SEARCH

SEARCH BY CITATION

Keywords:

  • Cirrhosis;
  • exercise;
  • liver transplantation;
  • oscillatory ventilation;
  • oxygen uptake

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure
  10. References

Our aim was to determine preoperative aerobic capacity (oxygen uptake [V′O2]) and prevalence of exercise oscillatory ventilation (EOV), underlying clinical characteristics of patients with EOV, and significance of reduced aerobic capacity and EOV in predicting mortality after liver transplantation. We prospectively studied 263 patients who underwent elective liver transplantation. Patients were followed up for 1 year. Despite minor impairment of resting cardiopulmonary function, preoperative aerobic capacity was reduced (peak V′O2: 64 ± 19% predicted). EOV occurred in 10% of patients. Model for End-Stage Liver Disease score tended to be higher in patients with EOV compared to patients without, but failed to reach significance (p = 0.09). EOV patients had lower peak V′O2 and higher ventilatory drive. EOV was more frequent in nonsurvivors than in survivors (30% vs. 9%, p = 0.01) and was independently associated with posttransplant all-cause 1-year mortality. Reduced peak V′O2 best predicted the primary composite endpoint defined as 1-year mortality and/or prolonged hospitalization and early in-hospital mortality. Multivariate analysis revealed EOV (χ2, 3.96; p = 0.04) and V′O22, 4.28; p = 0.04) as independent predictors of mortality and so-called primary composite endpoint, respectively. EOV and reduced peak V′O2 may identify high-risk candidates for liver transplantation, which would motivate a more aggressive treatment when detected.


Abbreviations
AT

ventilatory anaerobic threshold

CPX

cardiopulmonary exercise

DL,CO

diffusing capacity of the lung for carbon monoxide

EOV

exercise oscillatory ventilation

FEV1

forced expiratory volume in 1 s

FVC

forced vital capacity

MELD

Model for End-Stage Liver Disease

Pa,CO2

arterial carbon dioxide partial pressure

RER

respiratory exchange ratio

TLC

total lung capacity

V′CO2

carbon dioxide output

VD/VT

physiological dead space/tidal volume ratio

V′E

minute ventilation

V′O2

oxygen uptake

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure
  10. References

Liver transplantation has become the treatment of choice for patients with end-stage liver disease and hepatocellular carcinoma. Advances in surgical and anesthetic techniques along with new immunosuppressive drugs have contributed to improvement in patient survival after liver transplantation [1]. In cirrhotic patients on the waiting list for liver transplant, compelling data exist for predicting complications and mortality using the Model for End-Stage Liver Disease (MELD) score [2-4]. Whereas MELD score has been shown useful for adequate liver allocation, clinicians still face difficulties in accurately predicting patient outcome after liver transplantation [4, 5]. Preoperative evaluation of cardiopulmonary function and aerobic capacity (peak oxygen uptake [V′O2]) has been proposed to better predict immediate postoperative outcome and early mortality after transplantation [6-9]. However, these clinical studies have yielded conflicting data that mainly rely on subgroup analyses, which potentially lead to type II error induced by insufficient statistical power. In line with the clinical importance of preoperative assessments in liver transplantation, whether pretransplant impaired aerobic capacity can prognosticate posttransplant outcome was the first objective of the present study.

Peculiar patterns of ventilation, such as exercise oscillatory breathing or ventilation (EOV), have gained primary importance as markers of severity and prognosis of cardiovascular diseases [10, 11]. EOV is characterized by waxing and waning of tidal volume leading to an oscillatory kinetics of breathing with a period of 45–90 s. Theoretically, mechanisms of ventilation instability that lead to EOV include prolonged circulatory time (i.e. time lag), abnormal response of central and peripheral controllers that drive ventilation (i.e. controller gain) and impairment of the efficiency of gas exchange system to damp out changes in arterial gas tensions (i.e. controlled plant gain). In patients with heart failure, EOV has been attributed to prolonged circulatory time due to reduced cardiac output [10, 11]. In patients with neurological disorders such as hepatic encephalopathy, prolonged circulatory time may not be responsible for EOV as normal or even hyperdynamic circulatory status is typically present in advanced liver diseases [12]. Instead, in these clinical conditions, ventilation instability has been related to overstimulation of the ventilatory control center, which thus favors periodic breathing [12]. Specifically, such an overstimulation of the ventilatory drive has been consistently reported in severe cirrhotic patients and further related to increased chemoreceptor sensitivity in response to elevated catecholamine and progesterone plasma concentrations [13-15].

Despite the well-known impairment of respiratory control in cirrhotic patients, there is no study focusing on the prevalence of exercise periodic breathing in severe liver failure. Because EOV emerged as a strong predictor of survival in end-stage heart failure, whether EOV can prognosticate post–liver transplant outcome was the second objective of our study. Accordingly, our objectives were (1) to determine preoperative aerobic capacity and the prevalence of EOV in a large population of end-stage liver disease patients who underwent liver transplantation, (2) to describe the underlying clinical and functional characteristics of patients exhibiting EOV, and (3) to investigate the significance of reduced aerobic capacity and EOV in predicting prolonged hospitalization and early in-hospital mortality, and 1-year mortality after liver transplantation in this population.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure
  10. References

Patients

This study was conducted in accordance with the Declaration of Helsinki (http://www.wma.net/en/30publications/10policies/b3/). All patients completed a written informed consent for the procedure and for their medical data to be used in a study. A local institutional review board approved the protocol (CHRU Lille Pole Imagerie et Explorations Fonctionnelles Ethics Committee—approval number UF 5040; 2000-20). The cohort consisted of consecutive patients, who were prospectively studied, as they will undergo liver transplantation at the Lille's University Hospital, France. Liver disease severity was assessed by the MELD score. Ascites were quoted as in the Child-Pugh score: absent, mild-moderate and severe/refractory. Patients with ascites systematically underwent paracentesis the day before cardiopulmonary and exercise tests. Exclusion criteria included congestive heart failure, left ventricular ejection fraction <55% and severe pulmonary hypertension (pulmonary artery systolic pressure >50 mmHg). Eligibility criteria also included the ability to perform cardiopulmonary exercise (CPX) testing, which was stopped for fatigue and/or dyspnea. Exercise testing contraindications included unstable cardiovascular diseases, orthopedic impairment that compromises exercise performance and mental impairment leading to inability to cooperate, such as hepatic encephalopathy, as recommended by the ATS/ACCP [16].

Cardiopulmonary function, quality control and exercise protocols

Transthoracic echocardiography (Sonos 5500; Hewlett Packard, Paris, France) was systematically performed in the preoperative period according to the American Society of Echocardiography guidelines. Right ventricle systolic pressure was assumed to be equivalent to the pulmonary artery systolic pressure and can be calculated from the tricuspid regurgitant velocity using the Bernoulli equation and using right atrial pressure estimated by measurement of the diameter of the inferior vena cava at end-expiration and during sharp inspiration. Spirometry (forced expiratory volume in 1 s [FEV1], forced vital capacity [FVC], total lung capacity [TLC] and diffusing capacity of the lung for carbon monoxide [DL,CO]) was performed according to European Respiratory Society guidelines. Exercise was performed on a bicycle ergometer, and breath-by-breath respiratory gas exchange was measured using a computerized metabolic card (Medisoft ErgoCard, Sorinne, Belgium). Blood pressure was measured manually at rest and every 3 min during incremental exercise and at peak. The electrocardiogram and heart rate were continuously monitored at rest and throughout exercise. Before each test, oxygen (O2) and carbon dioxide (CO2) analyzers and flow mass sensor were calibrated using available precision gas mixture and a 3-L syringe, respectively. During the study period, mean values between qualified replicate tests performed weekly on control subjects were 5.1% ± 4.2%, 5.4% ± 3.2% and 6.1% ± 2.2%, for peak V′O2, carbon dioxide output (V′CO2) and minute ventilation (V′E), respectively. The exercise protocol was standardized: 2 min of rest, 2 min of unload cycling, followed by a progressively increasing work rate of 10 W/min until exhaustion. Arterial blood (240 µL) was sampled at rest and at peak exercise (Microsampler; Roche Diagnostics, Schaffhausen, Switzerland) and immediately analyzed using a blood gas analyzer/co-oximeter (Radiometer ABL700; Radiometer SAS, Neuilly-Plaisance, France).

CPX testing parameters

Breath-by-breath CPX data were measured at rest, warm-up, and incremental exercise testing. V′E, V′O2 and V′CO2 were recorded as concurrent 10-s moving averages, as was determined ventilatory anaerobic threshold (AT) by the V-slope method. V′E/V′CO2 slope was calculated off-line as a linear regression function using 10-s averaged values and excluding the nonlinear part of the relationship after the respiratory compensation point (where nonlinear rise in V′E occurred relative to V′CO2 in the presence of end-tidal PCO2 decrease). In the presence of EOV, the V′E/V′CO2 slope was calculated off-line from 30-s averaged V′E and V′CO2 values due to the scattered distribution of CPX data. Peak V′O2 was measured as the highest average V′O2 over 30 s, as determined from three sequential 10-s periods. Peak O2 pulse was delineated as peak V′O2 divided by peak heart rate and was expressed in mL per beat and as percentage of predicted value by dividing the predicted peak V′O2 by predicted peak heart rate. EOV was visually defined as three or more consecutive cyclic fluctuations of ventilation during unload cycling and exercise. For the definition of EOV, we chose the criteria defined by Leite et al [17]: (1) at least three oscillatory fluctuations in V′E during warm-up and exercise; (2) regular oscillations as defined by a standard deviation of three consecutive cycle length durations (time between two consecutive nadirs) within 20% of the average; (3) a minimal average ventilation amplitude of >5 L, defined as peak V′E of one oscillation minus the average of two adjacent nadirs. Patient effort was considered to be maximal if two of the following occurred: predicted maximal work is achieved, predicted maximal heart rate is achieved, V′E/V′O2 > 45, lactate level > 6 mmol L−1, respiratory exchange ratio (RER) > 1.10 and pH drop > 0.06, as recommended by the ATS/ACCP [16].

Endpoints

Patients were followed up through in-hospital and outpatient medical chart records at the reference center. Follow-up of those who did not attend their scheduled appointments was obtained by phone call interview of the patient, patient's relatives or primary care physician. All-cause 1-year mortality after liver transplantation was used as primary endpoint. The primary composite endpoint was built up as 1-year mortality and/or prolonged hospitalization (up to 28 days) and early in-hospital mortality. Results of the CPX testing did not influence organ allocation priority, and perioperative and postoperative care during the study period.

Statistics

Results are expressed as means ± SD or median (25th–75th percentile). Student's t-test or Mann–Whitney U-test for nonpaired values was used to compare means of groups for quantitative variables, assuming or not Gaussian distribution, respectively. For qualitative variables, chi-square test or Fisher's exact test was employed. Kaplan–Meier survival curves were constructed, and log-rank tests were performed to determine statistical significance between dummy variables in Kaplan–Meier analysis. Prognosis value of variables was determined using univariate and multivariable Cox proportional hazards regression analysis. First, variables included in the univariate Cox regression model were age, MELD and exercise variables (peak V′O2, O2 pulse, V′E/V′O2, V′E/V′CO2 and EOV). Variables that showed a significant association with the outcome at univariate Cox regression analysis (p < 0.20) were then included in the multivariable Cox regression analysis. Level of statistical significance was set at a two-tailed p-value < 0.05. Calculations were performed using software (SAS version 9.3; SAS Institute, Inc., Cary, NC).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure
  10. References

Preoperative and intraoperative characteristics of patients according to outcome

During the study period (January 2001 to December 2011), our institution performed 436 liver transplantations. Patients hospitalized in intensive care unit immediately before transplantation (70), transplanted for acute/fulminant liver failure (32) or combined kidney–liver transplantation (26), re-transplanted [18], screened outside our institution [17], and pediatric recipients [5] were not enrolled. Hence, our study population consisted of 268 patients with end-stage liver disease (with or without hepatocellular carcinoma) who will undergo liver transplantation. No patients were excluded on the basis of cardiac failure, EF < 55%, portopulmonary hypertension or encephalopathy. Five additional patients were excluded from the study as they failed to handle ergometer cycling.

Patients' demographics and clinical characteristics (n = 263) according to outcome are given in Table 1. Waiting time on list before transplantation was 2.0 [1-4] months (median [25th–75th percentile]). Importance of ascites scored as in Child-Pugh classification did not differ between survivors and nonsurvivors (χ2 model 1.51; p = 0.47). Pretransplantation values of mean arterial pressure (86 ± 14 mmHg vs. 87 ± 15 mmHg; p = 0.36), hemoglobin (11.5 ± 2.4 mg/dL vs. 11.9 ± 2.4 mg/dL; p = 0.47), albumin (3.4 ± 0.6 mg/dL vs. 3.5 ± 0.8 mg/dL; p = 0.48) and sodium (136 ± 11 mmol/L vs. 135 ± 5 mmol/L; p = 0.69) concentrations did not differ between survivors and nonsurvivors.

Table 1. Clinical characteristics of patients according to mortality after liver transplantation
 All patientsSurvivorsNonsurvivorsp-Value
  1. COPD, chronic obstructive pulmonary disease; MELD, Model for End-Stage Liver Disease; n, number; %, percent.

  2. Waiting time for transplantation and from exercise testing to transplantation, months.

  3. Results are expressed as mean ± SD, except for “Waiting time for transplantation and from exercise testing to transplantation” which are expressed as median (25th–75th).

Number of patients263243 (92.4%)20 (7.6%) 
Age, year58.8 ± 8.558.5 ± 8.660.7 ± 8.20.27
Body mass index25.3 ± 4.525.4 ± 4.224.8 ± 5.50.55
Male/female gender, n198/65183/6015/51.00
Etiology of liver disease n, %
Alcoholic cirrhosis187 (71%)172 (70%)15 (75%)0.80
Viral cirrhosis50 (19%)46 (19%)4 (20%)1.00
Hepatocellular carcinoma102 (39%)94 (39%)8 (40%)1.00
MELD score14.8 ± 1.914.9 ± 2.014.4 ± 2.00.28
History of tobacco use174 (66%)161 (66%)13 (65%)1.00
History of COPD81 (31%)75 (31%)6 (30%)1.00
History of heart disease
Ischemic34 (13%)31 (13%)3 (15%)0.73
Hypertensive60 (23%)55 (23%)5 (25%)0.78
Waiting time for transplantation2.0 (1–4)2.0 (1–4)2.0 (1–4)0.38
Waiting time from exercise testing4.5 (3–7)4.0 (3–7)5.0 (3–7)0.50

Relative functional testing time to transplant was 4.5 [3-7] months (median [25th–75th percentile]). Patients with a relative CPX time to transplant up to 8 months had hepatocellular carcinoma, which was considered stable over this period of time. Left ventricle ejection fraction was similar in survivors and nonsurvivors (Table 2). Lung function tests revealed similar values in survivors compared with nonsurvivors (Table 2). Objective criteria of maximal exercise (ATS/ACCP 2003 [[16]]) were achieved in all patients, which avoided inappropriate prognostic stratification due to poor patient motivation. Peak V′O2 and O2 pulse were significantly greater in survivors compared with nonsurvivors (Table 2). AT was not detectable in 13 (5%) survivors and 8 (40%) nonsurvivors (p = 0.0001). EOV was detected in 27 patients (10%), prevalence of which was higher in nonsurvivors. Figure 1 displays ventilation pattern of negative (A) and positive (B) EOV patients.

Table 2. Baseline and exercise cardiopulmonary functional tests of the patients before liver transplantation
PatientsAll patients (n = 263)Survivors (n = 243)Nonsurvivors (n = 20)p-ValueEOV negative (n = 236)EOV positive (n = 27)p-Value
  • AT, ventilatory anaerobic threshold; DL,CO, diffusing capacity of the lung for carbon monoxide (mL min−1 mmHg−1; % predicted); EOV, exercise oscillatory ventilation; FEV1, forced expiratory volume in 1 s (L, liter; % predicted); FEV1/FVC, (% of predicted); FVC, forced vital capacity (L, liter; % predicted); lactate, arterial lactate concentration; MAP, mean arterial pressure; MELD, Model for End-Stage Liver Disease; NA, not applicable; P(A-a),O2, alveolar-arterial O2 partial pressure gradient; Pa,CO2, arterial CO2 partial pressure; Pa,O2, arterial O2 partial pressure; peak, peak exercise; RER, respiratory exchange ratio; TLC, total lung capacity (L, liter; % predicted); VD/VT, physiological dead space/tidal volume ratio; V′CO2, carbon dioxide output; V′E, minute ventilation; V′O2, oxygen uptake; W, watt.

  • Ventilatory reserve (%) is calculated using the following equation (MVV − peak V′E)/MVV × 100, where peak V′E is the minute ventilation measured at peak exercise and MVV (maximal voluntary ventilation) is calculated as FEV1 multiplied by 35.

  • Results are expressed as mean ± SD (number of patients, % percent).

  • *

    p < 0.05.

Age, year58.8 ± 8.558.5 ± 8.660.7 ± 8.20.2758.6 ± 8.661.6 ± 7.60.08
MELD score14.8 ± 1.914.9 ± 2.014.4 ± 2.00.2814.6 ± 3.515.8 ± 4.10.09
Use of β-blockers132 (50%)122 (50%)10 (50%)1.00126 (53%)15 (55%)0.25
MortalityNA92.4%7.6% 7 (3%)13 (43%)*0.01
Echocardiography
LVEF, %67 ± 467 ± 566 ± 50.3967 ± 566 ± 30.31
LAD, mm41 ± 441 ± 442 ± 30.2742 ± 541 ± 40.39
LVEDD, mm46 ± 447 ± 446 ± 30.2846 ± 547 ± 40.38
PASP, mmHg28 ± 528 ± 527 ± 50.3927 ± 527 ± 31.00
Resting lung testing
FEV1, L2.8 ± 0.82.9 ± 0.82.8 ± 0.60.632.9 ± 0.82.5 ± 0.8*0.01
FEV1, %95 ± 1195 ± 1194 ± 110.6396 ± 1188 ± 9*0.01
FVC, L3.7 ± 0.83.7 ± 0.93.6 ± 0.80.653.7 ± 0.93.3 ± 0.9*0.02
FVC, %97 ± 1197 ± 1196 ± 110.6597 ± 1089 ± 11*0.02
FEV1/FVC, %76 ± 679 ± 578 ± 30.3778 ± 1177 ± 90.65
TLC, L5.1 ± 1.25.2 ± 1.15.1 ± 1.10.595.3 ± 1.14.7 ± 1.1*0.01
TLC, %100 ± 11102 ± 1198 ± 110.59106 ± 1189 ± 11*0.01
DL,CO, mL min−1 mmHg−123.8 ± 4.022.8 ± 4.024.2 ± 3.70.1123.7 ± 2.723.1 ± 3.00.29
DL,CO, %77 ± 1077 ± 1178 ± 90.1180 ± 979 ± 80.29
Exercise testing
Peak workload, W93 ± 2797 ± 2784 ± 30*0.0495 ± 2779 ± 30*0.01
Peak V′O2, mL kg−1 min−118.5 ± 5.018.6 ± 2.817.1 ± 3.3*0.0218.8 ± 4.915.8 ± 4.3*0.01
Predicted peak V′O2, %64 ± 1964 ± 1955 ± 12*0.0465 ± 1955 ± 15*0.01
AT V′O2, mL kg−1 min−112.2 ± 3.412.4 ± 3.511.9 ± 2.70.5312.5 ± 3.410.0 ± 2.0*0.01
ΔV′O2/ΔW, mL min−1 W−111.0 ± 2.311.0 ± 2.411.5 ± 1.80.3611.0 ± 2.310.3 ± 2.40.13
RERpeak1.23 ± 0.141.23 ± 0.131.23 ± 0.151.001.22 ± 0.131.23 ± 0.130.70
V′E/V′O2 peak48.1 ± 12.248.0 ± 12.348.2 ± 11.30.9447.4 ± 12.053.4 ± 12.8*0.01
V′E/V′CO2 peak32.1 ± 13.228.0 ± 12.333.2 ± 13.30.0734.4 ± 15.032.4 ± 13.00.51
V′E/V′CO2 slope32.2 ± 13.432.3 ± 13.231.2 ± 14.60.7232.0 ± 3.134.1 ± 4.9*0.01
Ventilatory reserve peak, L41 ± 2343 ± 2543 ± 271.0037 ± 2034 ± 220.50
Ventilatory reserve peak, %37 ± 1737 ± 1940 ± 150.4939 ± 1436 ± 170.37
VD/VT peak0.26 ± 0.070.26 ± 0.070.27 ± 0.060.530.26 ± 0.070.28 ± 0.070.16
Pa,O2 peak, mmHg102 ± 15103 ± 16103 ± 141.00101 ± 14105 ± 150.22
P(A-a),O2 peak, mmHg20.8 ± 13.021.0 ± 13.120.0 ± 12.40.7420.8 ± 13.221.5 ± 11.10.79
Pa,CO2 peak, mmHg32.4 ± 5.032.2 ± 5.133.2 ± 5.10.4032.7 ± 5.130.1 ± 3.9*0.01
Heart ratepeak, beat min−1124 ± 24124 ± 25127 ± 210.60125 ± 24121 ± 270.42
O2 pulsepeak, mL beat−111.2 ± 3.611.3 ± 2.710.0 ± 2.8*0.0411.4 ± 3.69.4 ± 3.0*0.01
O2 pulsepeak, %88 ± 2594 ± 2881 ± 14*0.0492 ± 2282 ± 24*0.03
Arterial pHpeak7.38 ± 0.057.38 ± 0.057.38 ± 0.061.007.38 ± 0.057.41 ± 0.05*0.01
Lactatepeak, mmol L−16.0 ± 2.16.0 ± 2.05.8 ± 2.40.676.2 ± 2.14.7 ± 1.4*0.01
MAPpeak, mmHg105 ± 19105 ± 18106 ± 200.81106 ± 19100 ± 160.11
EOV, n (%)27 (10%)21 (9%)6 (30%)*0.01NANA 
image

Figure 1. Representative minute ventilation (V′E) recordings at rest, warm-up and incremental exercise (10 W min−1) in a patient with normal ventilatory pattern (A) and EOV (B) are displayed.

Download figure to PowerPoint

Survivors and nonsurvivors had similar intraoperative management history in terms of blood loss (2.4 ± 1.4 L vs. 1.9 ± 1.3 L, respectively; p = 0.12) and transfusion of packed red blood cells (5.3 ± 3.9 units vs. 5.7 ± 4.2 units, respectively; p = 0.67) and platelets (1.27 ± 2.63 units vs. 0.16 ± 0.38 units, respectively; p = 0.17). No fresh frozen plasma was used. Quality of living donor organ was similar in survivors and nonsurvivors in terms of donor age (42.8 ± 15.3 L vs. 46.1 ± 13.2 L, respectively; p = 0.34), cerebrovascular cause of death (15% in both groups) and duration of cold ischemia (8.5 ± 2.6 h vs. 8.2 ± 2.7 h, respectively; p = 0.61).

Characteristics of patients with oscillatory ventilation

Comparisons of main variables between EOV-positive and EOV-negative patients are listed in Table 2. Age and MELD score tended to be higher in EOV positive patients but failed to reach significance. Importance of ascites was higher in patients with EOV as compared to those without (χ2 model 9.27; p = 0.01). Left ventricle ejection fraction was similar in EOV-positive and EOV-negative patients (Table 2). Occurrence of EOV did not differ regarding etiology of liver failure (data not shown), age, use of β-adrenergic blocking agents, or MELD (Table 2). Lung function tests displayed reduced values of FEV1, FVC and TLC in patients with EOV as compared to those without (Table 2). Patients with EOV, as compared to those without, had reduced peak exercise (peak V′O2) and O2 pulse (Table 2). Due to the scattered distribution of breath-by-breath values related to oscillatory ventilation pattern, AT was not detectable in 9 patients (33%) with EOV and 12 patients (5%) without EOV (p = 0.0001). Whereas alveolar-arterial O2 partial pressure gradient was not different, peak arterial CO2 partial pressure (Pa,CO2) was lower in patients with EOV compared to those without (Table 2).

Follow-up and Cox regression analysis for primary endpoints

No patient was unavailable for the follow-up period. There was no intraoperative mortality. Nine patients (two patients without EOV and seven patients with EOV) died from septic shock in intensive care unit within the first 30 days after liver transplantation. Six patients (three patients without EOV and three patients with EOV) died from acute or late acute graft rejection within 180 days, and five patients (two patients without EOV and three patients with EOV) died from hepatocellular carcinoma recurrence within the last month of the 1-year follow-up.

Using univariate Cox regression analysis, EOV was the only variable that emerged as significant predictor of 1-year mortality (Table 3), whereas peak V′O2, O2 pulse and EOV were significant prognostic predictors of the primary composite endpoint. Multivariate Cox proportional hazard analysis revealed EOV as the only independent predictor of 1-year mortality (p = 0.02; Table 3). Regarding the primary composite endpoint, peak V′O2 was identified as a predictor factor, whereas EOV failed to reach statistical significance (p = 0.09; Table 3). Kaplan–Meier analysis relating to EOV is illustrated in Figure 2. Patients with EOV had a significantly higher total mortality rate than those without EOV (χ2, 7.46; p = 0.006).

Table 3. Cardiopulmonary parameters related to mortality and primary composite endpoint at univariate and multivariate Cox regression analysis
 χ2p-ValueHR95% CI
A: Primary endpoint (mortality)
Univariate
Age2.050.151.040.985–1.103
MELD0.810.370.960.890–1.044
Peak V′O20.890.340.950.869–1.050
O2 pulse2.770.090.880.771–1.021
V′E/V′O20.050.821.000.969–1.041
EOV6.570.013.491.342–9.091
Multivariate
EOV3.960.042.731.016–7.354
O2 pulse1.900.160.900.781–1.044
Age2.020.151.0420.985–1.103
 χ2p-ValueHR95% CI
  1. CI, confidence interval; EOV, exercise oscillatory ventilation; HR, hazard ratio; MELD, model for end-stage liver disease; V′O2, oxygen uptake; V′E, minute ventilation; χ2, model chi-square2.

  2. Primary composite endpoint refers to 1-year mortality and/or prolonged hospitalization and early in-hospital mortality. Variable units are as previously displayed.

B: Primary composite endpoint
Univariate
Age1.090.2951.010.987–1.043
MELD2.550.1101.030.993–1.067
Peak V′O27.750.0050.930.884–0.979
O2 pulse5.490.0190.920.858–0.986
V′E/V′O23.380.0661.020.999–1.035
EOV5.630.0172.021.130–3.619
Multivariate
Peak V′O24.280.040.940.893–0.997
EOV2.790.091.670.915–3.036
MELD0.290.581.010.972–1.051
image

Figure 2. Plots of Kaplan–Meier survival curves in patients with exercise oscillatory ventilation (EOV) and without it (no EOV). The difference in survival between the two groups was statistically significant (χ2, 7.46; p = 0.006).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure
  10. References

We have investigated the preoperative suitability of peak V′O2 and EOV to predict perioperative complications, early in-hospital mortality and 1-year mortality in end-stage liver disease patients after liver transplantation. Novel findings of our study are twofold. First, exertional oscillatory ventilation may be observed in end-stage liver disease patients (about 10%). Second, whereas reduced peak V′O2 was able to predict the primary composite endpoint defined as 1-year mortality and/or perioperative complications and early in-hospital mortality, the presence of EOV identified candidates with an increased 1-year mortality rate after liver transplantation.

Identification of those patients with high priority and opportuneness for liver transplantation is a major challenge in transplantation medicine. The MELD score has been proposed for proper allocation of grafts for liver transplantation and its use decreases time to transplant and reduction in waiting list mortality [2-5]. However, no concomitant changes related to MELD utilization have been demonstrated in posttransplant outcome [3, 18]. Our study confirms that preoperative MELD score was unable to predict perioperative complications, early in-hospital mortality and longer-term mortality after liver transplantation. Limited but promising evidence suggests that impairment of preoperative aerobic capacity can prognosticate liver transplant postoperative course and early survival [6-9]. For example, the 6-min walk distance was significantly reduced in patients listed for liver transplantation and a preoperative distance less than 250 m was a risk for death on the waitlist [7]. Consistently, our group reported that liver transplant candidates with peak exercise V′O2 less than 60% of the predicted value had a lower survival rate than those who had peak V′O2 greater than 60% predicted [8]. Yet these data were not separated for those patients undergoing transplantation [7, 8]. More recently, determination of V′O2 at AT was proposed as an accurate predictor of early survival after liver transplantation [9]. In the present study, reduced peak V′O2 was clearly predictive of 1-year mortality and/or perioperative complications and early in-hospital mortality composite endpoint, but failed to predict 1-year mortality after liver transplantation. In addition, peak O2 pulse, an indirect index of left ventricle performance during exercise, was lower in nonsurvivors, but univariate analysis failed (p = 0.09) to associate impaired O2 pulse with perioperative complications plus early in-hospital mortality or longer-term mortality. Overall, our results, obtained in a large series of liver transplanted patients, support that preoperative peak V′O2 represents a valuable tool to predict perioperative complications and early in-hospital mortality. This finding may be relevant because impairments of peak V′O2 and O2 pulse may identify high-risk patients who would benefit from aggressive postoperative supporting care.

Consistently with its strong prognostic value in heart failure [10, 19], we found that EOV was an independent and valuable prognostic indicator of 1-year mortality after liver transplantation. Proposed mechanisms of ventilation instability leading to EOV include prolonged circulatory time that delays homeostatic reflex information transfer (time lag), abnormal response of central and peripheral controllers that drive ventilation (controller gain), and efficiency impairment of the gas exchange system to damp out changes in arterial gas tensions (controlled plant gain) [20-23]. Clearly, EOV cannot be due to cardiac output decrease (i.e. time lag) in patients with advanced liver diseases, which typically present with hyperdynamic circulatory. In line, patients with EOV had preserved or even increased left ventricle ejection fraction at rest and normal range O2 pulse (a surrogate of left ventricle ejection volume) at peak exercise, suggesting no major cardiomyopathy. Alternatively, mechanisms of EOV may include enhanced chemoreceptor sensitivity (i.e. controller gain), which may favor hyperventilation and ventilation instability in cirrhotic patients [13, 15]. Consistently, patients with EOV had excessive rise in V′E relative to V′O2 and V′CO2, and lower Pa,CO2 at peak exercise, suggesting inadequate ventilatory drive [20-24]. Hyperventilation observed in EOV patients was not due to increased physiological dead space/tidal volume ratio (VD/VT), reduced DL,CO or early onset of lactic acidosis during exercise [20, 22]. Hence, it is likely that these patients may have elected a lower set-point for Pa,CO2 at peak exercise. As an eventual cause of ventilation instability and EOV, we reasoned that efficiency of the gas exchange system to damp out changes in arterial gas tensions (i.e. controlled plant gain) might be impaired as the result of reductions in lung volumes [20, 22, 24]. Decreases in lung volumes were attributed to the compressive effects of ascites [25], which were found more severe in patients with EOV, despite equivalent volume of ascite drainage the day before functional testing. Whether ascites would develop more rapidly after paracentesis in EOV patients or other mechanisms of lung volume restriction may be present in EOV patients warrants further studies.

Study limitations

Our study results were confined to a population of end-stage liver disease patients who all underwent liver transplantation. A limitation of the study is the small number of deaths at 1 year in this population; however, the consistent findings regarding the primary composite endpoint (1-year mortality and/or perioperative complications and early in-hospital mortality) somewhat balance this limitation. The findings also confined to patients who were able to perform maximal exercise testing that excluded patients with poor physiological conditions. This was related to our objective that focused on the preoperative value of aerobic capacity and EOV to predict postoperative complications and outcome. Although the present study included a large number of patients, sample size of patients with EOV was small, which is a weakness of our report; nonetheless, both follow-up duration and substantial event rate authorize a reasonable interpretation of EOV prognosis significance. Different arbitrary definitions of EOV exist in the literature. Arbitrary use of specific EOV definition may lead to overestimation or underestimation of EOV prevalence in the studied population. We defined EOV according to suggestions from the literature that can be considered more liberal than those used by other groups. The observation that such liberal classification led to a high statistical significant prediction underscores the pathophysiological relevance of this abnormal ventilatory pattern. Our present study lacks a formal test–retest analysis for EOV reproducibility, which was difficult to perform due to poor functional and psychological patient conditions and short waiting time until transplantation. However, our specified intention was to assess the prognosis value of variables measured during CPX testing that are considered gold standards in preoperative risk stratification.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure
  10. References

In cirrhotic patients, preoperative aerobic capacity impairment and EOV delineated a high-risk population in liver transplant recipients. Reduced peak V′O2 and EOV may be considered a valuable guide in the management of patients with liver dysfunction and should suggest a more aggressive medical and surgical treatment policy when detected.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure
  10. References

Pr. Remi Neviere is the guarantor of the content of the manuscript, including the data and analysis. Pr. Remi Neviere, Dr. Jean Louis Edme, Dr. David Montaigne, Pr. Francois R. Pruvot, Emmanuel Boleslawski and Pr. Sebastien Dharancy take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. Dr. Jean Louis Edme is PhD in statistics and was responsible for statistical analyses (Department of Biostatistics of Lille University, France).

Disclosure

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure
  10. References

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure
  10. References