Heart transplantation is the last stage of cardiac failure treatment and improves survival. However, the waiting time is becoming longer over the years because donors have changed and 20-30% of patients die while being on the waiting list. The survival rate once the heart is transplanted is 85% after 1 yr and 76% after 5 yr . If the optimal medical treatment failed and if a graft is not immediately available, mechanical circulatory assistance is possible as a bridge to transplantation. Mortality under ventricular assist device (VAD) is usually related to multiorgan failure (MOF) secondary to an impaired splanchnic circulation [2,3]. Thus, monitoring the splanchnic tissue perfusion may assess the discrepancy between normal systemic perfusion and abnormal microcirculatory perfusion. Patients treated by VAD as an emergency are specifically at risk of hepatic, splanchnic or renal failure [4,5].
Gastric tonometry (measurement of gastric intra-mucosal pressure of CO2) allows splanchnic perfusion monitoring and has already been studied in many fields including cardiac surgery [6-8].
The purpose of this work was to determine the prognostic indicators including gastric tonometry monitoring during the early postoperative hours after implantation of a VAD.
This prospective non-randomized study was performed in a university hospital postoperative ICU. The protocol used in the present study was part of our routine clinical practice and ethical approval was given by the institutional review board (Comité Consultatif de Protection des Personnes dans la Recherche Biomédicale Lyon B) of our institution (Hospices Civils de Lyon, France). All consecutive patients (from the same university hospital) referred for VAD implantations were included in the study between 1999 and 2003.
The pneumatic system by Thoratec® (Thoratec Corporation, Pleasanton, CA, USA) was used in all patients. Implantation was performed under general anaesthesia and cardiopulmonary bypass (CPB). Preoperative inotropic support (dobutamine, dopamine, epinephrine) was administrated until CPB was started. Postoperative haemodynamic management was performed using volume expansion and vasoconstrictive drugs (norepinephrine).
Standard monitoring included cardiac rhythm and frequency, ST segment analysis, urine output, central temperature, pulse oximetry, expired end-tidal CO2 (petCO2) central venous pressure, an intra-arterial catheter and a pulmonary artery fiberoptic catheter (Swan-Ganz CCombo, CCO/SvO2; Edwards Laboratory, Irvine, CA, USA). During the postoperative period, the same monitoring was used and all patients were mechanically ventilated using the same mode (Bi-level positive airway pressure).
All patients were equipped with a gastric tube in the postoperative period so that a gastric tonometry catheter could be inserted both for the measurement of gastric mucosal partial pressure of carbon dioxide (PgCO2) and to ensure that the stomach was empty. This additional monitoring was initiated either in the operating room or in the ICU if transoesophageal echocardiography had been performed during the procedure. Gastric tonometry measured PgCO2 at steady state. The protocol for PgCO2 measurement has been described in detail elsewhere . Briefly, it involved a manual use of a 0.9% saline solution and a contact time of 30 min. At the end of this contact time, a sample of the saline solution was withdrawn and assayed for PgCO2 on an automat CORNING model 288 (Bayer Diagnostics, Tarrytown, NY, USA). Blood was simultaneously sampled for arterial blood gas analysis and lactate level. The gastric intra-mucosal pH (pHi) was calculated from the measured CO2 and calculated arterial bicarbonate using the Henderson-Hasselbach equation: pHi = 6.1 + log 10 (arterial) HCO3/PCO2 (tonometer) ×K (K = time-dependent equilibration constant).
Renal failure was monitored using serum creatinin and urea concentration . Hepatic dysfunction was monitored using total serum bilirubin concentration (normal range <20 μmol L−1) .
Data were collected at each of the following steps: during surgery, at admission to ICU (H0), 24 and 48 h after surgery (H + 24 and H + 48, respectively) and when norepinephrine was discontinued (T0). Preoperative patient and haemodynamic data were also collected as well as norepinephrine concentrations (μg kg−1 min−1) were collected.
Parsonnet  and Euroscore  were used as preoperative scoring system. SAPS II  and APACHE II scores [14,15] were used as postoperative scoring system.
Statistical analysis was performed using SPSS 10.0 for Windows (SPSS Inc., Chicago, IL, USA). Data are presented as mean ± SD (or extreme). The primary endpoint was mortality. Receiver operating characteristic (ROC) curves were generated for PgCO2, pHi, and lactate concentration by varying the discriminating threshold of each parameter to predict mortality and areas under the ROC curves were calculated. After analysis of variance, quantitative data were compared using the t-test. Correlations were assessed using Pearson's coefficient correlation. A P-value of less than 0.05 was considered to be statistically significant.
Fifty-six patients (50 men and 6 women) were included. Mean age was 46 ± 13 yr. The initial cardiac disease was diagnosed 154 ± 129 months before implantation. Mean time between admission to ICU and VAD implantation was 7 h (range from 1 to 360 h) and the mean VAD duration was 60 days ranging from 4 to 236 days. Patients' characteristics are summarized in Table 1.
In 91% of cases the VAD was biventricular, in one case it was right sided and the remaining ones were left assistance after preoperative evaluation. Indications for assistance were ischaemic cardiomyopathy in 38%, non-obstructive cardiomyopathy in 48% and miscellaneous in 14%.
Preoperative biologic data (regarding liver and kidney dysfunction) are summarized in Table 2. Surgery and CPB durations were 220 ± 120 and 113 ± 53 min, respectively.
Twenty-seven deaths occurred during the study. Among them, 16 occurred before cardiac transplantation (mean time = 18 ± 16 days after VAD) and 11 occurred after transplantation (mean time = 24 days (1 day-12 months)). Death was related to MOF in nine cases. The other causes were represented by septic shock (five cases), massive cerebral vascular stroke (four cases), massive haemorrhage (two cases) and miscellaneous (seven cases). Among survivors, VAD was a bridge to transplantation in 27 cases and a bridge to recovery in two cases.
There were no significant variations in PgCO2 (P = 0.44) over time compared to H0 (Fig. 1). Similarly, arterial pH (pHa) and mucosal pH (pHi) were stable during this period. Plasma lactate level progressively normalized within 48 h (Table 3).
Parameters significantly associated with mortality by univariate analysis are presented in Table 4. Obesity (body mass index>30) and low preoperative PaO2 value were significantly associated with death. There was no statistical relationship between mortality and body surface area. Preoperative liver and kidney dysfunction as well as preoperative lactate level was not associated with death.
Among the four scores studied (Parsonnet, Euroscore, SAPS II and APACHE II), only APACHE II was significantly associated with mortality. A cut-off value of 11 was able to predict mortality with a sensitivity of 75% and a specificity of 56%. Duration of CPB alone was not associated with death (Table 4). However, total length of surgery, duration of mechanical ventilation and red cell transfusion were significantly higher in non-survivors (Table 4).
Biologic data at the admission to ICU were significantly associated with mortality. Arterial lactate concentration at H0 had a high discriminating power to predict mortality (area under the ROC curve (AUC) = 0.771). PgCO2 measurement was only significantly associated with mortality at H0 (Fig. 2). Mean values of PgCO2 at H0 were 39 ± 14 mmHg in the non-survival group vs. 31 ± 7 mmHg in the survival group (P = 0.04). AUC for PgCO2 as a prognostic index of mortality was 0.695. A cut-off value of 31 mmHg corresponded to a sensitivity of 70% and a specificity of 62%. pHi value at H0 was also significantly different between non-survivors and survivors (7.30 ± 0.18 vs. 7.46 ± 0.07, respectively; P = 0.006). However, pHi predictive value was weak (AUC = 0.219). We have provided an analysis only for MOF-related mortality for PgCO2 and lactate concentration at admission to ICU. These parameters were associated with mortality but neither PgCO2 nor lactate concentration were specially associated with MOF-related mortality. The norepinephrine dose, 24 h after surgery, was higher in the non-survivors (Table 4).
One of the major causes of death in patients treated by VAD is MOF and related splanchnic ischaemia or hypoxia . This study shows that gastric tonometry is an accurate and predictive indicator of mortality following VAD implantation when measured in the early postoperative period. Gastric tonometry is a non-invasive tool that has potential clinical applications in this setting.
This study found that obesity, but not body surface area, was a predictive factor of mortality following VAD implantation. This observation is probably related to the difficulty to provide a sufficient cardiac output by VAD in this population. The Thoratec® system cannot provide more than 6 L min−1 output. According to Reinhartz and colleagues , weight and body surface area are not predictive factors whereas preoperative liver function was related to mortality in patients with VAD. In our sample of patients, liver and kidney dysfunction as well as preoperative lactate level were not associated with death probably because severe preoperative dysfunction is a contraindication to VAD implantation in our institution.
Many studies [6-8] have reported the interest of gastric tonometry in ICU patients and in various other clinical situations (cardiac surgery, major abdominal surgery, polytraumatized, sepsis or during cardiogenic shock). Gastric tonometry usually uses the gastric and arterial gradient of the PCO2. Lévy and colleagues  showed that a gradient >20 mmHg presented a sensitivity of 70% and a specificity of 72% to predict mortality in patients in a general ICU. In our study, we chose to focus on the absolute value of PgCO2. Janssens Uwe and colleagues  showed that PgCO2 and the P (g-a) PCO2 followed the same variations. Bernardin G and colleagues  showed that PaCO2 and PgCO2 were modified in an identical way by the modifications of the alveolar exchanges, explaining that the gradient remained stable. On the other hand, pHi was modified by the variations of these exchanges.
In this study we observed that gastric tonometry pHi and PgCO2 immediately after admission to ICU were the only significant tonometry values associated with mortality. In 1995, Gutierrez and Brown  came to the same conclusion in the abdominal aorta surgery setting.
Poeze and colleagues , in a randomized, double blind controlled study showed that the preoperative value of PgCO2 constituted the main predictive factor of mortality in high-risk surgical patients. These patients were operated on for major abdominal surgery except aortic surgery. A pHi lower or equal to 7.35 was significantly associated with mortality (P = 0.0001) and MOF. They found that a pHi value ≤7.35 was able to predict death with a sensitivity of 86.4% and a specificity of 63.6%. In our study, pHi had a lower predictive value compared to these previously published results.
Maynard and colleagues  monitored splanchnic ischaemia among patients with circulatory failure using gastric tonometry and compared its prognostic value to pulmonary artery monitoring alone. The principal criterion was death. Whereas the pulmonary pressures did not show any difference between survivors and non-survivors, pHi presented an 88% sensitivity to predict death. For note, only the value at 24 h was an independent predictive factor.
PgCO2 variations have been studied during circulatory support using an intra-aortic balloon pump (IABP) with opposite results. Janssens and colleagues  found that the IABP had no influence on the values of gastric tonometry until 36 h. After this period of time, a rise in PgCO2 was observed. Heinze and colleagues  in a recent published study have determined that IABP improved global and regional splanchnic (using lactate and gastric tonometry) perfusion in cardiac surgical patients. During circulatory assistance by Thoratec® there was no significant rise. However, absence of PgCO2 variation did not assure good tissue perfusion in the remainder splanchnic area. Creteur and colleagues  showed correlation between P (g-a) CO2 and splanchnic blood flow (determined by indocyanine green dilution) in septic and ventilated patients. Uusaro and colleagues  found similar results after cardiac surgery and suggested a non-homogeneous distribution of blood flow in the splanchnic area after cardiac surgery. This redistribution was due to CPB-induced vasoplegia and could explain why we observed a correlation between CPB duration and PCO2 value.
O'Malley and colleagues  published the first study concerning the prognostic value of gastric tonometry during left ventricular circulatory assistance by HeartMate I® (Thoratec Corporation Pleasanton, CA, USA). Although our two studies differ by the number of patients included and the type of assistance, a significant rise in PgCO2 was observed at the end of the surgery and this rise was significantly higher in the non-survivor group (P = 0.004). In this study, the CO2 gap at the end of operation was associated with a higher MOD score (r = 0.64, P = 0.0033). In our study, there was no relation between PgCO2 and MOF. This difference can be explained in a number of ways. We did not have the same population (no cardiac arrest in O'Malley's study vs. 16 in ours) and in our study, haematological dysfunction was not included as an organ failure. Stroke and haemorrhage are not only related to the device but also to the poor coagulation profile. Patients with MOF and especially those with liver failure or requiring dialysis are very difficult to maintain in the therapeutic heparin target range. O'Malley and colleagues found a significant relationship between MOF and PgCO2. However, their patients were equipped with left VAD. Our sample of patients included 91% of biventricular assist devices and this may explain liver and/or kidney failure being less frequent and occurring later in our study.
All of these previously published results lead to a careful interpretation of gastric tonometry data in patients treated by circulatory support. Wippermann and colleagues  found a pathologic pHi in 49% of patients after cardiac surgery and in 83% of children after cardiac surgery, whereas only 6% of them had symptoms of organ dysfunction. It must be emphasized that PgCO2 (or P (g-a) CO2) are measured values whereas pHi is a calculated one. That may explain why pHi is less frequently used in daily practice.
Vasopressor drugs are commonly used during early postoperative hours after VAD implantation because of vasoplegia and the inflammatory response. In a previously published review, Silva and colleagues  concluded that the effects of vasoactive drugs on gastric tonometry parameters (pHi) were unpredictable. O'Malley and colleagues demonstrated a correlation between the rise in PgCO2 and the use of norepinephrine (P = 0.001, r = 0.69) but did not clearly explain this observation. The results of our study suggest that a large dose of norepinephrine during the first 24 h after surgery is statistically associated with death. This certainly indicates a higher level of vasoplegia in the non-survivors group.
O'Malley and colleagues  noted that higher doses of norepinephrine were associated with an elevated CO2 gap at the end of operation whatever the systemic blood flow. This suggests that the redistribution of blood flow induced by vasopressor drugs interacts both on the splanchnic macro- and microcirculation.
Only APACHE II score was associated with mortality (P = 0.028) with a strong discriminating capacity (AUC = 0.745) in our study. This score was previously used by Gracin and colleagues  to select candidates for left VAD implantation. This could only be performed in ICU patients and not in an emergency. Patients who had APACHE II scores between 11 and 20 derived the greatest benefit from left VAD placement. In our study, a score of 11 corresponded to 75% sensivity and 56% specificity to predict mortality.
Lactate level on ICU admission was also significantly associated with mortality and its discriminating power was related to PCO2 (AUC = 0.771 vs. 0.695).
As lactate concentration gives information on global and macrocirculation, gastric tonometry offers information on regional and microcirculation. Splanchnic and, in particular, gastric tissue is the first to suffer from ischaemia due to blood flow redistribution. Moreover, gastric tonometry provides continuous monitoring. Patients with a VAD are different from patients in a medical ICU and so SAPS II, APACHE II or other scoring systems (that are also calculated during the first 24 h) are not valid.
Two methods are available for PgCO2 measurement: manual with saline solution and semi-automatic with air (Tonocap TC 200, Tonometric Division, Instrumentarium, Helsinki, Finland). We started the study with the first method and followed with the same to limit the bias related to the method. Janssens and colleagues  showed a good correlation between the two methods, but there was an overestimation of PgCO2 at high carbon dioxide concentration with saline solution.
VAD is a rescue technique for patients with cardiac failure, but remains under evaluation. Many parameters can be used as predictive factors. Gastric tonometry is a non-invasive and simple tool. Interpretation of tonometry results remains difficult because there is no linear correlation between PgCO2 and splanchnic blood flow. Nevertheless, several studies have assessed the interest of this monitoring in various clinical situations.
Only early postoperative values at ICU admission have a predictive value. Gastric tonometry can be combined with lactate or APACHE II score after emergency cardiac assistance implantation as predictive factors of mortality. Norepinephrine role in VAD needs further studies.
The authors thank Jean Ninet, MD, Jean François Obadia, MD, PhD, Roland Henaine, MD, Olivier Raisky, MD, Jacques Robin, MD, Lehot JJ, MD and Robert Rousson, MD, PhD, for their contribution to this study.
1. Davies RA, Badovinac K, Haddad H et al
. Heart transplantation at the Ottawa Heart Institute: comparison with Canadian and international results. Artif Organs
2. Fiddian Green RG. Associations between intramucosal acidosis in the gut and organ failure. Crit Care Med
3. Scalia S, Sharma P, Rodriguez J et al
. Decreased mesenteric blood flow in experimental multiple organ failure. J Surg Res
4. Nose Y, Ohtsubo S, Tayama E. Therapeutic and physiological artificial heart: furure prospects. Artif Organs
5. Minev PA, El-Banayosy A, Minami K et al
. Differential indication for mechanical circulatory support following heart transplantation. Intensive Care Med
6. Groeneveld AB, Kolkman JJ. Splanchnic tonometry: a review of physiology, methodology and clinical applications. J Crit Care
7. Wippermann CF, Schmid FX, Kampmann C et al
. Evaluation of gartric intramucosal pH during and after pediatric cardiac surgery. Eur J Cardiothorac Surg
8. Lebuffe G, Decoene C, Pol A et al
. Regional capnometry with air-automated tonometry detects circulatory failure earlier than conventional hemodynamics after cardiac surgery. Anesth Analg
9. Bellomo R, Ronco C, Kellum JA et al
. Acute renal failure - definition, outcome measures, animal models, fluid therapy and information technology needs: the Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care
10. Reinhartz O, Farrar DJ, Hershon JH et al
. Importance of preoperative liver function as a predictor of survival in patients supported with thoratec ventricular assist devices as a bridge to transplantation. J Thorac Cardiovasc Surg
11. Parsonnet V, Dean D, Bernstein AD. A metod of uniform stratification of risk for evaluation the result of surgery in aquired adult heart disease. Circulation
(Suppl 1): 3-12.
12. Nashef SAM, Roques F, Michel P et al
. European System for Cardiac Operative Risk Evaluation (EuroSCORE). Eur J Cardiothorac Surg
13. Le Gall JR, Lemeshow S, Saulnier F. A new simplified Acute Physiology Score (SAPS II) based on a European/North American Multicenter study. JAMA
14. Knaus WA, Draper EA, Wagner DP et al
. APACHE II: a severity of disease classification system. Crit Care Med
15. Aegerter P, Boumendi A, Retbi A et al
. SAPS II revisited. Intensive Care Med
16. Levy B, Gawalkiewicz P, Vallet B et al
. Gastric capnometry with air-automated tonometry predicts outcomes in critically ill patients. Crit Care Med
17. Janssens U, Graf J, Koch KC et al
. Gastric tonometry in patients with cardiogenic shock and intra-aortic balloon counter pulsation. Crit Care Med
18. Bernardin G, Lucas P, Hyvernat H et al
. Influence of alveolar ventilation changes on calculated gastric intramucosal pH and gastric-arterial PCO2
difference. Intensive Care Med
19. Gutierrez G, Brown SD. Gastric tonometry: a new monotoring modality in the intensive care unit. J Intensive Care Med
20. Poeze M, Takala J, Greve JWM et al
. Pre-operative tonometry is predictive for mortality
and morbidity in high-risk surgical patients. Intensive Care Med
21. Maynard N, Bihari D, Beale R et al
. Assessment of splanchnic oxygenation by gastric tonometry in patients with acute circulatory failure. JAMA
22. Heinze H, Heringlake M, Schmucker P et al
. Effects of intra-aortic balloon counterpulsation on parameters of tissue oxygenation. Eur J Anaesthesiol
23. Creteur J, De Backer D, Vincent JL. Does gastric tonometry monitor splanchnic perfusion? Crit Care Med
24. Uusaro A, Ruokonen E, Takala J. Gastric mucosal pH does not reflect changes in splanchnic blood flow after cardiac surgery. Br J Anaesth
25. O'Malley C, Frumento RJ, Mets B et al
. Abnormal gastric tonometric variables and vasoconstrictor used after left ventricular assist device insertion. Ann Thorac Surg
26. Silva E, De Baker D, Creteur J et al
. Effects of vasoactive drugs on gastric intramucosal pH. Crit Care Med
27. Gracin N, Johnson MR, Spokas D et al
. The use of APACHE II scores to select candidates for left ventricular assist device placement. J Heart Lung Transplant
28. Janssens U, Graf J, Koch KC et al
. Gastric tonometry: in vivo comparison of saline and air tonometry in patients with cardiogenic shock. Br J Anaesth