199

Donald H. Hulnick, M.D.Alec J. Megibow, M.D.

David P. Naidich, M.D.

Susan Hilton, M.D.Kyunghee C. Cho, M.D.

Emil J. Balthazar, M.D.

Abdominal Tuberculosis: CT

The computed tomography (CT) scans of27 patients with abdominal tuberculosiswere reviewed retrospectively to deter-mine the range of abdominal involve-ment. Most patientshad been at increasedrisk because of intravenous drug abuse,alcoholism, acquired immunodeficiencysyndrome (AIDS), cirrhosis, or steroidtherapy. The etiologic agent was Myco-bacterium tuberculosis in 23 patients andM. avium-intiacellulare in four patientswith AIDS. In five patients, tuberculosiswas limited to the abdomen. CT findingsincluded adenopathy, splenomegaly, he-patomegaly, ascites, bowel involvement,pleural effusiori, intrasplenic masses, andintrahepatic masses. Characteristic fea-tures were a tendency for adenopathy toprominently involve peripancreatic andmesentenic compartments, low-densitycenters withiti enlarged node�, complexnature of the ascites, and adenopathy ad-jacent to sites of gastrointestinal tract in-volvement. Recognition of these manifes-tations and maintenance of an index ofsuspicion, especially in patients at risk,should help optimize the correct diagno-sis and management of intraabdominal

tuberculosis.

Index terms: Abdomen, diseases #{149}Abdomen,

computed tomography, 70.121 #{149}Acquiredimmunodeficiency syndrome (AIDS) #{149}Lymphatic

system, diseases, 99.8231 #{149}Mycobacteria, 70.203Tuberculosis, intestinal, 70.23

Radiology 1985; 157:199-204

1 From the Department of Radiology, New York

University Medical Center, New York (D.H.H., A.J.M.,D.P.N., S.H., E.J.B.), and the Department of Radiology,Albert E. Einstein College of Medicine, Bronx, New

York (K.C.C.). Presented at the 70th Scientific Assem-bly and Annual Meeting of the Radiological Society ofNorth AmeriCa, Washington, D.C., November 25-30,1984. Received January 10, 1985; revision requestedMarch 5; accepted May 13.

C RSNA, 1985

M ANIFESTATIONS of tuberculosis within the abdomen are pro-tean. The gastrointestinal tract, lymphatic system, peritone-

um, and solid viscera are subject to varying degrees of involvement:alone, in combination, or in association within extraabdominal dis-

ease. Hence, clinical presentations of abdominal tuberculosis areextremely varied, and diagnosis may be obscured by lack of radio-

logic evidence of pulmonary tuberculosis (1, 2). Abdominal in-volvement is not rare, however, and complicates pulmonary tuber-culosis in 6%-38% of patients (3). computed tomography (CT)offers the advantage of examining the range of abdominal involve-

ment in a single examination, but only two small series have beenrepofted previously (4, 5). CT findiz�gs in 27 patients are presented

here.

MATERIALS AND METHODS

The CT scans of 27 patients with proved intraabdominal tuberculosisexamined between July 181 and October 1984 at six New York City hospi-tals were reviewed retrospectively. No patients with disease limited to themUsculoskeletal system or genitourinary tract were included. Twenty-threepatients were male, and four were female. Their ages ranged from 21 to 68,with an average age of 37 years. Seventeen patients were white, eight were

black, and two were oriental. Most patients had one or more risk factors forthe development Of tuberculosis. Ten had a history of intravenous drugabuse, eight were alcoholics, and eight suffered immunosuppression as aresult of acquired immunodeficiency syndrome (AIDS) (six cases), steroid

therapy (one case), or cirrhosis (one case). Infection was limited to theabdomen in five patients but was present also in the lung or other extraab-dominal sites in the remainder. The causative agent was Mycobacteriumtuberculosis in 23 patients and M. avium-intracellulare in four patients withAIDS. The diagnoses had been proved by microbiological or pathologicanalysis of abdominal tissue (19 cases) or by improvement seen on abdomi-nal CT studies following antituberctilous chemotherapy prescribed because

of the documentation of tuberculosis in extraabdominal sites (seven cases).Scans were performed on commercially available equipment using 10-

mm collimation at 10-15-mm intervals. Oral and intravenous contrast me-dia were administered whenever possible. Clinical charts and other radio-logic studies were reviewed in each case.

RESULTS

Adenopathy was the most common finding on CT study and waspresent in 24patients. In 13 patients there was no associated gastro-

intestinal or penitoneal involvement, but generalized adenopathyor an abnormal chest radiograph evidenced the presence of ftiber-culosis elsewhere. In three patients, abdominal lymph nodes werethe only sites involved. Enlarged nodes were found in the periaor-tic/penicaval retropenitoneum, mesentery, omentum, peripancrea-

tic and porta hepatis regions, and groin (Fig. 1). Most patients hadinvolvement of more than one compartment. Adenopathy patterns

varied widely, including increased numbers of nodes of normal

size, scattered mildly enlarged nodes, localized clusters of severalenlarged nodes, and large conglomerate masses (Fig. ld). Peripan-

Figure 1

200 #{149}Radiology October 1985

Tuberculous lymphadenopathy.

a. Contrast material enhanced scan at the level of the portal vein (PV) shows low-density adenopathy with ring enhancement in theperipancreatic and netroperitoneal compartments (arrows).

b. Multiple enlarged discrete soft-tissue density mesentenic lymph nodes (N) are associated with dilated small-bowel loops with thickenedfolds (SB). Note relative lack of retroperitoneal involvement.

c. Contrast material enhanced scan at the kidney level demonstrates multilocular low-density retroperitoneal masses with ring enhance-ment (M) and soft-tissue density mesentenic adenopathy (*).

d. Heterogeneous soft-tissue density mass in left upper quadrant (M) proved to represent a mesentenic tuberculous abscess. Note lack of

retroperitoneal involvement.

creatic or porta hepatis adenopathywas demonstrable in 21 patients (Fig.la), mesentenic/omental adenopathyin 20 (Fig. ib), and netropenitoneal ad-

enopathy in 19 (Fig. lc). A striking

finding was the relative severity of

the adenopathy in the mesenteric,

omental, and penipancreatic locations

compared with the degree of retro-penitoneal involvement. The retro-

penitoneum was the dominant site of

adenopathy in only three patients,while mesentenic/omental adenop-athy was dominant in eight, and pen-

pancreatic adenopathy was dominant

in seven. The remaining six patientshad diffuse abdominal adenopathybut no particular site of dominant in-volvement.

The centers of enlarged lymph

nodes were of low density in ten pa-

tients (Fig. la and c). Administration

of intravenous contrast material ac-

centuated this finding by enhancing

the inflammatory rim surrounding

the caseous center. Discrete abscesses

containing low-density material were

drained surgically from the mesen-tery (two cases), porta hepatis (two

cases), and uterine adnexa (one case)

(Fig. 2a and b). Abscesses of soft-tis-sue density material were found in

the mesentery (one case) (Fig. id) and

penipancreatic region (two cases) (Fig.

4c and d). Despite the considerablebulk of adenopathy in several pa-

tients, lymph-node masses were not

responsible for obstruction of the un-

nary, biliary, or gastrointestinal tracts

in any patient.

Splenomegaly, defined as a span 13

cm or greater (6), was a prevalent

finding (18 cases). In only three pa-

tients were discrete intrasplenic

masses identified by CT (Fig. 2c). He-

patomegaly, defined as a span greater

than 20 cm (7), was a less common

finding (ten cases). No patient had

hepatomegaly without associated

splenomegaly. In only one patient

did a CT scan demonstrate mntrahepa-tic masses (Fig. 3). Liver biopsy resultswere positive for caseating granulo-matous disease in three patients

whose CT scans showed enlarged but

homogeneous livers and in one pa-

tient with a liver that appeared as nor-

IIK

Hepatic tuberculosis. Multiple low-density intrahepatic masses are

associated with adenopathy (N) in the porta hepatis. The spleen

was enlarged and showed similar masses.

. .‘.-

Volume 157 Number 1 Radiology #{149}201

�‘. A

C

Tuberculous peritonitis.a. Contrast material enhanced scan at level of kidneys

shows relatively high-density ascites (A) and markedaccentuation of mesentenic vessels with mesentenic ad-

enopathy (arrows). Retropenitoneum is relatively unin-

volved.b. In the same patient, scan through the pelvis (B blad-

der, C = colon) demonstrates bilateral tubo-ovanian ab-

scesses (A), showing partial low density on the right.

c. Scan at the renal level (K) without contrast material

shows enlarged spleen with low-density mass (5) with-

in it and voluminous ascites complicated by soft tissuestanding and small irregular densities.

mal in size and density on CT scans.

Ascites was present in seven pa-

tients, three of whom also had pleural

effusion. The fluid was voluminousand diffusely distributed throughout

the peritoneal cavity in five patients

and was loculated in small scatteredpockets in two. Tuberculous salpingi-

tis apparently preceded diffuse pen-

tonitis in two patients (Figs. 2a and b),

and one patient had preexisting cm-

rhosis and ascites (Fig. 2c). Fluid-den-sity measurements ranged from 20 to

45 Hounsfield units (HU) and aver-

aged 30 HU for the seven patients.Penitoneal enhancement following

administration of intravenous con-

trast material was seen in three pa-

tients. In all patients with ascites,

there was abnormal thickening of softtissue and nodulanity associated withthe penitoneal surfaces, mesentery,and omentum (Fig. 2c).

CT scans demonstrated ileo-cecaltuberculosis in one patient (Fig. 4a

and b), diffuse small-bowel involve-

ment in another patient (Fig. lb), and

a peripancreatic tuberculous abscessassociated with a duodenal fistula in

two patients (Fig. 4c and d). In eachcase, there was adenopathy adjacent

to the involved bowel segment. In

one patient with endoscopically

proved colonic tuberculosis, the CTscan demonstrated diffuse abdominal

adenopathy with a mesentenic pre-

dominance, but the sites of colonic in-

#{149}#{149}�: -�::--__ �:-:: T

Figure 4

202 #{149}Radiology October 1985

Gastrointestinal tuberculosis.

a. Scan through lower abdomen shows thickening of cecal wall (C) and an adjacent mass of conglomerate adenopathy

(arrows).

b. In the same patient, small-bowel series demonstrates ulcerohypentrophic ileo-cecal tuberculosis.

c. A large, irregular, soft-tissue penipancreatic mass (M) deviates the stomach (S) anteriorly. A focus of extraluminal air

(arrow) is seen medial to the widened duodenal sweep (D).

d. Hypotonic upper gastrointestinal study shows spiculation of the medial aspect of the second and third duodenal

segments (arrows) and outlines extravasation into the penipancreatic abscess (curved arrows).

volvement could not be identified

from the scan.

DISCUSSION

Tuberculosis within the abdomen

potentially can affect the gastrointes-

tinal tract, peritoneum, lymph nodes,

liver, on spleen. Possible mechanisms

in the development of abdominal dis-

ease include ingestion of infected ma-

tenial such as sputum or milk, or he-

matogenous dissemination to the

abdominal viscera and lymphatic sys-

tem from a distant focus, usually in

the lung. Prior to the advent of CT

scanning, radiologic evaluation was

confined largely to studies of the gas-

trointestinal tract with contrast me-

dia. Extramucosal disease in lymph

nodes, peritoneal involvement, on in-

volvement of the liver and spleen

could not be evaluated directly by

this method. Even before the era of

chemotherapy, however, it had been

shown that up to two-thirds of pa-

tients with abdominal tuberculosis

had lymphadenopathy or penitoneal

disease without gastrointestinal in-

volvement (8, 9). Contrast material

studies alone, therefore, have been of

limited utility in detecting and locat-

ing the full range of abdominal dis-

ease. Furthermore, the virtual disap-

pearance of bovine tuberculosis in the

United States and the general avail-

ability and effectiveness of antituber-

culous chemotherapy have resulted

in a decreased incidence of intestinal

tuberculosis caused by ingestion.

Concurrently, tuberculosis has be-

come less readily recognized, perhaps

because of its decreased incidence or a

changing pattern of clinical and ra-

diological presentation since the ad-

vent of chemotherapy (10-12). Today,

immunodeficiency associated with ci-

then alcoholism, intravenous drug

abuse, diabetes, cancer, steroid thera-

py, and acquired states (e.g., AIDS)are prominent risk factors (13), and

indeed, in our series only 20% of the

patients did not have at least one of

these risk factors.

Volume 157 Number 1 Radiology #{149}203

Abdominal lymphadenopathy was

the most common manifestation of tu-berculosis on CT scans in our series

and was present in nearly all (23 of26) patients. Of course, adenopathy

within the abdomen is found in amultitude of pathologic conditions,

benign and malignant; but in our pa-

tients, tuberculous adenopathy exhib-

ited several characteristic but not pa-

thognomonic features. There was a

striking tendency for the involve-

ment of mesentenic and penipancrea-

tic node groups to accompany and

usually overshadow the extent of ret-

ropenitoneal involvement (Fig. la, b,and d). While a previous study (4) did

describe omental involvement, thefinding of mesentenic, porta hepatic,and penipancneatic adenopathy wasnot emphasized. Lymphangiographicstudies in patients with abdominal tu-

berculosis have noted a gradient of

retroperitoneal involvement, most se-vere in the upper abdomen (14), and

mesentenic adenitis and penipancrea-

tic involvements are well-recognizedmanifestations of abdominal disease(9, 15, 16). Perhaps this distributionreflects the lymphatic drainage ofsites in the small bowel and liver that

had been seeded hematogenously butmay not have developed frank clini-cally apparent infection. Iliac, lum-bar, and lower abdominal nodegroups are less commonly involved

because they drain the lower extrem-ities, pelvic viscera, and distal co-lon-areas less likely to be affected byhematogenous tuberculous dissemi-

nation, except when infection is over-whelming.

In our series, lymph node masses,even when large, did not cause ob-

struction of the biliary ducts, uneters,

or bowel. It should be noted, howev-en, that obstructive jaundice second-any to adenopathy in the porta hepa-tis, hepatoduodenal ligament, and

penipancreatic area has been reportedas a rare complication of tuberculosisin North American patients (17, 18)

and is apparently less rare in the Fili-pino population (19).

Low-density centers within tuber-culous abdominal lymph nodes havebeen noted in previous reports (4, 5)and were identified in 40% (ten of 24)of our patients (Fig. la and c). Thisfinding is not diagnostic, however,and may be found occasionally in

metastatic malignancy, lymphoma(especially if treated), pyogenic infec-tion, and Whipple disease (20). In tu-berculosis, presumably the low-den-sity nature of the involved nodes issecondary to caseation necrosis. Inour series, however, there was nostrong correlation between the site orextent of adenopathy and the density

of the involved nodes. The largest

mesentenic abscess (10 cm) was of

soft-tissue density (Fig. id), while the

smallest (2 cm) was of low density.

Gastrointestinal tuberculosis can

result from ingestion, extension from

the lymphatic system, or hemato-

genous dissemination; each of these

pathophysiologic mechanisms is rep-

resented in our series. One patient

whose chest radiograph was within

normal limits had a history of recent

travel to an area endemic for bovine

tuberculosis. This patient had ulcero-

hypentrophic ileo-cecal disease, pne-

sumably caused by ingestion of in-

fected milk (Fig. 4a and b). In two

patients, massive penipancneatic tu-

berculous adenopathy eroded into

the duodenum, resulting in fistula

(Fig. 4c and d). In a patient with AIDS,

diffuse bowel involvement (Fig. ib)

was associated with hematogenous

dissemination to the lymph nodes,

bone marrow, central nervous system,

liver, and spleen. Though CT was of

diagnostic importance in demonstrat-

ing ancillary findings such as adenop-

athy, barium studies were essential

for further characterizing the nature

of the bowel pathology in each pa-

tient.

Tuberculous peritonitis, as seen in

seven of our patients, is a well-de-

scnibed manifestation of abdominal

disease (8, 9, 15, 21-23) but occurs in

less than 4% of patients with pulmo-

nary tuberculosis (1, 2). Peritonitis

and ascites may be associated with

widespread abdominal and extraab-

dominal disease, as in three of our pa-

tients. Tubenculous salpingitis pro-

gressing to diffuse peritonitis, a

known mechanism of spread (2), wasevident in two of our patients (Fig. 2a

and b). The association of cirrhosis

and tuberculous peritonitis has alsobeen described (20) and was present

in one patient (Fig. 2c). In the seventh

patient, tuberculous peritonitis ap-

peaned to develop primarily, as no

pulmonary or other abdominal site of

infection was evident.

In these seven patients, ascitic fluid

density ranged from 20 to 45 HU, (av-

enage, 30 HU). It has been suggestedthat high-density ascites may be char-

actenistic of tuberculosis (4, 5). Theo-

retically this can be explained by the

high protein and cellular contents in

a tuberculous effusion, particularlywhen cell-mediated immunity has de-veloped. We emphasize, however,

that tubenculous ascites may also be

near water in density, perhaps reflect-

ing an earlier transudative stage of

immune reaction. The problems in-

herent in using CT density measure-

ments as absolute standards of refer-

ence should be recognized (24), and

this is especially evident when “mea-

suning” small areas such as pockets of

ascites. Moreover, as in our cases, an-

cillany findings of tubo-ovanian ab-

scesses, adenopathy, penitoneal en-

hancement, or a “dirty” appearance to

the mesentery indicate a complex na-

tune to the ascites and should suggest

the diagnosis of tubenculous penitoni-

tis. Recognition of these features

proved more relevant than merely

measuring ascitic fluid density or at-

tempting to classify peritoneal in-

volvement into the classic wet, dry, or

fibrotic-fixed forms (1, 2, 21).

Tuberculosis of the liver or spleen

is reportedly uncommon, except in as-

sociation with miliary dissemination

(2, 9). Though hepatic involvement

by tuberculosis reportedly tends to be

diffuse (2), the macronodular or pseu-

dotumon forms, as found in one of our

patients (Fig. 3), is rarely seen as well

(25). The high incidence of spleno-

megaly (18 of 27 cases) and hepatome-

galy (ten of 27 cases) in our series is

higher than in previous series (8, 9,

14, 15), and four of our patients did

not appear to have miliary disease or

even disease outside the abdomen.Our findings may in part be related tothe high incidence of coexisting alco-

holism and intravenous drug abuse in

our patients. Interestingly, no patient

had hepatomegaly without spleno-

megaly, and in many cases the en-

largement was mild, perhaps indicat-

ing that enlargement of the liven or

spleen in some cases represented a

nonspecific response to infection.

Patients with AIDS presented spe-

cial problems. There have been sever-

al recent reports of disseminated in-

fection with M. avium-intracellulare

complex in these patients (26-29), aswas seen in four of our cases. Yet two

of our patients with AIDS were in-

fected with typical M. tuberculosis. Pri-or to the AIDS epidemic, disseminat-

ed infection with M. avium was

exceedingly rare (30-32). In our series

each patient with M. avium infection

had been chronically ill with AIDS-

related diseases and exhibited multi-

system involvement. However, dis-

tinguishing M. tuberculosis from M.

avium infection was not possible byCT criteria alone. In all AIDS patients,differential diagnosis must include

the Lymph Node Syndrome, Kaposi

sarcoma, lymphoma, and other op-

portunistic infections as well as typi-

cal and atypical tuberculosis (33); ag-gressive biopsy, smear, and culture

studies of relevant tissue are neces-

sary for an accurate diagnosis (26, 27,

34-37).In any patient, diagnosis of intraab-

dominal tuberculosis is difficult be-cause symptoms are vague, signs non-

204 #{149}Radiology October 1985

specific, and clinical tests often notconclusive. Though the presence or ahistory of pulmonary disease may besuggestive, nearly one-fifth of our pa-tients (five of 27) had no evidence of

extraabdominal disease. Recovery ofmycobactenia from the stool may notnecessarily indicate gastrointestinalinfection. Results of purified proteinderivative skin tests may be falselypositive or negative. Results of liverbiopsy testing are usually negative inthe absence of miliary disease (2). Theimportance of reaching the correct di-agnosis, however, is formidable, as

untreated abdominal tuberculosis car-nies a 50% mortality rate (8). Laparoto-my should be avoided if possible, andeven those patients requiring surgery

benefit from preoperative antituber-culous chemotherapy (9).

Unfortunately, there are no CTfindings, alone or in combination,that are pathognomonic of tuberculo-sis. Abdominal adenopathy, predomi-nantly in the mesenteric and peripan-creatic areas, is a suggestive finding.Low-density centers within the en-larged nodes are also a characteristicfeature. Ascites-perhaps relativelyhigh in density, loculated, or accom-panied by peritoneal enhancement oradenopathy, and associated with mes-entenic and omental soft-tissuechanges-is suggestive of tubercu-bus peritonitis. Bowel abnormalityassociated with adjacent adenopathymust also raise the possibility of gas-trointestinal tuberculosis. And mostimportant, to make the correct diag-nosis, an index of suspicion must bemaintained, especially in those pa-tients at increased risk.

Acknowledgments: We gratefully thank Drs.L. Berliner, J. Lang, R. Meisell, and K. Zirinskyfor contributing cases for this study, and JennieLopez for assistance in manuscript preparation.

Send correspondence and reprint requests to:Donald H. Hulnick, M.D. Department of Radi-ology, New York University Medical Center,

560 First Avenue, New York, NY 10016.

References

1. Carrera CF. Young 5, Lewicki AM. Intesti-nal tuberculosis. Gastrointest Radiol 1976;

1:145-155.

2. Thoeni RF, Margulis AR. Gastrointestinaltuberculosis. Semin Roentgenol 1979;14:283-294.

3. Goldberg HI, Reeder MM. Infections and

infestations of the gastrointestinal tract.In: Margulis AR, Burhenne HJ, eds. Ali-mentary tract roentgenology. St. Louis:Mosby, 1973; 1575-1607.

4. Epstein BM, Mann JH. CT of abdominaltuberculosis. AJR 1982; 139:861-866.

5. Dahlene DH, Stanley RJ, Koehler RE, ShinMS. Tishler JMA. Abdominal tuberculo-sis: CT findings. J Comput Assist Tomogr1984; 8:443-445.

6. Johnson PM, Spencer PP. The spleen. In:Freeman LM, Johnson PM, eds. Clinicalscintillation imaging. 2d ed. New York:Grune & Stratton, 1975; 639-670.

7. Johnson PM. The liver. In: Freeman LM,Johnson PM, eds. Clinical scintillation im-aging. 2d ed. New York: Grune & Stratton,1975; 405-459.

8. Fraki 0, Peltokallio P. Intestinal and peri-toneal tuberculosis: report of two cases. DisColon Rectum 1975; 18:685-693.

9. Bhansali 5K. Abdominal tuberculosis: cx-periences with 300 cases. Am J Gastroen-

terol 1977; 67:324-337.10. Friedenberg RM, Isaacs N, Elkin M. The

changing roentgenologic picture in pul-monary tuberculosis under modern che-

motherapy. AJR 1959; 81:196-206.11. MacGregor RR. A year’s experience with

tuberculosis in a private urban teachinghospital in the postsanitorium era. Am J

Med 1975; 58:221-228.12. Miller WT, MacGregor RR. Tuberculosis:

frequency of unusual radiographic find-ings. AJR 1978; 130:867-875.

13. Miller WT. Tuberculosis in the immuno-suppressed patient. Semin Roentgenol1979; 14:249-255.

14. Beetlestone CA, Wieland W, Lewis EA,Itayemi SO. The lymphogram in abdomi-nal tuberculosis. Clin Radiol 1977; 28:653-658.

15. Gunn A, Keddie NC. Abdominal tubercu-losis. Br J Sung 1972; 59:597-602.

16. Kolawole TM, Lewis EA. A radiologic

study of tuberculosis of the abdomen (gas-trointestinal tract). AJR 1975; 123:348-358.

17. Kohen MD, Altman KA. Jaundice due to arare cause: tuberculous lymphadenitis. AmJ Gastroenterol 1973; 59:48-53.

18. Murphy TF, Gray GF. Biliary obstructiondue to tuberculous adenitis. Am J Med1980; 68:452-454.

19. Alvarez SZ, Carpir R. Hepatobiliary tu-berculosis. Dig Dis Sci 1983; 28:193-200.

20. Li DKD, Rennie CS. Abdominal comput-ed tomography in Whipple’s disease.Comput Assist Tomogr 1980; 4:630-633.

21. Auerbach 0. Pleural, peritoneal, and peri-

cardial tuberculosis. Am Rev Tuberc 1950;61:845-861.

22. Burack WR, Hollister RM. Tuberculous

peritonitis. Am J Med 1960; 28:510-523.23. Sochocky S. Tuberculous peritonitis: a re-

view of 100 cases. Am Rev Respir Dis 1967;95:398-401.

24. Levi C, Gray JE, McCullough EC, HatteryRR. The unreliability of CT numbers asabsolute values. AJR 1982; 139:443-447.

25. Zipser RD. Ray JE, Ricketts RR, et al. Tu-berculous pseudotumors of the liver. Am J

Med 1976; 61:946-951.26. GreeneJB, Sidhu GS, Lewin 5, et al. Myco-

bacterium avium-intracellulare: a cause of dis-seminated life-threatening infection in ho-mosexuals and drug abusers. Ann Intern

Med 1982; 97:539-546.

27. Zakowski P. Fligiel 5, Berlin GW, JohnsonBL. Disseminated Mycobacterium avium-in-tracellulare infection in homosexual mendying of acquired immunodeficiency.JAMA 1982; 248:2980-2982.

28. Pitchenik AE, Fischl MA. Disseminatedtuberculosis and the acquired immunode-ficiency syndrome, letter to the editor.

Ann Intern Med 1983; 97:112.29. Poon MC, Landay A, Prasthofer EF, Stagno

S. AIDS with Pneumocystis pneumoniaand Mycobacterium avium-intracellulare in-fection in a previously healthy patientwith classic hemophilia. Ann Intern Med

1983; 98:287-290.30. Saito H, Tasaka H, Osasa S, et al. Dissemi-

nated Mycobacterium intracellulare infec-tion. Am Rev Respir Dis 1974; 109:572-576.

31. Savage PJ, Dellinger PP. Disseminated hi-

berculosis caused by Mycobacterium intro-

cellulare, letter to the editor. Chest 1982;82:800-801.

32. Bone M, Stableforth D. Miliary infection

due to Mycobacterium avium-intracellulare.Tubercle 1981; 62:211-213.

33. Moon KL, Federle MP, Abrams DI, Vol-

berding P, Lewis BJ. Kaposi sarcoma andlymphadenopathy syndrome: limitations

of abdominal CT in AIDS. Radiology 1984;150:479-483.

34. Cohen RJ, Samoszuk MK, Busch D, LagiosM. Occult infections with M. intracellularein bone-marrow biopsy specimens frompatients with AIDS, letter to the editor. NEngl J Med 1983; 308:1475-1476.

35. Croxson TS, Ebanks D, Mildvan D. Atypi-cal mycobacteria and Kaposi’s sarcoma inthe same biopsy specimens, letter to theeditor. N Engl J Med 1983; 308:1476.

36. Brynes RK, Chan WC, Spira TJ, Ewing EP,Chandler FW. Value of lymph node biop-sy in unexplained lymphadenopathy inhomosexual men. JAMA 1983; 250:1313-

1317.37. Giron JA, Mandel U, Wollschlager C, et al.

Mycobacterial culture in acquired im-muno-deficiency syndrome, letter to theeditor. Ann Intern Med 1983; 98:1028.