REVERSIBILITY OF LUNG COLLAPSE AND HYPOXEMIA IN EARLY ACUTE RESPIRATORY DISTRESS SYNDROME AUTHORS: BORGES, JOÃO B. MD OKAMOTO, VALDELIS N. MD MATOS, GUSTAVO F. J. MD CARAMEZ, MARIA P. R. MD ARANTES, PAULA R. MD BARROS, FABIO MD SOUZA, CIRO E. MD VICTORINO, JOSUÉ A. MD KACMAREK, ROBERT M. PhD BARBAS, CARMEN S. V. MD CARVALHO, CARLOS R. R. MD AMATO, MARCELO B. P. MD Online Data Supplement
Methods: 1) Rationale for the use of the index PaO 2 + PaCO 2 400 mmhg: According to the alveolar gas equation, increments of PCO 2 at the alveolar space are expected to decrease the alveolar PO 2 in an approximate 1:1 ratio especially at 100% FIO 2. In a paper of Joyce et al (E1) the authors developed a theoretical mathematical model to investigate the effects of permissive hypercapnia on pulmonary gas exchange when simulating different levels of intrapulmonary shunt. The ultimate consequences on arterial blood-gases were investigated. According to their constant metabolism model, whenever the arterial PCO 2 increases, there is a linear decrease in alveolar content of O 2 caused by the increasing alveolar content of CO 2. The net result is a fall in arterial PO 2, in proportion to this increment in PaCO 2, but dependent on baseline shunt levels. For shunt levels around 5-10%, increments in arterial PCO 2 cause a decrease in arterial PO 2 in an almost 1:1 ratio. For higher shunt levels (> 10%) the effects on PaO 2 are more complex. Despite the necessary decrease in the alveolar content of oxygen, there is a stronger influence of the rightward shift of the oxygen-hemoglobin dissociation curve, which may result in a higher mixed venous PO 2. This latter effect may supplant the deficient alveolar oxygenation, ultimately resulting in increased arterial PO 2. These relationships were tested by our group in a pilot experiment in rabbits, revealing that, when using 100% oxygen, any degree of hypercapnia (produced by the addition of a dead space in the ventilator circuit) was associated to a decreased arterial PO 2, in proportion to the degree of hypercapnia. When dealing with low
levels of shunt (<10%), increments of 1 mmhg in arterial PCO 2 were usually associated to a decrement of 1.2 mmhg in the arterial PO 2. Transposing these considerations to our patients, we reasoned that some of our patients, although maximally recruited, might present intermediate levels of PaO 2 just because they were submitted to an associated permissive hypercapnia. By looking only at PaO 2 values, one could conclude that such patient was not maximally recruited. By considering the total sum, however, this underestimation could be avoided, since the increments in the arterial PCO 2 would cause a proportional fall in arterial PO 2, but with an almost constant sum. Generally speaking, a high PaCO 2 may indicate loss of functional alveolar units, especially when minute ventilation is normal/high and there is no permissive hypercapnia involved. During permissive hypercapnia, however, usually with low minute ventilation associated, the high PaCO 2 may just reflect the physician s ventilatory strategy, precluding one to observe a higher PaO 2 and creating a scenario where the degree of lung recruitment might be underestimated. FIGURE E5 illustrates the concepts discussed above, demonstrating the practical implications of such concept in rabbits submitted to lung lavage and permissive hypercapnia. 2) Complete Inclusion/Exclusion criteria for the patients: Inclusion: a) Need of mechanical ventilation and predisposing condition for ALI/ARDS. b) PaO 2 /FIO 2 < 300, at PEEP = 10 cmh 2 O and V T = 6-8 ml/kg.
c) Wedge pressure S 18 mmhg d) Stable doses of vasopressors, with mean arterial blood pressure > 65 mmhg and a stable arterial lactate level over the preceding 6 hours e) Mechanical ventilation S 7 days Exclusion: a) Chronic pulmonary disease, previous lung resection or active broncho-pleural fistulae b) Contraindications to sedation or paralysis c) Suspicion or clinical signals of elevated intracranial pressure d) Life-threatening arrhythmias 3) Respiratory monitoring: Airway flow was measured by a heated pneumotachometer (Hans Rudolph ; 0-160 L/min. Kansas City, Mo) connected to a differential pressure transducer (MP-45; ± 2 cmh 2 O, Validyne Co. Northridge CA); volume was determined by digital integration of the flow signal. Differential pressure transducers (Validyne MP-45; ± 100 cmh 2 O) measured airway opening pressure. Changes in end-expiratory lung volume were continuously assessed by inductive plethysmography (Respitrace, Ambulatory Monitoring Inc. Ardsley, New York).
4) Hemodynamic management: Fluid status was previously optimized according to a pre-defined protocol based on pulse-pressure variation (E2) conducted right before the recruiting maneuver and the baseline measurements: vasopressors (norepinephrine or dobutamine, depending on the case) were kept at a constant dose and volemia was maximized (colloid infusions up to 2L) till measuring a pulse pressure variation < 13 %, during a standardized mechanical ventilation period with PEEP = 10 cmh 2 O and V T = 8 ml/kg. We ignored the values for wedge pulmonary-artery pressure or central venous pressure. 5) Calculations of threshold opening pressures for FIGURE #3 We basically adopted the same procedure used by Crotti et al, but with a small difference (E3). In our study, the CT data for constructing the recruitment vs. pressure curves was not taken during an inspiratory pause, but taken at endexpiration, at PEEP levels of 5 and 25 cmh 2 O. Data from PEEP = 5 cmh 2 O was considered as the situation where we had not yet observed any recruitment (zero % of the potential-for-recruitment), whereas the data at the last step of the maximumrecruitment was considered as the situation where we observed 100% of the recruitment process. Additionally, all scans taken at PEEP = 25 cmh 2 O were considered as a reliable representation of the effects of the preceding inspiratory pressures. It means that threshold opening pressures were related to the previous plateau pressure, and not to current PEEP. This assumption was based on the observation that all eleven patients presented closing pressures (detected during the PEEP titration phase) well below 25 cmh 2 O. Therefore, we reasoned that, whenever
airway pressures exceeded the threshold-opening-pressure of a lung unit during inspiration, this unit would remain open at end-expiration, being detected in the expiratory CT scan. Consequently, the amount of recruited tissue at a certain recruitment step was related to the maximum inspiratory pressure observed at that step or just before, and not to its end-expiratory pressure. Recruitment pressure curves (from zero to 100% recruitment, according to the changes in the mass of collapsed tissue) were fitted with a sigmoid function ( y = a/{1+exp[-(x-x 0 )/b]}), where a corresponds to the vital capacity, b is the parameter proportional to the pressure range within which most of the volume change takes place, and x 0 is the pressure at the inflection point of the sigmoidal curve (where curvature changes sign), according to Venegas and coworkers (E4). Threshold opening pressures were then defined as the pressures at which new increment of recruitment was observed. The data was derived (using 1 cmh 2 O pressure intervals) from the fitted recruitment pressure curve obtained in each patient. Thus, data are not strictly experimental, but an estimate of threshold opening pressures according to the Gaussian function (y = a x exp { - 0.5 [(x x o ) / b ] 2 }) derived form the original sigmoid equation. The average curve for all patients was then constructed, with data expressed as mean ± SEM. 6) Details on CT analysis: A quantitative analysis of CT numbers of lung tissue was accomplished in two stages. In the first one, we employed the Osiris Medical Imaging Software program (version 3.6) (University Hospital of Geneva, available in the web: www.expasy.ch/uin). This software generated the histograms of CT numbers
distribution for each region of interest manually determined. In the second stage, the histograms data were analyzed in a customized program written in LabVIEW 5.1.1. The global accuracy of this procedure was previously validated in vitro and in vivo: a) we checked the internal volume of a supersyringe, with errors < 2%, and b) in an experimental study with rabbits, we checked the calculated weight gain as a result of the instillation of a known amount of liquid into the lung, also with errors < 2%. For each region of interest, we manually drew an internal outline in each hemithorax, excluding the chest wall, the mediastinum, pleural effusion and areas with partial volume effect. Iterative changes in the density window were tested to identify areas with partial volume effect. The histogram data of regions of interest was analyzed to generate the CT parameters. These parameters were extracted from the quantitative analysis of CT numbers representing densities of lung tissue. This analysis is based on quasi linear relationship between X-ray attenuation in a voxel and his physical density, i.e., the ratio of mass to the volume (E5, E6). The X-ray attenuation is expressed in CT numbers Hounsfield units (HU), representing the percent of radiation absorbed for each lung voxel. As the tissue density is near the water density, the relationship between the physical density and CT number, in any lung region of interest may be expressed as (assuming the specific weight of the tissue is equal to 1): volume gas mean CT number observed ------------------------------------ = ------------------------------------------- (Equation E1) (volume gas + volume tissue ) (CT number gas CT number water ) Rearranging Equation E1, it is possible to compute for any lung region of interest (contiguous voxels), in which total volume is known, the volume of gas, the volume
(and the weight) of tissue, and the gas/tissue ratio (E6). The mass of each voxel, representing his parenchymal mass, can be calculated as: Mass VOXEL = (CT number + 1000) x Volume VOXEL / 1000 (Equation E2) The Osiris software provides the histograms of CT numbers for each region of interest, grouping voxels less than 1 HU appart. By knowing frequency distribution of CT numbers in a region of interest, as well as his total volume, it is possible estimate the amount of tissue or parenchymal mass. In this study, we separately calculated the mass changes and volume changes for all lung compartments.
REFERENCES (ONLINE DATA SUPPLEMENT) E1. Joyce, C. J., and K. G. Hickling. 1996. Permissive hypercapnia and gas exchange in lungs with high Qs/Qt: a mathematical model. Br J Anaesth 77(5):678-83. E2. Michard, F., D. Chemla, C. Richard, M. Wysocki, M. R. Pinsky, Y. Lecarpentier, and J. L. Teboul. 1999. Clinical use of respiratory changes in arterial pulse pressure to monitor the hemodynamic effects of PEEP. Am J Respir Crit Care Med 159(3):935-9. E3. Crotti, S., D. Mascheroni, P. Caironi, P. Pelosi, G. Ronzoni, M. Mondino, J. J. Marini, and L. Gattinoni. 2001. Recruitment and derecruitment during acute respiratory failure: a clinical study. Am J Respir Crit Care Med 164(1):131-40. E4. Venegas, J. G., R. S. Harris, and B. A. Simon. 1998. A comprehensive equation for the pulmonary pressure-volume curve. J Appl Physiol 84(1):389-95. E5. Drummond, G. B. 1998. Computed tomography and pulmonary measurements. Br J Anaesth 80(5):665-71. E6. Gattinoni, L., P. Caironi, P. Pelosi, and L. R. Goodman. 2001. What Has Computed Tomography Taught Us about the Acute Respiratory Distress Syndrome? Am J Respir Crit Care Med 164(9):1701-11.
LEGENDS FOR FIGURES (ONLINE DATA SUPPLEMENT) FIGURE E1: Bivariate plot showing the relationship between two CT-estimates of lung collapse: the new proposed variable (percent-mass of collapsed tissue in the Y axis) versus the traditional variable used in previous studies (percent-volume of collapsed tissue in the X-axis). As demonstrated, the difference between both variables is larger for intermediate values of lung collapse FIGURE E2: Comparative plot of the two variables to assess lung collapse during critical steps in our protocol. Note that the percent-volume variable underestimated the amount of collapse (estimated as mass-percent) and probably the pulmonary shunt. FIGURE E3: Evolution of total lung mass along the recruitment protocol. As expected, the total lung mass (including the four compartments, collapsed, poorly-aerated, normally aerated and hyperinflated) did not change along the protocol. This suggest that there was no significant migration of blood or fluid out of the lung during the recruitment procedure. Absolute values represent an underestimation of true values (approximately 80%) because of our methodology: in order to avoid partial volume effects, the regions of interest kept a safety margin (approximately 5 mm) from the chest-wall, heart and diaphragm. This means that the periphery of the lung, close to visceral pleura, was systematically excluded. FIGURE E4: Evolution of the total amount of gas inside the thorax, before and after the recruitment protocol. Note that there was a large increment (1.5L) in the total amount of gas, despite the constant lung mass (FIGURE E3 above), demonstrating that the lung tissue was redistributed during the recruitment process.
FIGURE E5: Symmetrical relationship between arterial PO 2 and arterial PCO 2 during permissive hypercapnia in a pilot study performed in 12 rabbits. The study was conducted to demonstrate the theoretical model of Joyce et al (E1). The animals were previously recruited, maintaining shunt levels < 10%. Under this circumstance, any increase in arterial PCO 2 was accompanied by a symmetrical decrement in arterial PO 2, in an almost 1:1 ratio. FIGURE E6: ROC (receiving operator characteristic) curve for the PaO 2 + PaCO 2 index, expressing its accuracy to predict the presence of lung collapse. Lung collapse was transformed in a dichotomous variable: < 5% (absence) or 5% (presence) of lung collapse (according to the percent-mass calculations). Sensitivity and specificity for some critical thresholds for the PaO 2 +PaCO 2 sum were displayed (330, 400 and 560 mmhg). FIGURE E7: Evolution of quasi-static compliance (calculated by multiple linear regression) during the recruitment steps. Please, note that compliance calculations were always performed at the same PEEP level (25 cmh 2 O) and at the same driving pressure (15 cmh 2 O). Each step, represented below, was preceded by progressively higher values of plateau pressure, as indicated (60/25 means a previous plateau pressure of 60 cmh 2 O, followed by a current PEEP = 25 cmh 2 O). The 15% increment in static compliance suggests the recruitment of additional 15% of lung units. FIGURE E8:
Individual correlation (within-patient regression) between arterial PO 2 and static compliance during the recruitment steps presented in the FIGURE E7 above. Note that, for each individual patient, increments in compliance were followed by proportional gains in oxygenation, with a positive slope. Only the two patients that could not be recruited (marked with dashed red circles) did not present a positive slope. FIGURE E9: Evolution of functional residual capacity (referenced to the baseline period) calculated according to the inductance plethysmography, along the recruitment steps. FRC was always measured at the same PEEP level (25 cmh 2 O) and at the same driving pressure (15 cmh 2 O). Each step, represented below, was preceded by progressively higher values of plateau pressures, as indicated (60/25 means a previous plateau pressure of 60 cmh 2 O, followed by a current PEEP = 25 cmh 2 O). FIGURE E10: Hemodynamic consequences of the maximum-recruitment strategy in the eleven patients submitted to CT scanning. Cardiac index was continuously measured with the Vigilance device. Note that the slight reduction of cardiac-index is completely reversed afterwards, during optimum-peep conditions.
Results: a) Most extreme values for blood-gases during the protocol (see TABLE 3 in the main text): TABLE E1: variable baseline OLA Step-1 Step-2 Step-3 Step-4 Step-5 Titrated PEEP PCO 2 Max 84 96 116 132 134 149 147 92 Min 28 48 29 29 28 56 56 30 Arterial-pH Max 7.39 7.33 7.38 7.38 7.38 7.20 7.12 7.33 Min 6.96 6.96 6.94 6.91 6.87 6.85 6.84 * 6.85 *: This lowest ph value was obtained in a patient with previous renal failure and no lactic acidosis. At the moment of such blood gas sample, the arterial PCO 2 was 64 mmhg. : This highest PaCO 2 was observed simultaneously to a ph = 6.96.
FIGURE E1: Percent mass of collapsed tissue 1.0.8.6.4.2 0.0 0.0.2.4 R 2 = 0.9835.6.8 1.0 Percent volume of collapsed tissue
FIGURE E2: 0.8 Percent of collapse 0.6 0.4 0.2 0.0 Baseline OLA Maximum Recruitment Percent-Mass of collapsed tissue Percent-Volume of collapsed tissue
FIGURE E3: 1500 1400 1300 1200 1100
FIGURE E4: Total Lung Air 2500 2000 1500 1000 500 P < 0.001 (OVERALL TREND) 0 ZEEP AFTER RM
FIGURE E5:
FIGURE E6: ROC Curve 1.00 Sensitivity.50.25 ROC Area = 0.943 0.00 0.00.25.50.75 1.00 1 - Specificity
FIGURE E7:
FIGURE E8: R = 0.67 (within subject); P < 0.001 600 500 Arterial PO 2 400 300 200 100 10 20 30 40 Static compliance ( ml / cmh 2 O)
FIGURE E9:
FIGURE E10: