Correspondence INTACT EPITHELIAL BARRIER FUNCTION IS CRITICAL FOR RESOLUTION OF ALVEOLAR EDEMA IN HUMANS
To the Editor: I am writing concerning the recent paper by Matthay and WienerKronish (1). I am wondering why they chose to group patients with hydrostatic and permeability pulmonary edema in one group for their statistical analysis. One would expect the patients with hydrostatic pulmonary edema to have lower total protein and albumin concentrations than the permeability pulmonary edema subgroup. Thus, this population of patients would be expected to have a bimodal distribution with one mean for the hydrostatic patients and a separate mean for the permeability patients. As concerns the group who made clinical improvement, the distribution is skewed by the fact that there are more hydrostatic pulmonary edema patients in this group than permeability edema patients. In the group of patients who did not make improvement, the distribution is skewed by the greater frequency of permeability pulmonary edema. A single mean does not adequately represent either group. I am not sure that this makes a difference with performing a paired Student's t test. It may, however, influence an unpaired t test. A z-distribution is a theoretical probability distribution that is symmetrical and bell-shaped and is similar to a normal distribution, but more spread out. I am not sure that one can apply this to a group that has a bimodal distribution and may be skewed towards one end as in the present case.
Because the rise in edema fluid/plasma protein ratio correlated with improved clinical indices in all patients, the statistical test used for analyzing the data is not crucial. We hope that this explanation clarifies the issue. MICHAEL
JEANINE P. WIENER-KRONISH, M.D.
Cardiovascular Research Institute University of California San Francisco, CA 1. Matthay MA, Landolt CC, Staub NC. Differential liquid and protein clearance from the alveoli of anesthetized sheep. J Appl Physiol 1982; 53:96-104. 2. Matthay MA, Berthiaume Y, Staub NC. Long term clearance of liquid and protein from the lungs of unanesthetized sheep. J Appl Physiol1985; 59:928-34. 3. Berthiaume Y,Broaddus VC, Gropper MA, Tanita T, Matthay MA. Alveolar liquid and protein clearance from normal dog lungs. J Appl Physiol 1988; 65:585-93. 4. Smedira N, Gates L, Hastings R, Jayr C, Sakuma T, Matthay MA. Alveolar and lung liquid clearance in anesthetized rabbits. J Appl Physioll991; 70:1827-35. 5. Matthay MA, Weiner-Kronish JP. Intact epithelial barrier function is critical for the resolution of alveolar edema in humans. Am Rev Respir Dis 1990; 142:1250-7.
PETER R. ROGOL, M.D.
Pulmonary Associates of New Haven, RC New Haven, CT 1. Matthay MA, Wiener-Kronish JP. Intact epithelial barrier function is
critical for the resolution of alveolar edema in humans. Am Rev Respir Dis 1990; 142:1250-7.
From the Authors: In reply to Dr. Rogol's letter, we grouped patients with hydrostatic or increased permeability pulmonary edema together because we were interested in determining if a rise in the edema fluid/plasma protein ratio correlated with clinical improvement independent of the type of edema. In several of our experimental studies (1-4), a rise in alveolar fluid protein concentration has correlated well with net alveolar liquid clearance. In order to test this hypothesis in a clinical study, we needed to determine whether a rise-in edema fluid protein concentration would occur in patients in whom there was evidence of clinical improvement (improved oxygenation and decreased edema on chest radiograph). Therefore, the patients were grouped together, independent of the type of pulmonary edema. In the study, every patient who had a rise in edema fluid/plasma protein concentration in the second sample had evidence of improved oxygenation. However, we did analyze the data for the two types of edema separately as well.For example, there were 18patients with hydrostatic pulmonary edema, 15 of whom had an increase in the edema fluid/plasma protein concentration ratio and each patient had improvement in their oxygenation, whereas the three patients who had no change in the edema fluid/plasma protein concentration ratio had no improvement in oxygenation. Similarly, there were 16 patients with increased permeability pulmonary edema, and nine of those patients showed a rise in their edema fluid/plasma protein concentration ratio, which correlated in each patient with improved clinical indices (5). 468
SURFACTANT REPLACEMENT IMPROVES LUNG RECOIL IN RABBIT LUNGS AFTER ACID ASPIRATION
To the Editor: I am puzzled by the methods used to determine static air P-Vcurves and the interpretation of results in the report by Lamm and Albert (1) in the December 1990 issue of the Review. Why was surfactant instilled in degassed lungs? Degassing, if done correctly, will render the lungs airless and fully collapsed. Injecting gas or liquid in such a lung would require substantial pressure and would result in a very inhomogenous distribution of the material used due to opening pressure and recruitment phenomena. The pressure-volume (P-V) curves of the control and the HCl groups, which were determined in degassed lungs, cannot be compared with that of Hel and surfactant group, which was not done in the degassed state (because of surfactant instillation). How could the volume of the liquid surfactant instilled be added to the volume of the gas injected for lung inflation? This would have been acceptable if saline, not air, P-V curves were done. In any case, how could the addition of the surfactant volume underestimate the true TLC? Lungs were inflated to 25 and 40 em H 2 0 in in vitro and in vivo studies, respectively. Figures 1 and 2, however, show only pressures up to approximately 15 and 25 em H 20 in the two preparations, respectively. Did the P-V curves reach the plateau, and if so, why are not the entire scope of P-V relationships presented in the figures? The P-Vmeasurements were done by inflating and deflating the lungs in to-ml steps and recording pressure at each point, which would have been different for each lung according to its mechanical characteristics. Thus, how could one normalize pressures at exact percent TLC points? This could only be done accurately if lungs are deflated by stepwise withdrawal of exact percent TLC (90, 80, 70, etc.) rather than standard 10-ml volumes with pressure recordings at each point. Or, more conventionally, the P-V curves are constructed by stepwise inflation and deflation, to predetermined pressures, and recording the volume.