Relationship
between
and venous pressure
vascular
resistance
0161-7567/83/0000-0000$01.50
Copyright
0 1983 the American
Physiological
REFERENCES
1. MOFFATT,
flow and
Environ.
2. OLDHAM,
the same
3. ZERBE,
comparing
(Regulatory
D., A. GUTYON, AND T. ADAIR. Functional
diagrams
of
volume
for the dog’s lung. J. Appl. PhysioZ.:
Respirat.
Exercise
Physiol.
52: 1035-1042,
1982.
P. A note on the analysis of repeated
measurements
of
subjects. J. Chron. Dis. 15: 969-977,
1962.
G., P. ARCHER,
N. BANCHERO,
AND A. LECHNER.
On
regression
lines with unequal slopes. Am. J. PhysioZ. 242
Integrative
Comp. PhysioZ. 11): R178-R180,
1982.
Thomas C. Lloyd, Jr.
Departments
of Medicine and Physiology
Indiana University
School of Medicine
Indianapolis,
Indiana 46223
Alveolar surface, intra-alveolar
fluid pools,
and respiratory
volume changes
To the Editor: In a recent letter to the editor, B. A. Hills
(4) challenges the hypothesis
that a continuous
fluid
layer of surfactant
lines all of the alveolar surface. He
suggests that the largely dry alveolar surface is partly
covered by discontinuous
drops of nonwetting
surfactants and that the pressure vs. volume hysteresis
cycle
could in part be explained by translation
of a force vs.
distance hysteresis
as the drops of nonwetting
fluid ride
on top of an expanding tissue support without losing its
contact angle. Hills presents theoretical
considerations
to support his belief that surfactant
is nonwetting.
This theory raises difficult issues concerning the relative significance
of cohesion of alveolar surfactant
vs.
adhesion to the wall, which would be particularly
complicated if the underlying tissue itself were to undergo
the extreme elastic changes of dimension shown in Fig.
1 of Hills. In that illustration,
the linear dimensions
increase by 2.5 times from maximum deflation to maximum inflation, which means that an area of 1 ,um” would
be stretched to 6.25 ,um2. The purpose of this letter is to
summarize two morphological
observations
incompatible
with the analysis of Hills: 1) that the contact angle of
any intra-alveolar
fluids, normal or edematous, visible in
microscopy
is always close to zero and 2) that morphological studies of the internal surface of the lung at
different points of the pressure-volume
loop reveal complex alterations
which cannot be described as simple
stretching.
Society
321
Downloaded from http://jap.physiology.org/ by 10.220.33.1 on June 16, 2017
To the Editor: A recent report by Moffatt et al. (1) uses
a common method, that of calculating pulmonary
vascular resistance
(PVR) from arterial and venous pressures and flow and then plotting PVR as a function of
pulmonary
venous pressure (Ppv), to come to a commonly held conclusion,
i.e., that PVR is remarkably
dependent on Ppv. The reason for this relationship,
shown graphically in Fig. 3A of their article, may not be
as much determined by vascular distensibility
as it is by
statistical inevitability.
Such a plot is based on nonorthogonal indices, and the outcome will show a significant
relationship between the X and Y variables, even if these
are chosen randomly, as long as the range of values is
limited (2). In this case, as in many others in physiology,
the possibilities
are not limitless: the range of venous
pressures was defined beforehand and arterial pressures
are constrained by flow and anatomy. Computer-generated trials based on random values between limits reported by Moffatt et al. for the highest and lowest flows
will show this effect. For example, if one randomly
chooses any Ppv between
-10 and 13 Torr and any
arterial pressure (Ppa) between 16 and 22 Torr, subtracts
the two to obtain a pressure gradient, divides the result
by a flow of 0.3 l/min, and plots the result against the
chosen Ppv, the resultant (linear) graph will have a slope
of about -3.3 and a correlation coefficient of about 0.95
regardless of whether one uses 10, 20, or 100 data pairs.
Similar efforts using a flow of 1.0 l/min, a randomly
chosen Ppv between the previous limits, and a randomly
chosen Ppa ranging from 24 to 31 Torr, will provide
graphs with slopes of about -1.0 and the same degree of
correlation.
The slopes are close to the reported results
in each case. Indeed in the general random case the slope
will equal the negative reciprocal
of the chosen flow.
Thus the prominent feature of Fig. 3A, i.e., the “fan” of
isoflow lines each of which apparently shows a decreasing
PVR at higher Ppv, cannot readily be distinguished
from
graphs in which causality has been deliberately excluded.
The effect of Ppv resides not so much in the “fan” or the
falling PVR as it does in the difference between curves
generated randomly and curves generated from observations, but this has not been made evident. In similar
fashion, random selection of Ppa and flow between limits
appropriate for each Ppv will generate a family of hyperbolas resembling Fig. 3B, a second example of a spurious
relationship that occurs when nonorthogonal
indices are
used.
That the vessels are really not very sensitive to Ppv
(neglecting the sluice phenomenon) is apparent from Fig.
2, where one can infer that the apparent dimensions of
the vascular bed remain the same at all venous pressures
because the amount of pressure required to increase flow
(say, from 0.3 to 0.6 l/min) was essentially the same with
each Ppv. One way to see if perfusion gradient varies in
some systematic
way with vascular pressure is to plot,
for each flow, (Ppa - Ppv) vs. (Ppa + Ppv)/2. While the
latter may not express the real weighted average vascular
pressure, its use does result in plots of orthogonal variables from which the hypothesized
relationship
can be
surmised without
concern for spurious correlations.
Alternatively,
one could test for differences among flow vs.
arterial pressure curves generated at several defined venous pressures, but that is a more demanding task (3).