pre-test SBP (P value = 0.28) also did not correlate with apnea test failure. Table 2 summarizes these findings in addition to brain death etiology. Age, pre-test pH, and A–a gradient were associated with test completion failure after logistic regression analysis.

Table 2 Risk factors associated with apnea test failure and cause of death
Full size table

Discussion

Documentation of absent respiratory drive fulfills one major diagnostic criterion for the clinical determination of brain death [1, 2]. Safety concerns (e.g. hypotension, pneumothorax, cardiac dysrhythmia) while performing the apnea test were recently assessed, and the apnea test was found to be safe in most circumstances [3]. However, risk factors such as inadequate pre-test oxygenation may lead to complications [4]. In this study, we found that age, acidemia at the time of starting the apnea test, and an elevated A–a gradient are associated with inability to complete the apnea test.

The A–a gradient helps differentiate intra versus extrapulmonary causes of poor oxygenation by measuring differences between alveolar and arterial oxygen concentration. This calculation also accounts for changes in respiration as measured by \( P_{{{\text{a}}_{{{\text{CO}}_{ 2} }} }} \). Abnormally elevated gradients suggest a defect in gas exchange (e.g. ventilation/perfusion mismatch, right to left shunting). A previous study did not demonstrate the increased risk of complications from an elevated gradient preceding the apnea test, but it analyzed a smaller sample of patients [4]. The \( {{P_{{{\text{a}}_{{{\text{O}}_{ 2} }} }} } \mathord{/ {\vphantom {{P_{{{\text{a}}_{{{\text{O}}_{ 2} }} }} } {F_{{{\text{i}}_{{{\text{O}}_{ 2} }} }} }}} \kern-\nulldelimiterspace} {F_{{{\text{i}}_{{{\text{O}}_{ 2} }} }} }} \) ratio is useful at estimating severity of gas exchange at higher \( {\text{F}}_{{{\text{i}}_{{{\text{O}}_{ 2} }} }} \) levels. While the \( {{P_{{{\text{a}}_{{{\text{O}}_{ 2} }} }} } \mathord{/ {\vphantom {{P_{{{\text{a}}_{{{\text{O}}_{ 2} }} }} } {F_{{{\text{i}}_{{{\text{O}}_{ 2} }} }} }}} \kern-\nulldelimiterspace} {F_{{{\text{i}}_{{{\text{O}}_{ 2} }} }} }} \) ratio was not associated with apnea test failure in our analysis, it did demonstrate that neurologically devastated individuals undergoing brain death evaluation often have coexisting diffuse lung injury. Furthermore, we found that A–a gradients were two-fold higher in those who failed the apnea test, which was significant after logistic regression. These findings suggest that although most patients undergoing brain death evaluation have variable degrees of acute lung injury, the presence of greater intrapulmonary oxygen shunting and poorer ventilation (e.g., \( P_{{{\text{a}}_{{{\text{CO}}_{ 2} }} }} \)), as defined by the A–a gradient, lead to poorer tolerability during a carbon dioxide challenge.

A striking finding was that despite normalization of pretest \( P_{{{\text{a}}_{{{\text{CO}}_{ 2} }} }} \), the presence of concomitant acidemia was associated with apnea test failure. The superimposed respiratory acidosis that develops during the procedure likely contributes significantly to hemodynamic compromise in those with underlying acidemia. Aborted apnea tests occurred more often in younger individuals. This counterintuitive association may be explained by selection bias; more aggressive testing may have been pursued in the younger population despite poorer clinical conditions (e.g. polytrauma), while greater caution may have been exercised when pursuing the apnea test in older patients.

Physicians performing a brain death examination should be aware that acidemia and markedly increased A–a gradients preceding apnea test may compromise successful testing. Still, the A–a gradient is increased in most patients with uncomplicated apnea tests, and we could not identify a cutoff value that precludes apnea testing.