1. Journal of Veterinary Cardiology (2007) 9, 83e89
www.elsevier.com/locate/jvc
Clinical and echocardiographic findings
of pulmonary artery stenosis in seven cats
Donald P. Schrope, DVM, Dipl ACVIM a,*,
b
William J. Kelch, DVM, PhD, Dipl ACVPM
a
Oradell Animal Hospital, 580 Winters Avenue, Paramus, NJ 07652, USA
b
Department of Comparative Medicine, College of Veterinary Medicine,
University of Tennessee, Box 1071, Knoxville, TN 37901-1071, USA
Received 19 November 2006; received in revised form 19 November 2006; accepted 10 September 2007
KEYWORDS Abstract Objectives: Describe the clinical, electrocardiographic (ECG), radio-
Pulmonary artery; graphic and echocardiographic findings in cats with isolated pulmonary artery ste-
Stenosis; nosis. Assess the usefulness of systolic and diastolic Doppler measurements at
Feline; predicting stenosis severity.
Congenital Background: Pulmonary artery stenosis is an infrequent congenital cardiac defect in
humans that has not been reported in cats. In humans, pulmonary artery stenosis is
usually seen in conjunction with other cardiac defects and may lead to clinical signs
if severe.
Animals, materials and methods: Seven cats with pulmonary artery stenosis were
retrospectively evaluated. Medical records, radiographs, ECGs, echocardiograms
and angiocardiograms were reviewed. Severity of stenosis was assessed by two-
dimensional and color Doppler echocardiographic evaluation and clinical findings.
Peak systolic and diastolic gradients across the stenosis, and systolic and diastolic
pressure decay half-times were graded using echocardiography. In addition, the du-
ration of antegrade flow during diastole was subjectively assessed. Univariate anal-
yses were performed to assess the best variable to predict stenosis severity.
Results: Concurrent congenital defects were not identified. Only cats with severe
obstruction showed clinical signs including exertional dyspnea and lethargy. Dia-
stolic Doppler measurements were superior to systolic measurements at predicting
severity of stenosis. Antegrade flow throughout diastole and/or a diastolic pressure
half-time of >100 ms indicated severe obstruction. The prognosis for pulmonary
artery stenosis appears to be good regardless of severity.
* Corresponding author.
E-mail address: dpsdvm@optonline.net (D.P. Schrope).
1760-2734/$ - see front matter ª 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jvc.2007.09.001
2. 84 D.P. Schrope, W.J. Kelch
Conclusion: Among cats with pulmonary artery stenosis, clinical signs are uncom-
mon and prognosis is good. Doppler assessment of diastolic flow appears to be
superior to systolic flow at predicting severity.
ª 2007 Elsevier B.V. All rights reserved.
Congenital stenosis of the main and/or The P-wave and QRS complex amplitude and
branched pulmonary arteries has been identified duration, and the PR interval were measured on
in humans and is commonly identified with con- three consecutive cardiac cycles and averaged on
current lesions such as ventricular septal defect, the electrocardiograms that were available for
pulmonic stenosis, patent ductus arteriosus, or review. The mean QRS axis was also assessed.
tetralogy of Fallot.1e3 Congenital obstruction of The vertebral heart scale (VHS) was measured on
the right ventricular outflow tract is uncommon all lateral and ventrodorsal radiographs that were
in cats with an incidence of about 2e3% of congen- available for review.12
ital cardiac defects.4,5 To the authors’ knowledge, Complete echocardiograms were performed us-
stenosis of the main or branched pulmonary ar- ing standard echocardiographic views. Maximum
teries has not been reported in a cat. pulmonary valve annulus diameter (PVD) and
Stenosis of the pulmonary artery is hemody- pulmonary artery stenosis diameter (PSD) were
namically similar to coarctation of the aorta and obtained from right or left parasternal short-axis
was referred to as coarctation of the pulmonary two-dimensional (2D) images during systole. Pul-
artery in early human studies.1 As with other types monary artery stenosis diameter was evaluated
of stenotic lesions, Doppler echocardiography has with the assistance of color-flow Doppler (CFD) to
been used to assess the severity of aortic coarcta- identify the vena contracta in all cases. In cats
tion using the modified Bernoulli’s equation to es- with severe disease the stenotic orifice was often
timate the systolic pressure gradient across the so small that it was not easily identified without
obstruction, however, this does not consistently CFD. The ratio of stenosis diameter to pulmonary
reflect severity of the stenosis when compared to valve annulus diameter (PSD/PVD) was calculated.
gradients measured at cardiac catheterization.6,7 Maximum right atrial diameter (RAD) and left atrial
Further studies of aortic coarctation in humans diameter (LAD) were obtained from right para-
have identified variable degrees of diastolic flow sternal long-axis 2D images during diastole. Mea-
across the stenosis. Both subjective and objective surements of the RAD were performed using
assessment of this diastolic flow has been found to a method similar to published methods of measur-
be more accurate than systolic gradients at assess- ing the LAD in long-axis.13 The ratio of RAD to LAD
ing severity of coarctation in many patients.6,8,9 In was calculated and considered normal if <1.0.14
humans with pulmonary artery stenosis and in ex- Right ventricular end-diastolic diameter (RVDd)
perimental studies of pulmonary artery banding and left ventricular end-diastolic diameter (LVDd)
in dogs, there are varying conclusions about how were obtained from M-mode tracings. The RVDd
closely the Doppler systolic and cardiac catheteri- to LVDd ratio was generated and considered nor-
zation-derived gradients correlated.10,11 In cats mal if <0.33.14 The remainder of the left heart
with pulmonary artery stenosis, it was hypothe- measurements was made from standard 2D and
sized that diastolic Doppler evaluation would re- M-mode echocardiographic views. All 2D and M-
late better to severity than systolic evaluation. mode measurements were made on three consec-
utive cardiac cycles and averaged.
Peak systolic flow velocity through the stenosis
Animals, materials and methods was measured in all cats after normal pulmonary
valve flows were confirmed and a velocity step-up
Cats with an echocardiographic diagnosis of pul- was identified at the level of the stenosis. Estima-
monary artery stenosis were retrospectively re- tion of the peak systolic pressure gradient (PaS)
viewed. The majority of these cats were identified and peak diastolic pressure gradient (PaD) across
through an animal shelter that receives animals the stenosis was calculated using the modified
from multiple sites along the east coast of the Bernoulli’s equation. Peak diastolic flow velocity
United States. All of the cats were initially evalu- was measured at the onset of diastole identified by
ated because a murmur had been ausculted on the end of the T-wave on the electrocardiogram
physical exam. (Fig. 1).8 The systolic pressure decay half-time
3. Clinical and echocardiographic findings 85
multiple determination which estimates the pro-
portion of the response variation that can be
explained by the continuous independent variable;
and r-squared for an ordinal variable refers to the
uncertainty coefficient which, analogous to the co-
efficient of multiple determination for continuous
variables, also explains the proportion of the re-
sponse variation explained by the ordinal indepen-
dent variable.
Results
Figure 1 Spectral Doppler findings in cats 3 (A) and 7 The signalment, clinical signs, and findings of
(B) and technique used to identify diastolic Doppler cardiac auscultation are summarized in Table 1.
measurements. The onset of diastole was identified at None of the cats had evidence of jugular pulses,
the end of the T-wave and the peak diastolic gradient cyanosis, alterations in femoral pulse strength or
was measured at this point. The systolic pressure decay quality, abnormalities in S1 or S2, or a gallop
half-time (Spht) was measured from the peak systolic rhythm during a resting examination.
flow to the onset of diastole. The diastolic pressure Six of the cats were asymptomatic at the time
decay half-time (Dpht) was measured from the onset of presentation. One cat (diagnosed at 12 years of
of diastole to the end of diastolic flow.
age) was described as less active than expected all
of her life by one of two owners. One cat, that had
(Spht) was measured from the peak systolic flow to initially been asymptomatic, developed moderate
the onset of diastole (the end of the T-wave). The exertional dyspnea soon after diagnosis. An arte-
diastolic pressure decay half-time (Dpht) was mea- rial blood gas was performed after dyspnea was
sured from the onset of diastole to baseline or to induced by exercise. The arterial blood gas was
the onset of systole if antegrade flow was present assessed as a mixed metabolic and respiratory
throughout diastole. All measurements were per- acidosis with hypoxia (pH 7.1, pCO2 45.6, pO2
formed on three consecutive waveforms and aver- 84.0, HCO3 14.0, BEe15, sO2 90%). The calculated
aged. The presence or absence of diastolic flow alveolar to arterial gradient was nine. The findings
(DFA) was also subjectively assessed from the were consistent with hypoperfusion from pulmo-
spectral Doppler tracing and graded as; (0þ) no di- nary artery stenosis. The clinical signs in cat 7
astolic flow, (1þ) flow through 50% or less of dias- did not progress over time, and no treatment was
tole, (2þ) flow through >50% but <100% of initiated.
diastole, and (3þ) diastolic flow throughout dias- Three cats had an ECG available for review. No
tole resulting in continuous flow across the stenosis abnormalities were identified in two of the cats (#2
(Fig. 1). and 6). The ECG of cat 5 revealed deep S-waves in
Angiocardiography was performed to better leads I, II, III, aVF, a right mean electrical axis shift
characterize the lesion in the two initial cases. (170 e180 ), and widening of the QRS complex
The ratio of minimal stenosis diameter to PV (50 ms) suggestive of right ventricular enlargement
annulus diameter was obtained and averaged or a partial right bundle branch block. Five cats
from at least two cardiac cycles during systole. (#3e7) had chest radiographs available for review.
For statistical analysis, severity of stenosis was All five cats had normal VHS on lateral and ventro-
assessed based on the echocardiographic PSD/PVD dorsal views and one cat (#7) had subjective evi-
ratio. The diastolic and systolic Doppler data were dence of right ventricular enlargement and
compared to the PSD/PVD ratio. Univariate analy- a small pulmonary artery bulge.
ses were performed to assess the correlation with Echocardiography was performed in all seven
stenosis severity.c A p-value 0.05 was considered cats and identified pulmonary artery stenosis with
significant; r refers to the coefficient of linear cor- variable degrees of main pulmonary artery dilation
relation between two variables; r-squared for con- proximal to the stenotic lesion (Figs. 2 and 3).
tinuous variables refers to the coefficient of Color-flow Doppler confirmed turbulence during
systole at the suspected stenosis in all cats and di-
astolic flow at the level of the stenosis in cats 6
c
JMP, Version 5.1 statistical software (SAS Institute Inc., SAS and 7. No cat showed evidence of pulmonary valve
Campus Dr, Cary, NC 27513, USA). disease or left heart abnormalities. A summary of
4. 86 D.P. Schrope, W.J. Kelch
Table 1 Signalment, initial auscultation findings, and clinical signs in seven cats with pulmonary artery stenosis
Cat Breed Sex Age at Age as of Auscultation at initial Clinical signs
diagnosis publication (years) examination
1 DSH F 2 months 6.8 2e3/6 left and right basilar SM None
2 DSH F 8 months 7.4 2e3/6 right basilar SM None
3 DSH F 9 months 9 3/6 left basilar SM None
4 DSH M 3 years 9a 4/6 left basilar SM None
5 Persian M 5 months 4.6 3/6 right basilar SM None
6 DSH F 12.7 years 13.9b 4/6 left basilar SM Lethargy
7 DSH M 15 months 9.6 2e3/6 left basilar SM and 1/6 DM Exertional dyspnea
DSH ¼ domestic short hair, M ¼ male, F ¼ female, SM ¼ systolic murmur, and DM ¼ diastolic murmur.
a
Lost to follow-up.
b
Euthanized.
the 2D echocardiographic findings of the right just proximal to the bifurcation without involve-
heart and spectral Doppler data is presented in Ta- ment of the branched or peripheral segmental
ble 2. Visualization of the pulmonary artery bifur- pulmonary artery (Fig. 4). The average ratio of
cation and branched pulmonary arteries from the PSD/PVD on the angiocardiograms was 0.50 in cat
right parasternal short-axis view was accomplished 4 and 0.32 in cat 7.
in the majority of cats. In cat 6 the branched pul- Based on the univariate analyses, the Dpht
monary arteries could only be visualized from (p ¼ 0.002; r ¼ À0.94; r2 ¼ 0.88) was most closely
a left parasternal short-axis view. In cats 4 and 7 associated with stenosis severity. The DFA
the branched pulmonary arteries could not be visu- (p ¼ 0.025; r2 ¼ 0.94), PaD (p ¼ 0.003; r ¼ À0.92;
alized well from the right or left side with 2D. The r2 ¼ 0.85) and PaS (p ¼ 0.009; r ¼ À0.88; r2 ¼ 0.78)
CFD in both cats, though, did show flow conver- were also significantly associated with stenosis se-
gence and turbulence at the bifurcation with split- verity. Cats 6 and 7 had evidence of right atrial
ting of the color-flow pattern into the origins of and ventricular dilation as well as clinical signs. It
the branched pulmonary arteries. was felt that these two cats had severe disease.
Angiocardiography confirmed in cats 4 and 7 an Both cats with severe stenosis had a DFA of 3þ dia-
isolated lesion of the distal main pulmonary artery stolic flow and Dpht 100 ms (Table 2).
Figure 2 Two-dimensional (A) and color-flow Doppler (B) images of cat 1 with mild pulmonary artery stenosis. Im-
ages were obtained during systole. Note the mild narrowing of the pulmonary artery just proximal to the bifurcation
(A) and associated flow convergence using color-flow Doppler (B). (Ao ¼ aorta, MPa ¼ main pulmonary artery, RPa ¼
right-branched pulmonary artery, LPa ¼ left-branched pulmonary artery).
5. Clinical and echocardiographic findings 87
Figure 3 Two-dimensional (A) and color-flow Doppler (B) images of cat 6 with severe pulmonary artery stenosis. Im-
ages were obtained during systole. Note the severe stenosis in the body of the pulmonary artery and severe dilation of
the main pulmonary artery (A). The stenotic orifice was very difficult to identify using only 2D imaging in this cat.
Color-flow Doppler identified the orifice of the stenotic lesion and confirmed flow convergence (B). (Ao ¼ aorta,
MPa ¼ main pulmonary artery, RPa ¼ right-branched pulmonary artery, LPa ¼ left-branched pulmonary artery).
All of the cats were alive well into maturity and conjunction with other congenital cardiac anoma-
only cat 6 had died at the time of publication. Cat 6 lies. The cause of pulmonary artery stenosis is
had been euthanized approximately 15 months unclear. Postmortem studies in adults reveal intimal
after diagnosis due to repeated signs of anorexia, invasion by smooth muscle, medial hyperplasia, and
depression, and sneezing, as well as a suspicious increased and disorganized elastin fibers at the site
pulmonary nodule on radiographs. A necropsy was of stenosis. In contrast, postmortem studies in
not allowed. Cat 4 was lost to follow-up at approx- infants reveal fibrous intimal proliferation, medial
imately 8 years of age. At the time of publication, hypoplasia, and loss of elastin fibers.3 Postmortem
serial echocardiograms have been performed on the samples were not available from cats in this study
remaining five cats. No progression in the degree of for comparison with the findings in humans.
stenosis has been identified. A classification scheme for pulmonary artery
stenosis based on the stenosis location and number
of lesions has been developed in humans.1 The
Discussion classification divides the pulmonary artery into
four segments; (1) main pulmonary artery, (2) bi-
Pulmonary artery stenosis is a well-known congen- furcation, (3) left- and right-branched pulmonary
ital anomaly in humans that most often occurs in arteries, and (4) peripheral segmental pulmonary
Table 2 Echocardiographic findings at initial diagnosis
Cat Wt RVDd/LVDd RAD/LAD PVD PSD PSD/PVD PaS Spht PaD Dpht DFA
(kg) (cm) (cm) (mmHg) (ms) (mmHg) (ms) (þ)
1 0.8 0.19 0.92 0.50 0.30 0.60 13.0 37 0.00 0 0
2 2.5 0.31 0.85 0.90 0.45 0.50 16.2 47 5.86 30 1
3 2.8 0.20 1.10 0.90 0.45 0.50 21.5 33 4.67 43 2
4 2.3 0.13 0.98 0.85 0.36 0.42 37.7 43 13.69 30 2
5 2.3 0.27 0.83 0.74 0.31 0.41 21.2 58 4.75 50 2
6 3.7 0.40 1.15 0.75 0.22 0.29 59.9 61 20.61 132 3
7 4.3 0.46 1.18 1.04 0.27 0.26 45.2 50 28.94 130 3
The cats were ranked from least to most severe based on severity of the echocardiographic PSD/PVD ratio. RVDd/LVDd ¼ ratio of
right ventricular to left ventricular end-diastolic dimensions, RAD/LAD ¼ ratio of maximal right atrial diameter in long-axis to max-
imal left atrial diameter in long-axis (diastole), PVD ¼ pulmonary valve annulus diameter (systole), PSD ¼ pulmonary artery steno-
sis diameter (systole), PaS ¼ pulmonary artery stenosis peak systolic gradient, Spht ¼ pulmonary artery stenosis systolic pressure
half-time, PaD ¼ pulmonary artery stenosis peak diastolic gradient, Dpht ¼ pulmonary artery stenosis diastolic pressure half-time,
and DFA ¼ subjective grading of diastolic flow (0þ to 3þ). Weight (Wt) listed is that recorded at the time of initial diagnosis.
6. 88 D.P. Schrope, W.J. Kelch
Of the seven cats, six cats were mixed breed,
and no sex predilection was suggested (Table 1).
The age at the time of diagnosis for most of these
cats suggests a congenital lesion although a herita-
ble basis could not be assessed since the majority
were stray animals.
All seven cats presented with basilar systolic
murmurs although the point of maximal intensity
varied (Table 1). Only one cat with severe stenosis
and visible Doppler flow throughout diastole had
a diastolic murmur. In humans with pulmonary ar-
tery stenosis the presence of a diastolic murmur is
variable.1,2
Exertional dyspnea was seen in one of the cats
with severe stenosis. Exertional dyspnea is one of
the most common clinical signs identified in
humans with pulmonary artery stenosis, and is
likely related to hypoperfusion of the pulmonary
arterial tree.3 Based on the blood gas analysis, this
was also the likely cause for exertional dyspnea in
this cat. It is also possible that certain lesions
could result in asymmetric blood flow to branched
Figure 4 Selective right ventricular angiocardiogram pulmonary arteries further contributing to ventila-
from cat 4 with distal pulmonary artery stenosis. Opaci-
tioneperfusion mismatch.
fication of the right ventricle (RV), main pulmonary ar-
The available radiographs and ECGs did not
tery (MPa), and right- (RPa) and left (LPa)-branched
pulmonary arteries is evident. Mild opacification of the reveal specific findings that would differentiate
right atrium and the caudal vena cava (CVC) is also pres- pulmonary artery stenosis from other causes of
ent resulting from retrograde movement of the catheter basilar systolic murmurs. In fact, radiographic or
back into the right atrium during injection. Note the ste- electrocardiographic abnormalities were limited
nosis of the main pulmonary artery just proximal to the to cats with moderate to severe disease.
bifurcation (black arrow) with normal branched pulmo- Although PaS and PaD were correlated to dis-
nary arteries. ease severity, a ‘‘natural’’ cut-off between mod-
erate and severe disease was not obvious to the
arteries. Single lesions of the main pulmonary ar- authors (Table 2). Studies in human aortic coarcta-
tery or branched pulmonary arteries are classified tion support the value of evaluating other diastolic
as Type I pulmonary artery stenosis. Type I lesions flow parameters.6,8,9 In this study, Dpht and DFA
are further subclassified into Type Ia (isolated ste- were statistically significant and there appeared
nosis of the main pulmonary artery), Type Ib (iso- to be natural cut-offs between moderate and se-
lated stenosis of the right-branched pulmonary vere disease using Dpht and DFA. One human paper
artery), and Type Ic (isolated stenosis of the left- evaluating patients with aortic coarctation sug-
branched pulmonary artery). Isolated lesions lo- gested that a Dpht of 100 ms was consistent
cated at the bifurcation with extension into the with severe stenosis.8 Findings in this study suggest
proximal branched pulmonary arteries are classi- that a similar value may be appropriate in cats
fied as Type II. Lesions involving the peripheral with pulmonary artery stenosis. Both cats with
segmental pulmonary arteries with no abnormali- a Dpht 100 ms and a DFA of 3þ had the most se-
ties of the main pulmonary artery, bifurcation, or vere stenosis when graded by 2D echocardiography
left- or right-branched pulmonary arteries are and clinical signs.
classified as Type III. Lesions involving the periph- The discrepancy between systolic and diastolic
eral segmental pulmonary arteries with additional Doppler findings and obstruction severity in hu-
lesions in the main, left-branched, and/or right- mans with aortic coarctation is not clear. Causes
branched pulmonary arteries are classified as may be related to changes in cardiac output,
Type IV. Type Ia is one of the least common classes collateral blood flow, ductal flow, and the shape
of pulmonary artery stenosis seen in humans.1 In and length of the stenosis.6e8 Except for the mor-
contrast, the cats in this study showed no evidence phology of the stenosis, none of these would likely
of other concurrent cardiac defects and the lesion be a cause for similar findings in the pulmonary ar-
was classified as Type Ia in all cats. tery. It is possible that, with more severe stenosis,
7. Clinical and echocardiographic findings 89
the flexible walls of the main pulmonary artery 5. Liu SK. Pathology of feline heart disease. Vet Clin North Am
proximal to the obstruction absorb kinetic energy 1977;7:323e39.
6. Houston AB, Simpson IA, Pollock JC, Jamieson MP, Doig WB,
as the right ventricle contracts against the steno- Coleman EN. Doppler ultrasound in the assessment of sever-
sis. This could result in a lower peak systolic driv- ity of coarctation of the aorta and interruption of the aortic
ing force across the stenosis resulting in a lower arch. Br Heart J 1987;57:38e43.
relative gradient. 7. Marx GR, Allen HD. Accuracy and pitfalls of Doppler evalu-
The greatest limitations to this study are those ation of the pressure gradient in aortic coarctation. J Am
Coll Cardiol 1986;7:1379e85.
inherent to a retrospective study and the small 8. Carvalho JS, Redington AN, Shinebourne EA, Rigby ML,
number of cats available. Furthermore, grading of Gibson D. Continuous wave Doppler echocardiography and
severity was based on 2D echocardiographic find- coarctation of the aorta: gradients and flow patterns in
ings and clinical signs. Ideally, disease severity the assessment of severity. Br Heart J 1990;64:133e7.
would have been graded by angiocardiographic 9. Valdez-Cruz LM, Cayre RO. Coarctation of the aorta. In:
Valdez-Cruz LM, Cayre RO, editors. Echocardiographic
and intra-cardiac pressure data but these invasive diagnosis of congenital heart disease: an embryologic and
diagnostics were believed to not be in the best anatomic approach. Philadelphia: Lippincott-Raven; 1999.
interest for the majority of these cats. p. 475e82.
10. Houston AB, Sheldon CD, Simpson IA, Doig WB, Coleman EN.
The severity of pulmonary valve or artery obstruction in
children estimated by Doppler ultrasound. Eur Heart J
References 1985;6:786e90.
11. Valdes-Cruz LM, Horowitz S, Sahn DJ, Larson D, Oliveria
1. Gay BB, French RH, Shuford WH, Rogers Jr JV. The roent- Lima C, Mesel E. Validation of a Doppler echocardiographic
genologic features of single and multiple coarctations of method for calculating severity of discrete stenotic obstruc-
the pulmonary artery and branches. Am J Roentgenol Ra- tions in a canine preparation with a pulmonary arterial
dium Ther Nucl Med 1963;90:599e613. band. Circulation 1984;69:1177e81.
2. D’Cruz IA, Agustsson MH, Bicoff JP, Weinberg M, Arcilla RA. 12. Litster AL, Buchanan JW. Vertebral scale system to measure
Stenotic lesions of the pulmonary artery. Clinical and hemo- heart size in radiographs of cats. J Am Vet Med Assoc 2004;
dynamic findings in 84 cases. Am J Cardiol 1964 April:441e50. 216:210e4.
3. Kreutzer J, Landzberg MJ, Preminger TJ, Mandell VS, 13. Rishniw M, Erb HN. Evaluation of four 2-dimensional echo-
Treves ST, Reid LM, Lock JE. Isolated peripheral pulmonary cardiographic methods of assessing left atrial size in dogs.
artery stenosis in the adult. Circulation 1996;93:1417e23. J Vet Intern Med 2000;14:429e35.
4. Buchanan JW. Causes and prevalence of cardiovascular dis- 14. Boon JA. Acquired heart diseases. In: Boon JA, editor. Man-
ease. In: Kirk RW, Bonagura JD, editors. Kirk’s current veteri- ual of veterinary echocardiography. Baltimore: Williams
nary therapy XI. Philadelphia: WB Saunders; 1992. p. 647e55. Wilkins; 1998. p. 261e382.
Available online at www.sciencedirect.com