Allan

R. Bnitt,

PhD,

MD

e

Sarcoidosis:

Isaac

NHL

and

discrete

in sancoidosis.

Hepatomegaly was seen in six of the i6 patients (38%) with sancoidosis and splenomegaly was present in nine of i5 (60%). CT depicted hepatic lesions in only three of eight patients (38%) with biopsy-proved hepatic involvement. Splenic lesions were seen at CT in five of the 15 patients (33%). The authors believe that the overlap in nodal appearance and distribution poses a limitation for use of these criteria in accurate disease characterization. Index

terms:

80.1211 792.34, 775.1211 87.22

Radiology

MD

e

Gary

Abdominal

There are few data in the literature on the abdominal manifestations of sancoidosis at computed tomognaphy (CT). To determine whether differences in nodal distribution and appearance can be reliably used to distinguish between sarcoidosis and non-Hodgkin lymphoma (NHL), the authors retrospectively neviewed the abdominal and pelvic CT scans of i6 patients with biopsyproved sancoidosis and 20 patients with biopsy-proved NHL. Eleven of the 16 patients with sancoidosis had abdominal and/on pelvic lymphadenopathy, which was common at all nodal sites except for the netnocrunal and pelvic locations. There was a statistically significant lower fnequency of retnocrunal adenopathy in sancoidosis than in NHL. Mean nodal size was significantly greaten in NHL. Nodes tended to be confluent in

R. Francis,

Lymphatic system, CT, 70.1211, Lymphatic system, diseases, 792.2.2, 87.22, 87.34 #{149} Lymphoma, CT, 761.1211, #{149} Sarcoidosis, 761.22, 775.22, 792.22, #{149}

1991; 178:91-94

I From the Department of Radiology, Univensity of Michigan Hospitals, 1500 E Medical Center Dr. Ann Arbor, MI 48109-0030. From the 1989 RSNA scientific assembly. Received February 1, 1990; revision requested March 28; revision received August 6; accepted August 14. Address reprint requests to I.R.F. 2 Current address: Stanford University School of Medicine, Palo Alto, Calif. ©RSNA, 1991

Glazer,

M.

MD2

ARCOIDOSIS is a systemic (1,2) that predominantly the thorax. However, even

common

abdominal

disease affects the less

manifestations

can be visualized pelvic computed

on abdominal tomographic

and (CT)

scans. To our knowledge, published data regarding the CT findings in abdominal sarcoidosis have been limited to case incidental

reports findings

macic sarcoidosis inal

study,

or descriptions in studies

(3-10). we

manifestations

CT and

contrast

PATIENTS

of of tho-

In this

report

the

metroabdom-

of sarcoidosis the

nodal

ance and distribution with in non-Hodgkin hymphoma

AND

at

appear-

those seen (NHL).

METHODS

The abdominal and/or pelvic CT scans of 16 patients with biopsy-proved sarcoidosis and 20 patients with biopsyproved NHL were retrospectively re-

viewed.

James

H. Ellis,

Manifestations

S

spective

e

Three

of the authors

(A.R.B.,

G.M.G., I.R.F.) reviewed the CT scans from both groups. At the time of analysis, the investigators were cognizant of the diagnosis for each patient, did not refer to other imaging studies, and arrived at their conclusions by consensus. Nodal sizes and patterns of distribution (which included retrocrural, porta hepatis, gastrohepatic ligament, cebiac artery axis, superior mesentenic artery axis, paraaortic/ paracaval, mesenteric, and pelvic sites) were determined. Lymph nodes were considered to be enlarged if their shortest axis measured greater than 1 .0 cm; in the retrocrural region, nodes greater than 0.6 cm were considered enlarged. Where individual lymph nodes could be discerned, the largest lymph node was measured in its two largest dimensions and the average was calculated. When it was not possible to distinguish individual lymph nodes, the largest lymph node mass was measured in its two largest dimensions and the average was calculated. Patterns of visceral involvement in sarcoidosis were also evaluated. All 16 patients with sarcoidosis were referred for abdominal and/or pelvic CT scanning. Seven of the 16 patients were studied before a diagnosis had been made

MD

at

CT’

(biopsy-proved sarcoidosis was established subsequently). The clinical indications for scanning were suspected abscess and/or sepsis (n 2) or suspected malignancy (n = 5). The remaining nine patients had biopsy-proved sarcoidosis (mean duration, 4.3 years; range, 1-10 years) and underwent abdominal and/or pelvic CT examination because of symptoms that could not be attributed to samcoidosis alone. The clinical indications for scanning included suspected abscess and/ or sepsis (n = 4), suspected malignancy (n = 2), and miscellaneous (n 3). Five of these nine patients were being treated with steroids; the remaining four were untreated. Each patient with sarcoidosis had biopsy proof of the disease established from at least one site and in most cases from two or more sites; these included transbronchial nodal biopsy (n 8), and biopsy of hepatic nodes (n 8), paratracheal nodes (n 4), cervical nodes (n 4), bone marrow (n 2), spleen (n 1), retroperitoneal nodes (n = i), and pleura (n = 1). All biopsy results were negative for tuberculosis and fungi. The NHL group was arbitrarily chosen from a group of patients who had abdominal and/or pelvic lymphadenopathy demonstrated on CT scans obtained for staging either at the time of initial presentation or at the time of relapse of known and proved NHL. Nine of the 16 patients with sarcoidosis underwent both abdominal and pelvic

CT, while the remaining seven underwent only abdominal CT. In 15 of the patients who underwent abdominal CT, 10mm-thick contiguous axial scans were obtamed; 10-mm-thick sections were obtained at 15-mm-intervals in the remaining patient. Four of the nine patients who underwent pelvic CT were scanned with 10-mm-thick contiguous axial sections, two were scanned with iO-mmthick sections at i5-mm intervals, and three with 10-mm-thick sections at 20mm intervals. CT scans of nine of the patients were obtained with use of a bolus of contrast material, one patient was scanned with use of a drip infusion of contrast material, and six patients were

Abbreviation:

NHL

non-Hodgkin

lympho-

ma.

9i

scanned trast

without

use

of intravenous

con-

Table i Distribution

material.

Sixteen underwent

of the

20 patients

abdominal

and

with

NHL

pelvic

CT;

Hodgkin

four underwent only abdominal CT. Contiguous 10-mm-thick axial sections were obtained in all patients except one; in that patient, the abdomen was scanned with 8mm-thick sections at 12-mm intervals. The scans were obtained with use of a bobus of contrast material in 16 patients, with drip infusion of intravenous contrast material in two, and without use of intravenous contrast material in two. Chest radiographs obtained within 2 weeks of the CT scans were available for

14 of the patients

with

sarcoidosis.

frequency

of enlarged

fined anatomic confluent nodal osis

and

NHL.

lymph

nodes

in Patients No.

with

with

Lymph

Node

Site

and Non-

of Patients

Non-Hodgkin Lymphoma

ii)

(n

Sarcoidosis

Lymphadenopathy

Sarcoidosis

The

radiographs were graded by two observens (A.R.B., I.R.F.) for the presence or absence of parenchymab and nodal disease by means of the criteria of Pare and Fraser (ii). The Fisher exact probability test was used to test statistical significance of the

of Lymphadenopathy Lymphoma

20)

(n

Significance

Retrocrural Portahepatis Gastrohepatic ligament Celiac artery axis Superior mesenteric artery axis Paraaortic/

2 (18) 8(73)

14 (70)

6 (55) 9 (82)

10 (50) 13 (65)

NS NS

5 (45)

13 (65)

NS

paracaval Mesenteric Pelvic

8 (73) 6 (55) 3 (33)t

20 (100) 13 (65)

P < .04*

16 (89)

P < .02*

Note.-NS = not significant at the 95% confidence level. S Not significant when Bonferroni correction was used level). t Includes

one

abdominal

CT scan

with

visualization

P < .001 NS

11(55)

Numbers (requiring

of the

NS

in parentheses are percentages. P < .006 at the 95% confidence

common

iliac

at de-

pelvic scan. Excludes two patients without pelvic scans. I Includes two abdominal CT scans with visualization of the common complete pelvic scan. Excludes two patients without pelvic scans.

lymph

nodes

but

without

nodes

but without

a

complete

iliac lymph

a

sites and the frequency of masses between sarcoidThe

Bonferroni

correction

was used to correct for the number of comparisons being made simultaneously (eg, P value for null hypothesis at 95% confidence bevel is .006 [0.05/8]). The Student t test was used to test statistical significance between the mean lymph node sizes in these two diseases.

Table

2

Morphologic

of Lymph

Features

size (cm) Mean Range Confluent mass

Nodes

in Patients

with

Lymphadenopathy

Sarcoidosis (n 11)

Non-Hodgkin Lymphoma (n 20)

Significance

2.6 ± 1.7 i.i-7.5 1 (9)

8.0 ± 5.5 3.0-17.0 10 (50)

P < .03

Node

RESULTS Table tion osis with

i summarizes

of hymphadenopathy and NHL. Eleven sarcoidosis had

dominal nopathy without cluded paring

athy

the

distnibu-

in sarcoidof the patients evidence of ab-

Note.-Numbers

and/on pelvic lymphadeat CT. The five patients lymphadenopathy from statistical the distribution

in sancoidosis

were analysis of adenop-

and

NHL.

node

All 20

significant lower frequency of metrocrural (P < .05), pamaaontic/paracaval (P < .05), and pelvic lymphadenopathy (P < .05) in patients with sarcoidosis. However, results from the Bon-

lower

frequency

of

lymphadenopathy in sarcoidosis. The differences in distributions for the two diseases were not significant (P > .006) at other lymph node sites. Table 2 compares some momphohogic features of enlarged lymph nodes in sarcoidosis and NHL. The mean size of lymph nodes in NHL (8.0 cm ± 5.5) was significantly larger than that in sarcoidosis (2.6 cm ± 1 .7) (P < 92

e

Radiology

masses

nificantly

in NHL higher

act probability

percentages.

(50%)

was

(P < .05,

Fisher

test)

than

sigex-

in sarcoid-

osis

(9%). Figure 3 illustrates the number of lymph node sites involved in mdividual patients with samcoidosis and NHL. NHL involved four or more sites in 85% of patients, while sancoidosis was nearly equally distnibut-

ed, with

fenroni correction for the number of comparisons being made simultaneously showed that only the metrocrural region had a statistically sig(P < .006)

are

.01, Student t test) (Fig 1). There was some overlap in the range of nodal sizes between the two diseases (Fig 2). The frequency of confluent lymph

excom-

patients with NHL had abdominal and/or pelvic hymphadenopathy. Results from the Fisher exact pmobabihity test demonstrated a statistically

nificant

in parentheses

55% of patients

showing

in-

volvement of four on more 45% showing involvement sites or fewer. All but one

sites and of three patient in

each

a single

group

showed

either

P < .01

site or contiguous nopathy. In both

sites of hymphadeof these patients,

external iliac and val hymphadenopathy

paraaortic/pamacawere present

without interposed common iliac hymphadenopathy. Lymphadenopathy was uniform at all lymph node sites in all but one case in each group. Table 3 summarizes the hepatic

and splenic involvement in sarcoidosis. CT was not sensitive for detecting samcoid hepatic granulomas, as hesions were detected in only three of eight patients (38%) with biopsyproved hepatic involvement. The lesions that were detected at CT were usually seen as small focal lesions distributed throughout the hepatic pamenchyma (Fig 4). Splenic lesions

were

seen

at CT in five

of the

15 pa-

tients (33%) (one patient had previously undergone sphenectomy). Only one patient underwent splenectomy after CT; splenic samcoidosis was

proved

in this

patient.

Ten of 14 patients who underwent chest radiography had evidence of sarcoidosis (stage I 3, stage II 4, stage III = 2, and stage IV = 1) and concomitant abdominal disease as detemmined at CT, three had radio-

graphic

evidence

of thoracic

involve-

ment (stage I) without CT evidence of abdominal involvement, and one had CT evidence of abdominal disease without radiographic evidence of sarcoidosis (stage 0). Of the two patients with sancoidosis in whom

January

1991

b. Figure

1.

enopathy (arrowheads)

CT scans

show

in non-Hodgkin in sarcoidosis.

(a) barge,

confluent

lymphoma

and

paraaortic/paracavab (b)

small,

discrete

and

mesenteric

mesentenic

lymphad-

lymphadenopathy

90% of patients ifestations have

b. Figure 2. (a) CT scans show bulky, confluent porta hepatis, celiac axis, and paraaortic/paracaval lymphadenopathy in sarcoidosis that resembles non-Hodgkin bymphoma. Ascites and spbenomegaly are also present. (b) Small, discrete mesenteric lymphadenopathy (arrowheads) is seen in non-Hodgkin bymphoma that resembles sarcoidosis.

..c

in the radiology literature (10-13). However, extmathoracic manifestations, particularly infradiaphragmatic involvement, are also common and may be the sole or predominant feature (2), an aspect that has not been reported in depth or detail in the imaging literature. The distribution of abdominal lymphadenopathy in sarcoidosis has been studied in cadavers by Iwai and Oka (14). They meported intraabdominal lymph node involvement in nine of 10 autopsy cases, with the nodal involvement mainly clustered in the upper abdomen. Mesentenic nodal involvement was not seen in these cases, although this finding was frequent in patients

in our

study

morphology Figure 3. Number of sites involved in mdividuat sarcoidosis and non-Hodgkin bymphoma patients. Percentages for patients with sarcoidosis do not add up to 100% due to rounding.

chest madiogmaphs for review, one

thy.

had

were no

samcoidosis abdominal

Theme

was

no

not available evidence of

at CT, and lymphadenopacorrelation

be-

of chest abnonmahiand abdominal in-

volvement

Theme

at CT.

was

difference in the degree ity on chest radiographs

Volume

whose diagnosed

178

has

been

also

Number

Figure bobus proved small small

4.

1

CT scan

obtained

with

use of a

of contrast material shows biopsyhepatic sarcoidosis demonstrated as focal lesions. The spleen also contains focal lesions; no biopsy was per-

formed.

the diagnosis had already been estabhished, some of whom were being treated for the disease.

DISCUSSION Sarcoidosis involves

is a systemic the thorax

and

coidosis

group,

(44%)

group group

as there in both

NHL

groups.

only

were

evidence of the ings of abdominal

at CT, the than

and

seven

in The a

was the

a sesam-

In the

sam-

of the

scanned

pa-

at the

time

was arbitrarily selected from of patients with biopsy-proved

sancoidosis disease in more

findto deter-

of presentation. The other nine were scanned during varying phases in the course of the disease. The NHL

enopathy. Although

no

that

population, bias involved

coidosis tients

me-

(8).

be identified with these two diseases. in this study represent

patients skewed hection

of abnormalbetween pa-

sancoidosis had not yet and those in whom

#{149}

and

could

patients

the

tween the degree ty at radiography

tients been

(55%)

ported in a prior case report We compared the abdominal ings in sarcoidosis and NHL mine whether any distinguishing features in nodal distribution

LELhJ

abdominal other had

(2). The thomacic manbeen well described

scanning

disease and CT and/or pelvic

31% of the did

clinical in this

not

with

adenopathy

indications group

findad-

patients

show

a

were

for not

Radiology

dis#{149} 93

similar from those in the group adenopathy. A higher frequency enlarged abdominal nodes was ported in an autopsy study (14).

with of me-

Moreover, theme was a wide range in the distribution and size of lymph nodes in our study. For these reasons,

we believe

it is unlikely

that

a partic-

ular subset of adenopathy patterns was selected for CT scanning by the clinical indications. Despite the overlap of findings in the two diseases, we noted that metrocrural involvement was common in

NHL

but

was

atypical

in sarcoidosis,

being seen in only two patients in our study. Retrocrural lymph node involvement has been previously meported in a case of sarcoidosis (5). To a lesser extent, pelvic and paraaontic/ paracaval adenopathy were also more frequently seen in NHL in our study. Pelvic lymphadenopathy, though relatively uncommon, has also been reported in sarcoidosis (4,10); we observed this feature in three of nine patients in whom enough of the pelvis was scanned to make a detemmination. Mean nodal size was also useful to distinguish the two diseases; the nodes in NHL tended to be larger than those seen in sarcoidosis. As has been reported previously (15), this study showed large, confluent lymph nodes in NHL, in contrast to the smaller, discrete nodes seen in samcoidosis. Hepatomegaly and splenomegaly in samcoidosis as determined by means of CT have been previously described (3,5,6,9). Enlargement of liver and/or spleen was seen in 1 1 of our 16 patients (69%). Autopsy stud-

ies have

revealed

that

the

prevalence

of sarcoid and spleen

granulomas in the is approximately

However,

in patients

with

sarcoidosis, results of splenic biopsies have involvement in only There are few data detection with CT of

sions

of sarcoidosis.

liver 70% (ii).

suspected

fine-needle demonstrated 24% (16). regarding the focal hepatic he-

The

involved

liv-

em appeared normal in two case meports, one describing a patient scanned with drip infusion of contrast material (5) and the other describing a patient scanned with use of a bolus of contrast material (6). A third case report described multiple small low-attenuation areas in the liver and spleen on a bolus contrastenhanced CT scan (i7). In our study, these hepatic sarcoid gmanulomas were usually in the form of small fo-

94

Radiology

#{149}

cal

low-attenuation

throughout

lesions

the

nonspecific

liver

appearance

hates that croabscesses

scattered

at CT. This that

of disseminated such as those

is a

was

subsequently

proved

crete

to be

of contrast

material.

ing

four patients were of an intravenously

The

remain-

material

(6,9,17).

involvement

tients

(33%)

may

involvement

data

not

suggesting

of 15 pa-

reflect

the

because

histologic

help

abdominal concomitant

appreciated.

diagnosis

should

be considered

abdominal adenopathy and/on splenic focal at CT. Hepatosplenomegaly

mon

in sarcoidosis,

when

Sarcoid

4.

hymphadenopathy

Parker,

Kenneth

MD,

for

E. Cuire

for

analysis.

RC.

Sarcoidosis.

In:

Braunwald

medicine.

6.

7.

11th

ed.

New

McGraw-Hill, Lynch JP III.

1987; 1445-1450. Extrapulmonary

of sarcoidosis.

I. Intern

5:163-189. Miller LK, Rochester

E, Is-

Med

York:

manifestations Specialist

D, Miller

lymphadenopathy

1984;

JW.

Extensive

in sarcoidosis.

Am J Castroenterol 1981; 75:367-369. Duszlak EJ Jr. Costello P. Crossan AW, Ctouse ME. Pelvic sarcoidosis. J Comput Assist Tomogr

9.

1 1.

1982; 6:1032-1033.

Meranze 5, Coleman B, Anger P. Mintz M, Markowitz L. Retropenitoneal manifestations of sarcoidosis on computed tomography. J Comput Assist Tomogr 1985; 9:50-52. Mathieu D, Vandestigel M, Schaeffer A, Vasile N. Computed tomography of splenic sarcoidosis. J Comput Assist Tomogr 1986; 10:679-680. Deutch SJ, Sandler MA, Alpern MB. Abdominal lymphadenopathy in benign diseases: CT detection. Radiology 1987; 163:335-338. Saksouk FA, Haddad MC. Detection of mesenteric involvement in sarcoidosis using computed tomography. Br J Radiol 1987;

can

Alvarez PJ, Esalada J, Comet R, Carcia F. Biosca M. CT diagnosis of splenic multifocal lesions

and

ondary

to sarcoidosis.

Craph Hamper

abdominal

lymphadenopathy

Comput

t988; 12:255-257. UM, Fishman

EK,

sec-

Med

Khouri

Imaging NF,

Johns

CJ, Wany KP, Siegelman 55. Typical and atypical CT manifestations of pulmonary sarcoidosis. J Comput Assist Tomogr 1986; 10:928-936. Pare JAP, Fraser RC. Diseases of the chest of unknown origin. In: Synopsis of diseases of the chest. 2nd ed. Philadelphia: Saunders, 1979; 1658-1690.

12. 13.

14.

15.

Muller NL, Kutlnig P. Miller RR. The CT findings of pulmonary sarcoidosis: analysis of 25 patients. AJR 1989; 152:1179-1182. Rockoff SD, Rohatgi PK. Unusual manifestations of thoracic sarcoidosis. AJR 1985; 144:513-528. Iwai K, Oka

18.

Sarcoidosis:

report

of ten

au-

Assist

Tomogr

1983;

7:846-850.

Taavitsainen M, Koivuniemi A, Helminen J, et al. Aspiration biopsy of the spleen in patients

17.

H.

topsy cases in Japan. Am Rev Respir Dis 1964; 90:612-622. Neumann CH, Robert NJ, Canellos C, Rosenthal D. Computed tomography of the abdomen and pelvis in non-Hodgkin lymphoma. Comput

16.

CT is a rela-

occur at any nodal site in the abdomen or pelvis. However, retmocrumal adenopathy occurs commonly in NHL and infrequently in sarcoidosis, making it the best single anatomic

and

statistical

abdominal

this

tivehy poor modality to use for the detection of focal hepatic sarcoid he-

sions.

Crystal

internal

2.

in-

on hepatic lesions are seen is com-

but

Timothy

a case,

Joa list of

selbacher KJ, Petersdorf RC, Wilson JD, Martin JB, Fauci AS, eds. Harrison’s principles of

10.

sarcoidosis thoracic

Therefore,

patients, the

The authors thank for contributing

III, MD,

60:1135-1136.

proof

volvement, either in the form of pulmonary pamenchymal or nodal (mediastinal or hilar) involvement, theme was no correlation between the degree of thoracic and abdominal involvement. In conclusion, mnfradiaphragmatic sarcoidosis is more common than generally

in

mdi-

U

References

true

of the presence or absence of disease was available in only one patient. Although the majority of our pa-

tients with also showed

P. Lynch

contributing

8.

in five

characterize

patient.

sarcoidosis

of contrast

Our

Nevertheless,

Acknowledgments: seph

5.

splenic

is a hess dis-

feature.

the

ficult to detect at CT because of their small size, as suggested by Nakata et al (i7).

of a bolus

of nodes

tending on dis-

ty to reliably

em than that detected at CT, even with the use of newer scanners and bolus-enhancement techniques. Some sarcoid granulomas may be dif-

ministration

nature

vidual

1.

sarcoidosis has also been in three case reports as foof low attenuation after ad-

discrimina-

in nodal in samtheir abihi-

3.

Splenic described cal areas

significant

theme is sufficient overlap appearance and distribution coidosis and NHL to limit

scanned with administered

use bolus. This suggests that the frequency with which sarcoid noncaseating granubomas involve the liven is great-

between size is also

ton, with the nodes in NHL to be hanger. The confluent cniminating

secondary to the noncaseating granulomas of sarcoidosis. Two of these three patients underwent CT with use of a bolus of contrast material, while the remaining patient was scanned without use of contrast materiah. Of the five patients with biopsy-proved hepatic sancoidosis, but no evidence of focal hepatic lesions at CT, only one was scanned without

use

for discriminating conditions. Nodal

a statistically

simu-

hepatic miseen in

Candida, Staphylococcus, and Aspergillus infections (18). This type of involvement, seen in three of our pa-

tients,

feature the two

with

sarcoidosis.

Acta

Radiol

1987;

28:723-725. Nakata K, Iwata K, Kojima K, Kanai K. Cornputed tomography of liver sarcoidosis. J Cornput Assist Tomogr 1989; 13:707-708. Mathieu

D. Vasile

N, Fagniez

bly D, Larde D. Dynamic patic abscesses. Radiology

P. Segui

S, Gra-

CT features of he1985; 154:749-752.

January

1991