Thoracic Charles

V. Zwirewich,

MD

a

Sverre

Vedal,

MD

Solitary Pulmonary and Radiologic-Pathologic

round

solitary been or oval

chyma

in the

HE

has

atelectasis,

pulmonary.

Radiology

From

the

University

of

Canada requested

V5Z 1M9. January

fellowships gists. C

Address RSNA,

British

from

Columbia

of malignancy

significantly

benign

malignant

be-

AND

METHODS

and hospital

investigated

records

for

of 260 pa-

a solitary

nodule between June 1985 and April 1989 were reviewed. We defined a solitamy pulmonary nodule as a single pamen-

a

chymal opacity in phy, with margins

N.L.M.), General

graphs

the lung at radiograthat were distinct

measurement included only

complete

Medicine Hospital,

(S.V.), 855

CT

scan

of diameif (a) a of the

(b)the

for review; on a CT 9800

scanner

and

(R.R.M.),

W 12th

Pathology Ave.

Vancouver,

BC,

in

performed

this

(c)

through

study

or

Ninety-three

the

attempt

patients

tam-

high-res-

to com-

pare the performance of HRCT ventional CT in the assessment nary nodules. (48

and conof pulmo-

men

and

45

women) ranging in age from 44 to 83 years (mean, 64 years) fulfilled the inclusion criteria. The 93 patients were among 158 ing

studied in a recent series investigatthe prevalence of multicentric adeno-

carcinoma at our institution (10). A total of 96 nodules were studied; 90 patients had a single pulmonary nodule, while three patients had two nodules each, one each lung. All scans were

performed

at

1-cm

inter-

vals from the apexes to the lung bases with use of intravenously administered contrast material. Technical scan parametens

pulmo-

scans

Milwaukee); collimation

were

120

kVp,

140-200

mA,

2-second

scan time, 10-mm scan thickness, 40-cm field of view, and standard reconstruction algorithm. One or more additional 1.5mm collimation scans were performed through the nodule in each patient and reconstructed

with

use

of a high-spatial-

resolution algorithm and a 15-20-cm field of view. All scans were photographed at window and level settings appropriate for mediastinum (level 35 HU, width 450 HU) and lung parenchyma (level -690 to -600 HU, width 1,500 HU).

Two observers (C.V.Z., N.L.M.) who were unaware of the pathologic findings independently

reviewed

each

CT scan.

From the 1989 RSNA scientific assembly. Received November 20, 1990; revision 4, 1991; revision received January 24; accepted January 28. C.V.Z. supported by the

reprint 1991

in prevalence

and

nary

(C.V.Z., Vancouver

(7-9),

in

preoperative,

and

in bea few associa-

risk

enough to permit ten. Patients were

of Radiology

has been in dem-

tics

SUBJECTS

179:469-476

Departments

com-

(HRCT) effective

the presence of a variand internal charactemis-

chest was available scan was performed

I

Re-

sions.

60.236

1991;

(1).

correlation of the HRCT findings with pathologic specimens has been limited. The aim of the present study was to perform HRCT-pathobogic correlation on a series of surgically mesected nodules, emphasizing the pathologic correlates of a variety of edge and internal characteristics seen at CT, and to analyze which parame-

terms:

culosis,

nodule were available for review; (d) the nodule was surgically mesected; and (e) thorough pathologic examination of the inflation-fixed resected lung specimen was performed. We excluded patients in whom (a) multiple nodules were present in a single lung at CT, (1,) the nodule was surrounded by areas of pneumonitis or atelectasis, and (c) the CT scans demonstrated significant artifact from cardiac structures, respiratomy motion, surgical clips, or mediastinal calcification. Patients with single bilateral nodules were not excluded. We did not evaluate the plain chest radio-

high-resolution

differ

CT

olution

of adenopathy,

tion between ety of edge

tems

PhD

absence

onstrating diffuse calcification nign nodules (3-6). Although studies have documented an

between

MD,

(GE Medical Systems, geted, 1.5-mm-thick

tomography to be highly

the

L. Muller,

High-Resolution Correlation’

or pneumonia

and

#{149} Nestor

pulmonary nodule defined as a single lesion of lung paren-

Thin-section

puted shown

tients

Bronchiectasis, 60.26 a Coccidioidomycosis, 60.2052 a Computed tomography (CT), high-resolution, 60.1211 a Hamartoma, 60.314 a Lung, nodule, 60.2052, 60.281, 60.3221 Lung neoplasms, CT, 60.1211, 60.3221 #{149} Lung neoplasms, diagnosis, 60.1211, 60.3221 a Tuber-

MD

views of large series of mesected nodules indicate that while most solitary nodules are benign, approximately 40% are malignant (2).

CT scans Index

R. Miller,

Nodule:

T

Edge and internal characteristics of pulmonary nodules evaluated with high-resolution computed tomography (HRCT) were correlated with the pathologic specimens in 93 patients. Spiculation correlated pathologically with irregular fibrosis, localized lymphatic spread of tumor, or an infiltrative tumor growth pattern and was observed in six of 11 benign nodules (55%) and 74 of 85 malignant nodules (87%). Pleural tags were observed in three benign nodules (27%) and 49 malignant lesions (58%); pathologically, these represented fibrotic bands usually associated with juxtacicatricial pleural retraction. Bubblelike areas of low attenuation within the nodule were observed in 21 malignant lesions (25%) and only one benign nodule (9%). They were observed most commonly in bronchioloalveolar carcinomas (seven of 14) and were due either to patent small bronchi or small, cystic spaces within neoplastic glands. Malignant nodules as a group were larger than benign lesions (P = .02) and more commonly demonstrated a spiculated contour (P < .05), lobulation (P < .001), and inhomogeneous attenuation (P < .05).

#{149} Roberta

Radiology

British requests

Columbia to

Lung N.L.M.

Association

and

the

Canadian

Association

of Radiolo-

Abbreviations: carcinoma,

HRCT

BAC bronchioloalveolar high-resolution CT.

469

Observers

recorded

teristics and the

a number

of fommahin or Bouin fixative segment through a large-bore specimens were then sectioned

of charac-

of the nodule including location presence or absence of a variety

of margin and internal characteristics. Discrepancies in interpretation between observers were resolved by consensus. The

location

of the

pulmonary

vemsely

nodule

by lobe and within a specific lobe was mecorded. Nodules located within the peniphenal

two-thirds

of

the

lung

at 1.0-1.5-cm

same

plane

into each needle. The trans-

intervals

as a CT scan.

in

The

pearance of the nodule was each case, and microscopic examination of the pathologic material was performed for histologic

diagnosis.

were arbitrarily considered to be penipheral. Those within the central one-third or in contact with or infiltrating lobar or segmental bronchi were considered central.

measured

and

in each

internal

HRCT

case,

and

the

were

pathologic

then

margin

assessed

correlated

with

Nodule

Size

and

at

the

specimens.

em (P

= .02; 95% for the observed 1.31 cm, 0.30-2.32 benign nodules cell carcinomas

overall

mean

Margin

Nodule

margins

defined

were

if the nodule

distinctly

lung

Characteristics described

from

and

hazy or indistinct was considered

margins,

the

in patient

age, nodule

size,

and the prevalence of irregularity of the nodule margin, lobulation, homogeneous

surrounding

In the presence

of

attenuation,

nign

nodule

to be poorly defined. The presence of lobulation or notching of the contour was recorded. Observers also mecorded the presence of coarse spiculation, which was defined by the presence of 2mm or thicker linear strands extending from the nodule margin into lung parenchyma without contacting the pleural surface. A halo was said to be present if there was an area of prominent low attenuation or emphysema contacting and partially or completely surrounding the margins of a lesion, regardless of its location

and

and

with

Fisher

exact

observer.

and

among

nodules

use of the Student test

For

the

spiculation

malignant

pared

independently

each

Fisher

test

were comt test and for each

x2

observer,

exact

be-

were

analysis

used

to test

as the

lung

area of high attenuation that (a) was sunrounded by aerated lung, originated from the margin of the mass, and (c) ex-

(b)

tended

peripherally

to contact

the

pleural

Taylor in the tervals

in lesions

probability

Characteristics

neous

ence

Nodule

ther

of

attenuation

was

homogeneous

to be ei-

or mnhomogeneous;

inhomogeneous,

cavitation, bubblelike

noted

the

presence

if

of frank

central air bronchograms, or areas of low attenuation (pseuwas recorded. The latter is

docavitation) characterized

by

small,

low attenuation

(oval,

within

that

the

areas

of

or linear)

have been variably reported to represent small foci of cavitation, necrosis, bronchiectasis, or focal em-

physema sence

mass

scattered

round,

(8,9,12).

The presence

of calcification

within

the

cification

mass

fat

also

recorded.

was evident

characterized

(“popcorn” (central

attenuation

visually,

as diffuse,

ed, on globular distribution

was

or ab-

and

central,

type)

or eccentric)

If calit was laminat-

and its was

noted.

Pathologic

Study

All surgical specimens were the inflated state by transpleural

470

a

Radiology

fixed in infusion

the

series approximations were calculation of 95% confidence (13).

attenuation,

bobulation,

of calcification

were

used in-

analysis if patient homogeand

in the

next

left

Of the

three

series,

all

was

independently di-

NC).

common

lobe,

right

to the

The most commonly encountered neoplasms among the 85 malignant nodules were acinar adenocarcinomas (n = 29; 34%), squamous camcinomas (n = 24; 28%), and bronchioloalveolar carcinomas (BACs) (n 14; 16%) (Table 1). Inflammatory processes accounted for eight (73%) of the 11 benign nodules. The HRCT findings are summarized in Tables 2 and 3. Mean nodule size measured at pathologic sectioning and the prevalence of homogeneous internal attenuation, coarse spiculation, and bobulation differed significantly between

nodules (91%)

peripheral

twolocated zones.

lesions

peripheral

subpleumal

one lobe.

BACs were peripheral

metastatic

were

and

lower

in this

and

in location.

one

Neither

the location of a nodule by lung lobe nor its position within a lobe differed significantly between benign and malignant lesions or among the different types of neoplasms.

Margin

A sharp, tween the

Characteristics

well-defined nodule and

lung was observed frequency among and benign (82%)

An RESULTS

nodules

lobes.

most

upper

in the

localized

Nodule

pres-

associated with a benign or malignant agnosis (14). Analyses were performed with use of Statistical Analysis System computing software (SAS Institute,

Carey,

CT

divided by the probability of lung in nodules lacking that feature.

was

Internal Nodules

of

exhibiting

Multiple logistic regression performed to determine age, nodule size, spiculation,

surface.

±

25% of malignant and lesions. Tubercubous found exclusively in the Two metastases were lo-

thirds of the lung. exclusively in the

teristic

feature cancer

cm

(32%) were located lobe and 30 (31%) lobe. The right bow-

the

located

were

mm of the dominant tail or tag at CT was

is defined

was

site, harboring 18% of benign lesions were upper lobes.

upper

Seventy-five malignant (88%) and 10 benign lesions

characteristics that differed in prevalence between the malignant nodule groups. risk of malignancy for each

calculated. The relative for a given CT charac-

cancer

(1.7 of all

in the

Thirty-one lesions in the right upper in the left upper

was

of these was then risk of malignancy

A pleural as a linear

smallest

percent

located

cated

(11). Satellite lesions were defined as one or more distinctly separate nodular areas of high attenuation observed within 5 nodule. defined

(5.7 cm ± 2.3) and

the

Sixty-three

were

em lobe

for differences in nodule characteristics among malignant nodules of different cell types. The P value and 95% confidence intervals were also calculated for those edge and internal significantly benign and The relative

confidence interval mean difference of cm) than that of (2.5 cm ± 1.2). Large showed the greatest

1.1).

Analysis

Differences

as well

was sharply

separated

parenchyma.

Statistical

for

Location

size

hamartomas

Nodule

groups 4).

The mean diameter of malignant lesions (mean ± standard deviation 3.8 cm ± 1 .7) was significantly great-

All specimens

characteristics

malignant (Table

the

gross apevaluated in

were prosected by a single pathologist (R.R.M.). The maximum nodule size was

at CT

benign and each observer

ill-defined

one of the pseudotumor

cases

interface surrounding

with almost equal malignant (85%) lesions at HRCT.

margin two

be-

was

seen

in

cases of inflammatory and one of the five

of tuberculosis,

where

patho-

logically pneumonia Among margins primary

it was due to organizing at the edge of the lesion. malignant lesions, ill-defined were observed in all types of carcinomas in which an in-

regular

infiltrative

edge

“pushing” edge was ically (Fig 1). All three sions had well-defined Coarse spiculation bobulation (P < .001) significantly higher

among The the

malignant

relative presence

rather

than

found pathobogmetastatic bemargins. (P < .05) and occurred with a frequency

nodules

(Table

risk of malignancy of coarse spiculation

4). in

May

1991

a

was 1.29 1.06-1.57), bobulation

(95% confidence and in the it was 2.07

dence

interval,

tion 90%

was observed of primary

bronchovasculam lymphangitic

interval, presence of (95% confi-

1.08-3.96).

fibrotic response. Lobulation of the

two

and metastatic. late of lobulation lesions was the excrescences

Solitary

Pulmonary

in 96 R esected Nodules

Malignant

(7: = 85) Acinar adenocarcinoma Squamous carcinoma BAC Large cell carcinoma Neuroendocnine carcinoma* Metastasist Benign (n = 11) Active tuberculosis Inflammatory pseudotumor Chondroid hamartoma Coccidioidomycosis Cystic bronchiectasis1

Category

includes

and

small cell One each myosarcoma. I Fluid-filled nodule in this t

atypical

29 24 14 9 6 3

(n

in 85 Resected

at its ad-

of the

benign

he-

dently sions.

three

cases.

with

a benign identified

appeared appearance

Pleural

tags

a desmoplastic

or malignant with use

to in

also

me-

attenuation HRCT images

cummed with a significantly frequency among benign malignant (20%) lesions ative risk of benignancy

associated For one

with observer,

spiculation

lesion of a multi-

age were

and the indepen-

malignant lethe presence

was

also

1.57).

confidence Sixty-eight

demonstrated uation on

HRCT

Among BACs, low attenuation correlate with

analysis. Reboth observers

that increasing of bobulation

of coarse

inde-

pendently associated with mahignancy. An increased nodule diameter was no longer significantly associat-

Malignant

Homogeneous nodule on

of

within oc-

greater (55%) than (P < .05; mel1.25 with

to the presence of bubblehike areas of low attenuation (Fig 5), air bronchograms, frank cavitation, or apparent tumor necrosis without cavitation (Fig 6). Pathologically, bubblelike aneas of low attenuation were due to small, patent air-containing bronchi within the nodule in some cases.

showed presence 4)

Characteristics

of cartilage, respectively, account for the HRCT

pie logistic regression gression models for

carcinoma (n 2). of colonic adenocarcinoma, leioand melanoma. bronchi simulated a pulmonary case.

Table 2 HRCT Findings

tumor

the

after acof bobula-

a 95%

with were

5 2 2 1 1 carcinoid

associatprimary

sponse to the nodule and were observed most commonly in malignant lesions (58%), but were also present in three of the 1 1 benign nodules (27%) (Fig 4). Pleural tags were not observed in pulmonary metastases. Patient and nodule characteristics that were independently associated

No.of Cases

Diagnosis

Internal Nodules

contour

pathologic comeamong malignant presence of nodular

Three

a malignant lesion for the presence

sions (27%) displayed lobulated bondens, including two tubencubous hesions and one hamamtoma. Coalescent granubomas and coalescent nodules

these

a

edge.

correlated Diagnoses

nodule

The

of the

vancing

ed with counting tion.

not distindue to a

was a feature predominantly ed with malignancy, both

metastases, but was also present in four of the five tubercubous lesions and in one of the two inflammatory pseudotumors. Pathologically, spiculation most commonly correlated with a desmoplastic response in the nodule, resulting in fibrotic strands radiating into the surrounding lung panenchyma (Fig 2). Among mabignant lesions, spiculation was occasionally associated with direct infiltmation of the tumor into adjacent

Table 1 Pathologic

or localized (Fig 3); on

HRCT images, these were guishable from spiculation

Spicuba-

in approximately carcinomas and

sheaths extension

interval malignancies inhomogeneous images,

of 1.01(80%) due

atteneither

bubblelike areas of were also found to small air-containing

cystic spaces associated with papillary tumor architecture (Fig 5). Bubblehike areas of low attenuation were observed more frequently among BACs (50%) than among any other malignant lesions (P < .05; relative risk = 2.67 with 95% confidence in-

terval

of 1.08-6.58).

of low attenuation but were occasionally bronchioboalveolam in benign lesions.

HRCT

was

Bubblehike

areas

were

uncommon seen in nonmalignancies and Frank cavitation at

demonstrated

exclusively

Nodules Diagnosis

Acinar Adenocarcinoma

Parameter Patientsex(M/F) Mean patient age (y) Mean diameter (cm)

BAC

Squamous Carcinoma

Large Cell Carcinoma

Total

1/2 60 3.0

44/38 65 3.8

6/7 66 3.1

16/7 69 4.1

24 (83) 5(17)

14 (100) 0

22 (92) 2(8)

7 (78) 2(22)

5 (83) 1(17)

3 (100) 0

75 (88) 10(12)

Poonlydefined Smooth

25 (86) 4(14) 0

13 (93) 1(7) 0

19 (79) 5(21) 1(4)

7 (78) 2(22) 1(11)

5 (83) 1(17) 1(17)

3 (100) 0 1(33)

72 (85) 13(15) 4(5)

Irregular

29 (100)

14 (100)

23 (96)

8 (89)

5 (83)

2 (67)

81 (95)

Spiculation Lobulation Pleural tag

27 27 20 0 2

(7)

13 (93) 13 (93) 1 1 (79) 1(7) 0

21 (87) 23 (96) 15 (62) 1(4) 0

6 (67) 8 (89) 2 (22) 0 0

5 (83) 6 (100) 1 (17) 0 0

2 (67) 2 (67) 0 0 0

74 (87) 79 (93) 49 (58) 2(2) 2(2)

7 (24) 9 (31) 1 (3) 4(14) 2 (7)

4 (29) 7 (50) 0 0 1 (7)

3 (12) 3 (12) 0 2(8) 1 (4)

0 1 (1 1) 1 (11) 0 0

1 (17) 1 (17) 0 0 0

2 (67) 0 0 0 0

17(20) 21 (25) 2(2) 6(7) 4 (5)

location

Peripheral

Central Nodule

margin

Well defined

Halo Satellites Internal features Homogeneous attenuation Bubblelike areas of low Air bronchogram Cavitation Calcification Note-Data

Volume

1/5 68 3.6

Metastasis

14/14 62 3.5

Nodule

6/3 62 5.7

Neuroendocrine Carcinoma

arrived

179

a

attenuation

at by consensus

Number

2

between

(93) (93) (69)

two observers;

numbers

in parentheses

are

percentages.

Radiology

a

471

Table

3

HRCT

in 11 Resected

Findings

Benign

Nodules Diagnosis

Active Parameter Patientsex(M/F) Mean patient age (y) Mean diameter (cm)

Nodule

Inflammatory

Tuberculosis

Cystic

Pseudotumor

2/3 54 2.3

Hamartoma

1/1 58 3.9

Coccidioidomycosis

1/1 58 1.7

Bronchiectasis

0/1 71 1.5

Central Nodule

defined

Poorly Smooth

defined

Irregular Spiculation

Lobuhation Pleuraltag Halo Satellites

Internal

attenuation

Bubblehike areas Air bronchogram Cavitation Calcification Note-Data

oflow

arrived

attenuation

at by consensus

between

in malignant nodules and was common in acinar adenocarcinomas (14%) than squamous carcinomas

2 (100)

2 (100)

1 (100)

0

0

0

0

1 (100)

1 (9)

4 (80) 1 (20)

1 (50) 1 (50)

2 (100) 0 2 (100) 0 0 1 (50) 0 0 0

1 (100) 0 1 (100) 0 1 (100) 0 0 0 0

1 (100) 0 1 (100) 0 0 0 1(100) 0 0

9(82) 2(18) 4(36) 7 (64) 6(55) 3(27) 3(27) 0 0

(100)

1 (100) 0 0 0 1 (100)

6 (55) (9)

0

0

5 (100) 4 (80) 2 (40) 1(20)

2 (100) 1 (50) 0 1(50)

0

0

0

0

3 (60) 0 0 0 0

0 1 (50) 0 0 0

two observers.

more

Numbers

Table

(50) 0 0 0 u

in parentheses

4 That

Parameters

Differed

Significantly

between

Benign

Nodules Parameten*

Mean

Lobulation spiculation Homogeneous attenuation Coarse

bronchial wall. Fat attenuation was not observed in either of the two hamartomas. The multiple logistic regression models showed that, for both observems, neither calcification nor homogeattenuation after

was

with

the

accounting

and the presence nodule contour.

for

diag-

patient

of lobulation

age

in the

DISCUSSION

in a solitary

cabby nodule

associatof ma-

pulmonary

advancing age, increasing size, and a history of smoking The radiologic features classiassociated include

with benignancy absence of growth

in a

over a 2-year period (16) and presence of a diffuse, laminated, central, or popcorn pattern of calcification (17).

In a large

however,

proportion

previous

not

available

472

a

Radiology

to assess

a

of patients,

radiogmaphs the

mate

For all other

t Taken

are of

from

in parentheses

parameters, measurements

growth,

standard

attenuation

=

characteristic

benign

tomography

of lung

in assess-

ing

the

nodules

(3). 384

In a cooperative series evaluating nodules, Zemhouni et al con-

firmed that combined use of a calibrated reference phantom and thincollimation CT can confidently allow identification of 31% of apparently noncalcified nodules on plain radiographs as benign (6). While quantitative CT densitometmy is undoubtedly of value in the assessment of nodules in geographic areas where granulomatous diseases are endemic, this technique is of limited use in regions

Nodules

11)

P

2.5

3.8

5 (45) 2 (18) 7 (64)

80 (94) 58 (68) 21 (25)