Non-Invasive Estimation of Pulmonary ArterialHypertension in Chronic ObstructivePulmonary Disease
K. Spiropoulos,1 N. Charokopos,1 T. Petsas,2 G. Trakada,1 D. Dougenis,3
A. Mazarakis,4 J. Christodoulou,4 A. Peristerakis,1 P. Ginopoulos,1
N. Mastronikolis,1 and D. Alexopoulos4
Department of Internal Medicine, Divisions of1Pulmonology and4Cardiology, and the Departments of2Radiology and3Cardiothoracic Surgery, University Hospital of Patras, PC 26500 Patras, Greece
Abstract. The feasibility and reliability of the combination of several noninvasivemethods using a multivariate method of analysis to predict pulmonary artery hy-pertension (PAH) is evaluated in 20 patients with chronic obstructive pulmonarydisease. These methods comprised arterial blood gases (PaO2, PaCO2), pulmonaryfunctional parameters (FEV1), echo-Doppler parameters (tricuspid regurgitationjets, acceleration time on pulmonary valve), computed tomography measurements(transhilar distance, hilar thoracic index, and measurement of the descending branchof the right pulmonary artery to the lower lobe). A multiple stepwise regressionanalysis (including one Doppler parameter, two parameters of arterial blood gases,and one functional parameter) revealed a coefficient of determination (R2) equal to0.954 for mean pulmonary artery pressure (MPAP) with a standard error of estimate(S.E.E.) of 5.25 mmHg. A stepwise regression analysis including computed tomog-raphy and radiographic parameters revealed anR2 equal to 0.970 for PAP with aS.E.E. of 4.26 mmHg. Logistical regression analysis classified correctly 80% ofpatients with PAH using noninvasive methods such as the diameter of the mainpulmonary artery and the diameter of the left pulmonary arterial branch calculatedby computed tomography. Not only the presence of PAH but also the level of MPAPcan be estimated by the combination of multiple stepwise and logistical regressionanalyses.
The assessment of pulmonary arterial pressure (PAP) is important in clinical manage-ment and prognostic evaluation of patients with chronic obstructive pulmonary disease(COPD) [4, 29]. The need to determine the prevalence of pulmonary hypertension inCOPD, its rate of progression over time, and the long term effects of various treatmentshas stimulated many attempts to define reliable, noninvasive methods of mean pulmo-nary arterial pressure (MPAP) assessment. Until today, noninvasive diagnosis re-mained a clinical challenge [3, 25]. Recent advances in echo-doppler technology en-abled scientists to improve the estimation of MPAP, but it is still considered inferior toa right heart catheterization [17, 30]. Studies using combinations of various noninva-sive methods to predict MPAP have been carried out in the past [5, 13, 22], but theyhave failed to predict MPAP accurately even using multiple regression equations. Toour knowledge, there is no study that incorporates Doppler and computed tomography(CT) measurements in predicting MPAP using a multiple stepwise regression ap-proach. The goals of this study were to assess the respective value of arterial bloodgases (ABG) radiography, Doppler echocardiography indexes, and CT in the diagnosisof MPAP in COPD patients; to examine whether combining the results of these meth-ods could improve the detection of pulmonary hypertension; and to investigate how topredict MPAP using a certain statistical approach.
Methods
Twenty patients with COPD were selected for the present study from the outpatient clinics of the UniversityHospital of Patras. Their ages ranged from 53 to 75 years old (mean ± S.D., 67.65 ± 6.45). COPD wasdiagnosed by history, physical examination, and standard pulmonary function tests according to the Euro-pean Respiratory Society criteria [23]. Each patient was in a clinically stable phase of the disease. Patientswith clinical evidence of cor pulmonale, long term oxygen therapy, or exacerbation of their disease up to thelast 2 months were excluded from the study. Those with anemia, coagulation disorders, malignant neoplasms,ischemic heart disease, left ventricular failure of any type, or hepatic or renal failure were also excluded. Allinvestigations were performed within 24 h, with the patient taking no respiratory medication. Informedconsent was obtained from all of the subjects. Baseline noninvasive and hemodynamic data in COPD patientsare shown in Table 1. Pulmonary arterial hypertension (PAH) was considered to be present if the MPAPexceeded 20 mmHg [19].
During the study, each patient underwent a clinical evaluation, arterial blood gases analysis (NOVA statprofile 5), spirometry, diffusing capacity of the lung by the single-breath technique (Transferscreen Drager),standard posteroanterior and left chest radiographs, CT, cardiac ultrasound examination, and right heartcatheterization.
The chest radiographs were evaluated by measuring the transhilar diameter (THD), the width of theright descending pulmonary artery to the lower lobe (RLLD) as described by Lupi et al. [15], and the hilarthoracic index (HTI) as the ratio of the transhilar diameter to the transverse diameter of the thorax.
The computed tomograms were analyzed by measuring the diameters of the main pulmonary artery(MPAD), right pulmonary artery (RPAD), left pulmonary artery (LPAD), right descending pulmonary artery(RDPAD), and left descending pulmonary artery (LPAD). The widest diameters perpendicular to the longaxis at the main, left, and proximal part of the pulmonary artery were measured at about the level of thebifurcation of the pulmonary artery.
66 K. Spiropoulos et al.
Tab
le1.
Pat
ient
char
acte
ristic
s.
Sub
ject
no.
Age
PO 2
(mm
Hg)
PC
O 2(m
mH
g)F
EV
1
(Lt)
FE
V1
(%)
FE
V1/
FV
CD
LCO
/VA
(%)
RLL
D(R
o)(m
m)
TH
D(R
o)(c
m)
HT
I(R
o)M
PA
(CT
)(m
m)
LPA
(CT
)(m
m)
LDP
AD
(CT
)(m
m)
RP
AD
(CT
)(m
m)
RD
PA
D(C
T)
(mm
)
PA
SP
(US
)(m
mH
g)
AT
(US
)(m
s)
EF
(US
)P
AP
(mm
Hg)
PA
SP
(mm
Hg)
MP
CP
(mm
Hg)
PA
DP
(mm
Hg)
1M75
58.6
34.6
940
330.
4736
1811
.60.
403
3730
2826
1432
800.
6827
368
102M
6669
.443
.11,
430
490.
4372
110.
363
2621
922
1071
0.56
2534
105
3M70
50.7
45.3
880
400.
5255
0.39
3023
1721
1739
.374
0.60
3043
84
4M60
46.1
55.4
1,28
038
0.46
6415
13.6
0.34
843
3524
3022
800.
5539
5412
65M
5370
.444
.51,
620
530.
5870
1510
.70.
348
2826
1724
1632
.585
0.57
2332
1210
6M65
64.6
39.2
790
26.3
0.37
6413
.60.
371
4331
2623
2248
.370
0.78
2736
1010
7M69
74.2
401,
650
590.
5165
2518
1520
1330
.312
50.
6719
319
108M
6068
.535
.91,
470
640.
6473
2824
1723
1632
.775
0.59
2134
109
9M62
73.1
38.1
1,22
039
0.43
3215
10.9
0.37
128
2312
230
1225
850.
8416
268
010
M70
68.6
49.5
1,06
044
0.48
6820
11.4
0.33
529
2318
1818
104
0.75
2033
72
11M
7160
40.9
1,39
049
.10.
4662
.612
.30.
376
3325
1828
1795
0.60
2943
109
12M
7546
.248
.783
030
.30.
4659
1612
.50.
414
2826
1726
170.
6022
363
113
M69
64.8
47.7
1,25
043
0.44
4915
110.
324
3122
2023
1913
00.
6018
257
814
M75
67.2
33.8
1,22
036
0.52
6917
11.1
0.35
331
2116
2216
910.
6618
308
1215
M71
51.7
35.1
1,00
044
0.35
4417
10.5
0.36
3023
1923
180.
6223
3010
316
M73
5746
.81,
120
440.
5567
11.4
0.35
530
2216
2317
38.5
650.
6027
4915
817
M74
65.7
48.5
1,31
058
0.44
3510
.80.
368
2822
1521
1438
.690
0.60
2337
124
18M
5750
571,
130
33.9
0.30
3716
11.2
0.34
829
2417
2316
37.7
110
0.65
2125
112
19M
7379
.336
.21,
220
480.
4477
.712
9.9
0.33
631
2414
1814
31.3
112
0.75
1431
51
20M
6574
.543
1,02
051
.70.
4374
.412
8.8
0.32
723
1813
1914
40.7
700.
6323
3210
11M
ean
63.0
343
.17
19.0
544
.17
0.47
57.9
215
.67
11.3
10.
3630
.55
24.0
57.
4022
.80
16.1
035
.58
89.5
60.
6023
.25
34.8
510
6.25
±S
D±9
.98
±6.7
5±2
44.9
±9.9
8±0
.08
±16.
16±2
.27
±1.2
±0.3
±5.1
6±4
.11
±4.5
1±3
.05
±2.9
9±6
.13
±19.
47±0
.81
±5.6
5±7
.53
±2.6
7±3
.88
FE
V1,
forc
edex
pira
tory
volu
me
in1
s;D
LCO
/VA
,tra
nsfe
rfa
ctor
for
CO
(sin
gle
brea
thm
etho
d);M
PA
D,d
iam
eter
ofm
ain
pulm
onar
yar
tery
onC
T;L
PA
D,d
iam
ete
rof
left
pulm
onar
yar
tery
onC
T;R
DP
AD
,dia
met
erof
right
desc
endi
ngpu
lmon
ary
arte
ryon
CT
;LD
PA
D,
diam
eter
ofle
ftde
scen
ding
pulm
onar
yar
tery
onC
T;
RD
PA
D,
diam
eter
ofrig
htde
scen
ding
pulm
onar
yar
tery
onC
T;
RLL
D,
diam
eter
ofrig
htde
scen
ding
pulm
onar
yar
tery
onpo
ster
oant
erio
rra
diog
ram
s;A
T,
acce
lera
tion
time
onpu
lmon
ary
valv
e;P
AS
P,
mea
sure
men
tof
pulm
onar
yar
tery
syst
olic
.
Diameters were measured using a window level 450 HU and the center at 45 HU. Each patient wasevaluated while resting supine at end-respiration with continuous sequential 4-mm-thick scans obtained at4-mm intervals (time, 4 s; 125 KV, 310 mAs). All CT scans were conducted using CT (Somatom DRH,Siemens, Erlagen, Germany).
The echocardographic study was performed using an ultrasound system (HP Sonos 1000). The pulsedDoppler mode was used to detect pulmonary flow with a left parasternal position at the level just proximalto the pulmonary value without any angle correction. A continuous Doppler wave was used to detecttricuspid regurgitation and to calculate pulmonary artery systolic pressure (PASP). The velocities in thetricuspid regurgitant jet were obtained from the apical for chamber view and from the left lower parasternalposition.
The right heart catheterization was performed using 7F Swan-Ganz catheters. Patients were evaluatedin the supine position during the morning hours. Right ventricular end diastolic pressure (RVEDP), systolicpulmonary arterial pressure (systPAP), right atrial pressure (RAP), pulmonary artery diastolic pressure(PADP), MPAP, and mean pulmonary capillary pressure (MPCP) were recorded altogether.
Statistical Analysis
The Kolmogorov-Smirnov test was used to test the normality of the data distribution. The statistical analysiswas carried out in two steps.First Step.Spearman’s correlation coefficients were calculated to assess therelationships between values of the noninvasive methods and pulmonary pressures measured during cardiaccatheterization.Second Step.Regression analysis (multiple stepwise regression as well as logistical regres-sion analysis) were performed [9, 20].
Regression analysis models reveal the dependence of one variable on another. This relationship issummarized by the regression equation, a mathematical representation of a straight line that passes throughthe data and can be used to predict the dependent variable when the values of the independent one are given.
The adjustedR2 is a measurement of the amount of variance explained by the model. The overall modelis tested for significance with an ANOVA. All calculations were carried out with the statistical packagesStatgraphics version 7.0 and Solo version 6.0.
Results
The subjects as a whole could be described as having severe airflow obstruction. Table2 shows the average values of the arterial blood gases, pulmonary volumes, diffusingcapacity, roentgenogram, CT and echocardiographic measurements, and hemodynamicdata.
The correlation of RLLD > 15 mm with MPAP was not significant (r 4 0.177,p4 0.58) (Table 3), whereas the correlation between THD > 11 cm and MPAP washighly significant (r 4 0.52,p 4 0.033) (Table 3). The correlation between HTI andMPAP was significant but weak (r 4 0.428,p 4 0.077) (Table 3).
On CT measurements, only the RPAD correlated significantly with MPAP (r 40.45, 0.047) (Table 3).
In 60% of our patients it was possible to evaluate tricuspid regurgitation, and thatenabled us to calculate the systPAP from the cardiac ultrasound examination. Thecorrelation of the systPAP using ultrasound examination with MPAP was highly sig-nificant (r 4 0.693,p 4 0.013) (Table 3).
In 90% of our patients it was possible to evaluate the acceleration time of thepulmonary valve, and its correlation with MPAP was significant (r 4 −0.62, p 40.0061) (Table 3).
Intraobserver and interobserver variability of Doppler measurements were checked
68 K. Spiropoulos et al.
in all participating subjects and correlated well with each other (p < 0.04). Heart rateduring Doppler and cardiac catheterization was not significantly different, 79 beats/minvs 80 beats/min. Of the measurements of arterial blood gases, pulmonary volumes, anddiffusing capacity, the only parameter that correlated significantly with MPAP wasPaO2 (r 4 0.565,p 4 0.0094) (Table 3).
Paraclinical examinations were divided into two groups.A stepwise multiple equation included four variables and allowed the calculation
of MPAP with anR2 equal to 0.954. The S.E.E. was 5.25 mmHg (Table 4 and Fig. 1).Four variables were included: AT(US), PaO2, FEV1, and PaCO2.
A stepwise multiple regression equation included ten variables and allowed thecalculation of MPAP with anR2 equal to 0.970. The S.E.E. was 4.26 mmHg (Table 4and Fig. 1). The ten variables were PaO2, PaCO2, FEV1, THD (Ro), HTI (Ro), logD-MPA (CT), LDPA (CT), logRPA (CT), RDPA (CT), and RVDD (US).
Table 2. Results of spirometric data, ABG, roentenogram, CT, ultrasound measurements, and hemody-namic data.
Unit Mean ±S.D. Range Median
Age Years 67.65 6.45 53–75 69.5PaO2 mmHg 63.03 ±9.98 46.1–79.3 65.25PaCO2 mmHg 43.17 ±6.75 33.8–57 43.05FEV1 ml 1191.05 ±244.93 790–1650 1,220FEV1% % 44,17 ±9.98 26.3–64DLCO/VA ml/mmHg/sec 57.92 ±16.16 32–77.7 64.5MPA(CT) mm 30.55 ±5.16 23–43 29.5LPAD(CT) mm 24.05 ±4.11 18–35 23LDPAD(CT) mm 17.40 ±4.51 9–28 17RPAD(CT) mm 22.80 ±3.05 18–30 23RDPAD(DT) mm 16.10 ±2.99 10–22 16RLLD(Ro) mm 15.67 ±2.27 12–20 15.5THD cm 11.31 ±1.2 8.8–13.6 11.1HTI 90.36 ±0.3 0.32–0.41 0.36AT(US) MSEC 89.56 ±19.47 65–130 85PASP(US) mmHg 35.58 ±6.13 25–48.3 35.2EF(US) % 60.00 ±0.81 55–84 60RVDD(US) cm 3.38 ±0.56 2.51–4.73 3.39MPAP mmHg 23.25 ±5.65 14–39 23PASP mmHg 34.85 ±7.53 25–54 33.5PVEP mmHg 6.25 ±3.88 0–12 7MPCP mmHg 10.00 ±2.67 3–15 10
FEV1, forced expiratory volume in 1 s; DLCO/VA, transfer factor for CO (single breath method); MPAD,diameter of main pulmonary artery on CT; LPAD, diameter of left pulmonary artery on CT; RDPAD,diameter of right descending pulmonary artery on CT; LDPAD, diameter of left descending pulmonary arteryon CT; RDPAD, diameter of right descending pulmonary artery on CT; RLLD, diameter of right descendingpulmonary artery on posteroanterior radiographs; AT, acceleration time on pulmonary valve; PASP, mea-surement of pulmonary artery systolic pressure from tricuspid regurgitation; EF, ejection fraction of leftventricle; THD, transhilar diameter; HTI, hilar thorax index; PAP, mean pulmonary artery pressure (invasivemethod); RVEP, right ventricle end-diastolic pressure (invasive method); PASP, systolic pulmonary arterialpressure (invasive method); RVEP, right ventricular end-diastolic pressure (invasive method); MPCP, meanpulmonary capillary pressure (invasive method).
Non-Invasive Estimation of Pulmonary Arterial Hypertension 69
The ten variables included in the stepwise multiple regression analysis were alsoincluded in the logistical regression analysis. We also included HTI (Ro), AT (US), andPASP (US). HTI (Ro) and PASP (US) were deleted because of insufficient data. Byusing logistical regression analysis we classified correctly 80% of the diagnosis of PAHusing noninvasive methods such as LPAD (CT) and MPA (CT) (Table 4).
Table 3. Correlation analysis of various paraclinical examinations and MPAP.
The fact that pulmonary hypertension develops insidiously, producing few diagnosticclues until cor pulmonale becomes clinically evident, makes the early detection ofpulmonary hypertension in COPD patients an important clinical challenge. SeverePAH (MPAP > 40 mmHg) is generally detected by conventional methods (such asEKG), but MPAP rarely exceeds 40 mmHg in COPD patients investigated during astable phase of their disease [21, 31]. In our study none of the patients experiencedMPAP greater than 40 mmHg. Although this group of patients had severe airflowobstruction the spirometric indices correlated poorly with MPAP. Measurements ofTHD were more highly correlated with MPAP (r 4 0.52,p 4 0.033) and in agreement
Fig. 1. Predicted and observed values for MPAP using the equations derived from the paraclinical exami-nations.
Non-Invasive Estimation of Pulmonary Arterial Hypertension 71
with the correlation coefficient of 0.62 reported by others [6]. The lack of a detectablecorrelation in RLLD and MPAP was probably related to a higher proportion of ourpatients with a ‘‘dirty lung’’ appearance on chest radiograms. A statistically significantcorrelation was observed between RPAD on CT and MPAP (r 4 0.45,p 4 0.047).Kuriyama et al. [14] reported a very good correlation between MPAD and MPAP; themeasurement methology was identical in the two studies, but our scans were of 4-mmthickness instead of the 1-cm thickness of Kuriyama; moreover, our study was pro-spective.
Tricuspid regurgitant jet gave the best correlation with MPAP (r 4 0.693,p 40.013), and was in excellent agreement with the correlation coefficient of 0.72 reportedby others [24, 26], although inferior to that reported for patients with congenital oracquired heart disease [24, 28]. Hinderliter et al. [11] reported that Doppler-derivedcalculation of PASP from the tricuspid regurgitant jet area and invasive measurementsof systPAP correlated significantly (r 4 0.57,p < 0.001). In our series the correlationbetween PASP and systPAP wasr 4 0.57, p 4 0.05. Our differences could beassigned to the differences in the selection of patients with primary pulmonary hyper-tension vs secondary pulmonary hypertension because at a given MPAP pulse pressurevaries with arterial compliance and wave reflection depending on the type of pulmo-nary hypertension [16]. The less direct method of estimating PAP from accelerationtime in the right ventricular outflow tract was technically easier but slightly inferior tothe tricuspid jet method. At the same time, it must be emphasized that the correlationcoefficients of Doppler measurements with PAP were higher than those reported forrespiratory function tests, arterial blood gas tension, X-ray chest examination, M-modeechocardiography, and other methods of noninvasive assessment of PAP in COPDpatients [13]. Doppler echocardiographic examination and heart catheterization werenot performed simultaneously. It has been shown that the pressure in COPD patientsexhibits, at rest, a coefficient of variation up to 10.6% [12], thus it must been taken inaccount that in individual patients, PAP could be considerably different at the momentof Doppler-derived and manometric measurements. This is probably expected in pa-tients with marked differences in heart rate between noninvasive and invasive methods.In our study the heart rate was not statistically significant between manometric andDoppler measurements, but one should also consider the potential drawbacks of si-multaneous Doppler and catheter studies, related to nonoptimal positioning of patientsduring echo-Doppler examination and consequently, the reduced consistency of cath-eter measurements. Finally, the catheter itself can cause tricuspid insufficiency, and inthis way it artificially increases the prevalence, as well as the quality of Doppler-detected transtricuspid regurgitant jets [10].
An attempt to predict PAH using multiple regression equations has been carriedout in the past but has failed to predict the presence of PAH in an appreciable per-centage of cases, particularly when the MPAP ranged from 20 to 30 mmHg [3–5, 13,30]. In addition, studies in the past used paraclinical examinations (electrocardiograph,kinetokardiography, thallium-201 myocardial scintigraphy, and orthostatic changes inCO transfer factor) that are of poor predictive value in COPD patients as far as PAHis concerned, so they have been abandoned in the present study. Tartullier et al. [25]noted that the detection of PAH was reinforced by the inclusion in regression analysisof ABG. According to their results, prediction of PAH could be improved by theinclusion of ABG or pulmonary function parameters. Oswald-Mammoser et al. [22],
72 K. Spiropoulos et al.
using stepwise regression analysis, noted that the multiple regression correlation co-efficient was only 0.659 when investigated by noninvasive examinations to predictPAH. Our multiple stepwise regression analysis, where we have included four nonin-vasive parameters (AT, PaO2, PaCO2, FEV1), revealed coefficient of determination (R2)equal to 0.954 and a S.E.E. of 5.25 mmHg. In an attempt to improve the detection ofMPAP we incorporated in the multiple stepwise analysis radiographic and other ul-trasound indices. The results were better as determined from anR2 equal to 0.970 andS.E.E. of 4.26 mmHg.
Logistical regression analysis using the same parameters allowed us to classifycorrectly 80% of the diagnoses of PAH using noninvasive methods such as LPAD (CT)and MPA (CT). In contrast to other studies that have been carried out in the past, weincorporated in multiple stepwise analysis parameters independent of their correlationcoefficient with MPAP. Omission of a variable specified by the truth introduces biasin all of the least squares estimates [1]. In addition we must recognize that MPAP couldbe affected by several parameters so the impact on MPAP of simultaneous changes ofnoninvasive parameters must be studied.
Finally, we wish to derive insight into causative mechanisms by discovering whichset of noninvasive parameters has the most influence on MPAP. Predictors of benefitsuch as changes in MPAP have been studied in the past. Weitzenblum et al. [29] statedthat pulmonary hypertension worsens the prognosis in COPD patients. The results ofthe Nocturnal Oxygen Therapy Trial Group [21] indicated that the benefits of continousO2 therapy were most evident in COPD patients who had relatively mild disturbancesof pulmonary hemodynamics, but these changes did not correlate with mortality. Are-examination of the above study [27] mentioned that patients on continuous O2
therapy displayed an increase in resting and exertional right ventricular stroke volumetogether with decreased pulmonary vascular resistance and that changes in MPAPobserved during the first 6 months were predictive of survival. Our results showed thatMPAP can be predicted with anR2 equal to 0.95–0.97 if we utilize examinations suchas arterial blood gases, Doppler ultrasound, and CT. This could be useful in screeningPAH in COPD patients.
In 1992 Ashutosh and Dunsky [2] measured pulmonary hypertension in 43 patientswith COPD commencing LTOT. A fall in MPAP of more than 5 mmHg was associatedwith a 3-year mortality of 30% compared with 90% in nonresponders. Cooper et al. [7]reported that survival was impaired significantly in patients who started LTOT within2 months of the first episode of edema. Moreover, detection of MPAP progressioncould be useful to study its relevance in terms of survival, which is not yet clear.Likewise, the noninvasive assessment of pulmonary hypertension among patients suf-fering from severe pulmonary emphysema, may prove to be a clinically useful param-eter in defining whether patients undergo lung volume reduction surgery or lungtransplantation. This is particularly important, because various centers have differentselection criteria for lung volume reduction, and not all patients will benefit from theprocedure [8, 18].
References
1. Armitage G, Berry (eds) (1987) Statistical Methods in Medical Research, pp 296–357. BlackwellScientific Publications, Oxford
Non-Invasive Estimation of Pulmonary Arterial Hypertension 73
2. Ashutosh K, Dunsky M (1992) Noninvasive tests for responsiveness of pulmonary hypertension tooxygen. Chest 3:393–399
3. Bishop JM, Cross KW (1981) Use of other physiologic variables to predict pulmonary artery pressurein patients with chronic respiratory disease. Eur Heart J 2:509–517
4. Bishop JM, Cross KW (1984) Physiological variables and mortality in patients with various categoriesof chronic respiratory disease. Bull Eur Physiopathol Respir 20:445–500
5. Bishop JM, Csukas M (1989) Combined use of non-invasive techniques to predict pulmonary arterialpressure in chronic respiratory disease. Thorax 44:85–96
6. Concise Clinical Studies (1982) Am Rev Respir Dis 126:338–3417. Cooper CB, Waterhouse J, Howard P (1987) Twelve-year clinical study of patients with hypoxic cor
pulmonale given long term domiciliary oxygen therapy. Thorax 42:105–1108. Cooper FD, Trulirk EP, Triantafillou AN, Patterson GA, Pohl MS, Deloney PA, Sundaresan RS, Roper
9. Fisher L, Van Belle G (1993) Biostatistics: a methology for the health sciences. In: Fisher L, Van BelleG (eds) John Wiley and Sons, New York
10. Himelman RB, Stulbary M, Kircher B (1989) Noninvasive evaluation of pulmonary artery pressureduring exercise by saline-enhanced Doppler echocardiography in chronic pulmonary disease. Circulation79:863–871
11. Hinderliter AL, Park W, Willis IV, Robyn J, Barst et al. for the Primary Pulmonary Hypertension StudyGroup (1997) Circulation 95:1478–1486
12. Jezekova J, Michaljanic A, Jesek V (1987) Natural variability of pulmonary hemodynamics. Eur HeartJ 8(Suppl II):40
13. Keller AC, Shepard W, Duck P, Chun S, Vasquez P, Dolan G (1986) Pulmonary hypertension in chronicobstructive pulmonary disease. Chest 90:185–192
15. Lupi E, Dumont C, Tejada VW, Horwitz S, Gallant F (1975) A radiological index of pulmonaryhypertension. Chest 75:656–662
16. MacDonald’s (1990) Blood flow in arteries. In: Nichols WW, O’Rourke MF (eds). 3rd ed. EdwardArnold, London
17. MacNee W (1994) Pathophysiology ot cor pulmonale in chronic obstructive pulmonary disease. Am JRespir Crit Care Med 150:833–853
18. Mckenna RJ, Brenner M, Fischel RJ, Singh N, Yoong B, Gelb AF, Ojann KE (1997) Patient selectioncriteria for lung volume reduction surgery. J Thorac Cardiovasc Surg 114:964–967
19. Michael TR, Summer WR (1985) Pulmonary hypertension. Lung 163:65–8220. Morisson DF (1976) Multivariate Statistical Methods. McGraw-Hill, New York21. Nocturnal Oxygen Therapy Trial Group (1980) Continuous nocturnal oxygen therapy in hypoxaemic
therapy chronic obstructive lung disease. Ann Intern Med 93:391–39822. Oswald-Mammoser M, Oswald T, Nyankin E, Dickele MC, Gragne D, Weitzenblum E (1987) Nonin-
vasive diagnosis of pulmonary hypertension in chronic obstructive pulmonary disease. Eur J Respir Dis71:419–429
23. Siafakas NM, Vermiere P, Pride NB, Paoletti P, Gibson J, Howard P, et al. (1995) Optimal assessmentand management of chronic obstructive pulmonary disease. Eur Respir J 8:1398–1420
24. Skjaerpe T, Hattle L (1986). Noninvasive estimation of systolic pressure in the right ventricle in patientswith tricuspid regurgitation. Eur Heart J 7:704–710
25. Tartullier M, Greiner O, Tartullier G, Boutarin J (1985) Noninvasive diagnosis of pulmonary arteryhypertension in chronic obstructive lung disease. Prog Respir Res 20:126–135
26. Tramarin R, Torbicki B, Marchandise J, Laaban P, Morpugo M (1991) Doppler echocardiographicevaluation of pulmonary artery pressure in chronic obstructive pulmonary disease. Eur Heart J 12:103–111
27. Timms RM, Khaja FU, Williams GW (1985) Nocturnal Oxygen Therapy Trial Group. Hemodynamicresponse to oxygen therapy in chronic obstructive pulmonary disease. Ann Intern Med 102:29–36
28. Vazquez de Prada JA, Ruano J, Martin Duran R (1987) Noninvasive determination of pulmonary arterysystolic pressure by continuous wave Doppler. Int J Cardiol 16:177–184
74 K. Spiropoulos et al.
29. Weitzenblum E, Hirth C, Ducolone A, Mirhom R, Rasaholinjanahary J, Ehrhart M (1981) Prognosticvalue of pulmonary artery pressure in chronic obstructive pulmonary disease. Thorax 36:752–758
30. Weitzenblum E, Oswald M, Oswald T, Apprill M (1989) Noninvasive diagnosis of pulmonary hyper-tension in chronic lung disease. In: Wagenvoort CA, Denolin H (eds) Pulmonary Circulation: Advancesand Controversies. Elsevier, New York
31. Weitzenblum E, Sautegeau A, Ehrhart M, Mammoser M, Hirth C, Roegel E (1984) Long term courseof pulmonary arterial pressure in chronic obstructive pulmonary disease. Am Rev Respir Dis 130:993–998
Accepted for publication: 13 August 1998
Non-Invasive Estimation of Pulmonary Arterial Hypertension 75
