Eur. Radiol. 7, 1419±1429 (1997) Springer-Verlag 1997
European Radiology
Original article Intestinal disease in acquired immunodeficiency: evaluation by CT F. D. Knollmann1, Th. GruÈnewald2, A. Adler3, J. MaÈurer1, R. E. Hintze3, H. D. Pohle2, R. Felix1 1
Strahlenklinik und Poliklinik, Virchow-Klinikum, Medizinische FakultaÈt der Humboldt-UniversitaÈt, Augustenburger Platz 1, D-13 353 Berlin, Germany 2 Medizinische Klinik II mit Schwerpunkt Infektionskrankheiten, Virchow-Klinikum, Medizinische FakultaÈt der Humboldt-UniversitaÈt, Augustenburger Platz 1, D-13 353 Berlin, Germany 3 Klinik fuÈr Innere Medizin mit dem Schwerpunkt Gastroenterologie, Zentrale InterdisziplinaÈre Endoskopie, Virchow-Klinikum, Medizinische FakultaÈt der Humboldt-UniversitaÈt, Augustenburger Platz 1, D-13 353 Berlin, Germany Received 17 October 1996; Revision received 20 January 1997; Accepted 24 February 1997
Abstract. Intestinal symptoms affect most AIDS patients at some point in their disease. The purpose of this study was to evaluate the use of CT in this setting. A total of 339 abdominal CT exams were reviewed for signs of intestinal disease. Abdominal CT scans of 45 patients with intestinal symptoms were compared with colonoscopy and histologic data. The CT results were correlated with CD4 + T-lymphocyte counts and patient survival. More than 14 % of all abdominal CT exams displayed signs of enteric disease. Of the 45 patients studied with both CT and colonoscopy, 35 (78 %) had signs of intestinal disease by CT. Of these 35 patients, colonoscopic signs of an intestinal lesion were found in 29 and histologic proof of disease was established in 30 cases. Colonoscopy and histology detected 8 lesions missed by CT. There were 14 cases of unspecific colitis, 15 cases of cytomegalovirus (CMV) colitis, and 4 cases of enteric tuberculosis as per biopsy. Five patients presented with Kaposi's sarcoma and 1 with a non-Hodgkin's lymphoma. Neither colonoscopic nor CT signs of intestinal disease did reliably distinguish between histologic subgroups. Specifically, CMV colitis could not be distinguished from unspecific colitis. CD4 + T-lymphocyte counts for histologic subgroups were not significantly different, either. No colonoscopic or histologic feature predicted survival, whereas low CD4 counts and ascites on CT indicated a poor prognosis. Whereas CT detects signs of intestinal disease in most AIDS patients, these signs remain largely unspecific. Colonoscopy and biopsies provide no consistently valid standard with which to compare CT because of controversial sensitivity and specificity of these methods. The CT technique detects small bowel as well as extraintestinal disease. Therefore, CT is an important diagnostic modality in abdominal disease of immunocompromised patients.
Correspondence to: F. D. Knollmann
Key words: Human immunodeficiency virus ± Bowel disease ± CT ± AIDS ± Cytomegalovirus ± Colitis
Introduction Presently, we have to face an estimated 20 million human immunodeficiency virus (HIV)-infected people worldwide [1] and over half a million reported cases of the acquired immunodeficiency syndrome (AIDS) in the United States alone since 1981 [2]. The HIV-related disease poses a very serious threat to human health. Signs of abdominal disease have been found in most adult HIV-infected patients [3]. Persistent diarrhea is the most commonly encountered presenting problem. Although potential pathogens have been identified in the majority of cases, the causative agent often cannot be unequivocally determined in the individual patient [4]. The usual diagnostic approach is microbiological analysis of stool which detects potentially causative organisms in up to 85 % of patients [3]. If no pathogen is found, intestinal endoscopy is the next step. The most commonly identified pathogens are cytomegalovirus, cryptosporidium, and Mycobacterium avium-intracellulare [3]. Although several more descriptive reports on radiological signs of intestinal disease are available [5±10], a more rigorous analysis of the overall benefit abdominal imaging offers in the evaluation of HIV-infected patients with enteric symptoms [11, 12] is incomplete at best. The purpose of this study was to investigate the role of abdominal CT in HIV-related inflammatory bowel disease. Subjects and methods From January 1986 until July 1994, 339 HIV-infected patients (52 females; mean age 38.6 years, range 20± 68 years) were referred for abdominal CT either as a
1420
1a
F. D. Knollmann et al.: Intestinal disease in acquired immunodeficiency syndrome
1b Fig. 1 a, b. Abdominal CT scan of a 62-year-old male acquired immunodeficiency syndrome (AIDS) patient presenting with watery diarrhea. a The CT technique documents ascites, air±fluid levels and marked bowel wall thickening. The ascending colon presents the target sign (arrow), i. e., bowel wall layering of alternating density. Histologic workup revealed necrotizing colitis ascribed to cryptosporidium. Colonoscopy detected disseminated soft polyps and inflammation. b A rectal polyp-like tumor is also noted (arrow) on a more caudal CT scan
2 staging examination in patients with extra-abdominal disease or for the workup of abdominal symptoms. Inclusion criteria were referral from either a day-care clinic for HIV-infected patients or from the associated infectious disease department on an inpatient basis during the period between opening of the day-care clinic in June 1990 and July 1994. The inclusion of previous exams resulted from selection of the earliest available CTexam for each patient; thus, 339 CT studies were included. From January 1991 until December 1995, 90 patients (7 females; mean age 42.7 years, range 26±64 years) were studied by 111 colonoscopies. In all cases intestinal biopsies were sampled from each section of the large intestine and submitted for histologic workup. Also, cultures of biopsy material were started. At colonoscopy, lesions were classified as inflammatory or neoplastic, with aphthous ulcerations, erythema, mucous adhesions, erosions, ulcerations, pseudopolyps, fistula, and hemorrhagic colitis implying inflammation and angiomas, exulceration, exophytic tumors, and submucosal protuberation indicating neopastic disease. The type and site of bowel involvement were noted. Of these 90 patients, 45 (5 females; mean age 39.3 years, range 24±64 years) were also studied by abdominal CT. Imaging protocol Continuous 8- or 10-mm-thick slices were chosen for pre- and postcontrast scans (Somatom DRH or Somatom Plus, Siemens, Erlangen, Germany) applying
Fig. 2. This 46-year-old male hemophiliac presented with anal bleeding. On colonoscopy, anal angioma was detected. A CT scan revealed widespread perirectal and sigmoidal inflammation which was classified as unspecific colitis at histologic workup. There is proliferation of perirectal fat (large arrows) and perirectal stranding (small arrows)
200 ml intravenous (Ultravist 370, Schering, Berlin, Germany) and 2000 ml oral barium contrast. CT image analysis Scans were reviewed retrospectively by two experienced investigators blinded to clinical information other than serologic confirmation of HIV infection. All CT studies were evaluated for the presence of enteritis. All studies of patients studied by both CT and colonoscopy were checked for peritoneal stranding, bowel wall edema, contrast enhancement of the bowel wall, target phenomena, mesenteric lymphadenopathy, proliferation of mesenteric fat, ascites, intraintestinal air±fluid levels and retroperitoneal lymph node enlargement as indicators of inflammatory or neoplastic intestinal disease. Peritoneal stranding was noted if streaks of soft tissue density were scattered in the peritoneal fat. Bowel wall edema was assumed if the bowel wall displayed inhomogeneous density. The target sign denoted layering of the bowel wall due to either submucosal edema or fat proliferation (Fig. 1). Proliferation of mesenteric fat was characterized by prominent peritoneal tissue of fat density (Fig. 2). It was often accompanied by peritoneal stranding. The presence of specific features was established by consensus by jointly discussing discrepencies in rating a CT study. Bowel wall thickening was defined as an intestinal wall diameter of over 3 mm in distended or 5 mm in collapsed loops. Lymphadenopathy denoted the presence of not normally visible mesenteric, paracav-
F. D. Knollmann et al.: Intestinal disease in acquired immunodeficiency syndrome
1421
Table 1. Features of cytomegalovirus (CMV) colitis (NHL non-Hodgkin's lymphoma; KS Kaposil's sarcoma; n. a. not available) Patient Gender Age (years)
Clinical presentation
Clinical and colonoscopic features 2 M 51 Suspected toxic megacolon
Colonoscopic features Within normals limits
4
M
38
Inflammatory lesions on CT
Ulcerating enteritis from terminal ileum to transverse colon
14 15
M M
29 57
Diarrhea, painful defecation Diarrhea
Moderately inflamed bowel at all sites, rectal angioma at 10 cm Within normal limits
19
M
35
Diarrhea
Proctosigmoiditis
25
M
37
Bloody diarrhea
Ulcerating colitis with most severe involvement at descending colon, sigmoid, and rectum. Induration of sigmoid bowel wall, suspected NHL
27 30
M M
26 31
Diarrhea Diarrhea
Pseudopolyps in all segments including terminal ileum Within normal limits
32
M
44
Frequent bowel movements, painful defecation
Submucosal petechiae of sigmoid and rectum, perianitis
35
M
54
Bloody diarrhea
Severe segmental ulcerating colitis of terminal ileum, cecum, transverse and descending colon, and sigmoid of 10 cm length each. Perianal fistula
36
M
33
Diarrhea
Pale intestinal mucosa, otherwise within normal limits
41 42
M M
39 41
Bloody diarrhea Diarrhea
Hemorrhagic colitis of descending colon, sigmoid, and rectum Bowel wall edema and moderate inflammation in all segments
44
M
30
Rectal bleeding
Rectal ulceration
45
F
34
Diarrhea, subileus
Ileitis terminalis
Patient
CD4 count CT features
Patient outcome
CT features and patient outcome 2 26 Intestinal air-fluid levels, normal bowel wall
Still alive on day 167
4 14
n. a. 3
Died on day 234 Still alive on day 223, colonic KS
15
9
Perienteric stranding, mesenterial fat proliferation, intestinal air-fluid levels Perienteric stranding, mesenterial lymphadenopathy and fat proliferation, retroperitoneal lymphadenopathy Small bowel wall thickening/edema, enteric air-fluid levels
19
44
Intestinal wall thickening of the small bowel
Alive on day 394
25 27
6 83
Retroperitoneal lymphadenopathy Bowel wall thickening of sigmoid and rectum
Alive on day 75 Alive on day 1160
30
18
Small bowel wall thickening
Alive on day 22
32 35
12 n. a.
Alive on day 54 Lost to follow-up
36
15
Small bowel wall thickening Colonic wall thickening, perienteric stranding, mesenterial lymphadenopathy, and fat proliferation Perienteric stranding, mesenterial lymphadenopathy
41
n. a.
Within normal limits
Died on day 396
42 44
96 1
Within normal limits Small and large bowel wall thickening, retroperitoneal lymphadenopathy
Alive on day 297 Died on day 137
45
0
Small bowel and colonic wall thickening/edema and contrast enhancement
Proof of concomitant microsporidial enteritis, alive on day 321
al, or para-aortic nodes less than 10±12 mm in diameter depending on location. Larger nodes were included as grossly enlarged. For this analysis both readers were blinded as to all data other than the presence of HIV infection and unspecified abdominal complaints. Clinical data For correlation with clinical status, the 1993 classification scheme for HIV-infected patients from the Centers for Disease Control and Prevention (CDC) was employed [13]. This scheme designates category A to as-
Died on day 354
Died on day 332
ymptomatic individuals or those exhibiting generalized lymphadenopathy. Category B denotes exclusion of criteria for both categories A and C. Any of 26 defining conditions represents category C. Diagnosing AIDS relies on the demonstration of either clinical category C or a CD4 + T-lymphocyte count of less than 200/mL. CD4 + T-lymphocyte counts from within 2 weeks of the CT study were recovered from the patient medical charts. Clinical category was chosen to reflect the prestudy (CT) assessment. Survival data were assembled from medical records giving either the time of death or the last visit to the clinics. The records of the local bureau of vital statistics
1422
F. D. Knollmann et al.: Intestinal disease in acquired immunodeficiency syndrome
were searched for death claims concerning all patients lost from follow-up during the study period. Statistical analysis Initially, associations of colonoscopic, histologic, and imaging criteria were established using contingency tables and the chi-square test. CD4 + T-lymphocyte counts were compared for CT, histologic, and colonoscopic subgroups by the Mann-Whitney U-test. To determine the value of CT features for predicting the etiology of intestinal disease, discriminant analysis was performed. Kaplan-Meier survival analysis was employed for all colonoscopic, histologic, and imaging parameters. Median survival and its 95 % confidence interval were determined for both the presence and exclusion of any single clinical or imaging parameter. Finally, a proportional hazards regression model [14] was used to elucidate the determinants of survival among clinical and imaging data. The Logrank or the Breslow-Gehan-Wilcoxon test were applied depending on the respective hazard function. The null hypothesis was rejected at p < 0.05. Results
whereas 29 displayed signs of colitis and 6 of colonic neoplasia. Most commonly, ulcerating colitis (n = 13) or hemorrhagic colitis (n = 12) were found. In 4 cases inflammatory pseudopolyps were detected. Less common were aphthous mucosal lesions (n = 3), toxic megacolon, anitis interna, hemorrhagic proctitis, and fistula (one case each). Neoplastic findings comprised polyps, angioma, Kaposi's sarcoma (KS) and a cecal valve mass. There were two granulomas due to mycobacteriosis. Histologic and microbiologic diagnoses Histologic workup indicated normal biopsy specimen in 7 cases, whereas cytopathologic effects taken as proof of cytomegalovirus (CMV) colitis were noted in 15 patients. Nonspecific inflammatory alterations were found in 14 patients as the second most commonly encountered histologic result. There were 5 patients with biopsy proven KS. Seven patients revealed other infectious agents, with 4 cases of mycobacteriosis including 3 of Mycobacterium avium-intracellulare and 1 of mycobacterium africanum. Other agents were microsporidia, enterotoxic E. coli, and cryptosporidia in one case each.
Incidence of intestinal disease as by CT
Clinical features
Of 339 HIV-infected patients referred to abdominal CT, 46 displayed signs of inflammatory or neoplastic bowel disease using the criterion of bowel wall thickening. Within this cohort, bowel wall thickening did not predict survival. Bowel wall thickening revealed no correlation with clinical stage, nor with any other pathologic imaging finding. Bowel wall thickening was associated with low CD4 + T-lymphocyte counts (p = 0.006).
In the cohort of 45 patients with both CT and colonoscopy, all but 4 patients in CDC clinical stage B were classified as AIDS patients. Of these remaining 4 patients, 2 had CD4 T-lymphocyte counts of less than 200/mL and were therefore also defined as having AIDS. One patient underwent colectomy for severe hemorrhagic colitis.
CT features of enteric disease No demographic, histologic, or colonoscopic differences could be established between patients having been studied by both CT and colonoscopy as compared with those studied with colonoscopy alone. Thus, 45 patients were included for correlation of colonoscopic with CT findings. Of these 45 CT studies, 10 were entirely normal. In 33 studies bowel wall thickening was found. Mesenteric stranding was present in 12 cases, mesenteric fat proliferation in 8 cases. Bowel wall edema was found in 10 studies, mucosal contrast enhancement in 9. There were 10 instances of mesenteric and 8 of retroperitoneal lymph node enlargement. Ascites were found in 6 patients. Six patients displayed intraintestinal air±fluid levels. Colonoscopic findings In the cohort of 45 patients with both CT and colonoscopy, 8 patients had an entirely normal colonoscopy,
Correlation of CT with colonoscopy Although both CT and colonoscopy found abnormal findings in 29 cases, only 2 were unequivocally diagnosed as normal. In 6, only CT was considered pathologic, and in 8 patients, only colonoscopy detected abnormalities. Thus, there was no statistically significant association of an overall abnormal CT with abnormal colonoscopy using chi-square testing. Of the 24 cases of bowel wall thickening, only 2 were colonoscopically normal (p = 0.12); 18 displayed signs of inflammatory bowel disease (p = 0.13). Colonoscopic evidence of an intestinal tumor (n = 6) concurred with ascites (2 of 6; p = 0.18) and mesenteric lymphadenopathy (3 of 6; p = 0.1). Correlation of CT with histology In 30 patients, both CT and histology indicated enteric disease, whereas 5 abnormal CT scans did not match with pathologic biopsy results and 8 cases of biopsy proven pathology were missed by CT (Tables 1±6). Sen-
F. D. Knollmann et al.: Intestinal disease in acquired immunodeficiency syndrome
3a
3b
4a
4b
1423
Fig. 3 a, b. This 54-year-old male AIDS patient presented with hemorrhagic diarrhea. Colonoscopy found a severe ulcerating colitis in segmental distribution which affected all parts of the colon. a, b A CT scan depicts bowel wall thickening, mesenterial fat proliferation (single arrows) and pericolonic stranding (double arrows) of the sigmoid. Histologic diagnosis was cytomegalovirus (CMV) colitis
5 sitivity of CT for biopsy proven colonic disease was 79 %. Two patients showed no evidence of intestinal disease by either method. Of 15 cases of biopsy proven CMV colitis, 13 had signs of enteric disease on CT (Fig. 3, Table 1). Of 14 biopsy proven cases of unspecific colitis, 9 were considered abnormal by CT (p = 0.14; Fig. 4, Table 2). Of 5 cases of biopsy proven intestinal KS, 4 displayed conspicuous CT findings (Table 4). Three of the five instances of mesenteric lymph node enlargement had histologic proof of Kaposi's sarcoma (p = 0.06). Of 8 patients with retroperitoneal lymphadenopathy, none had a normal biopsy reading. All had histologic proof of enteritis (p = 0.17), and half of these patients had CMV colitis. None of the patients with CMV colitis had ascites (p = 0.16). Of 6 patients with ascites, 3 had histologic proof of enteritis (p = 0.1), 2 had a
Fig. 4 a, b. A 32-year-old male presenting after partial colectomy for toxic megacolon due to human immunodeficiency virus (HIV)-associated nonspecific ulcerating colitis. a A CT scan reveals wall thickening (arrow) of large bowel loops. b There are also intramural air inclusions in the remaining descending colon (arrow). There is bowel wall thickening in this location as well. Colonoscopy revealed severe hemorrhagic colitis spreading into the terminal ileum. Histologic diagnosis was unspecific colitis Fig. 5. A CT scan of a 34-year-old AIDS patient with biopsy proof of enteric Mycobacterium avium-intracellulare infection. Colonoscopy revealed confluent ulcerating inflammation of the cecum with some smaller ulcerations in the ascending, transverse, and descending colon. The cecum displays significant bowel wall thickening (long arrow) and cecal ulceration (short arrows)
normal biopsy diagnosis (p = 0.16). Intraintestinal air± fluid levels were found in 2 of the 4 patients with mycobacterial enteritis (p = 0.08), but not in a single case of unspecific colitis (p = 0.16). Of these 4 patients, 3 displayed bowel wall edema and mucosal contrast enhancement (p = 0.02), 2 ascites (p = 0.08) on CT. All cases of mycobacteriosis had a thickened bowel wall (p = 0.1) on CT (Figs. 5, 6).
1424
F. D. Knollmann et al.: Intestinal disease in acquired immunodeficiency syndrome
Table 2. Features of unspecific human immunodeficiency virus (HIV)-related colitis colitis Patient Gender Age (years) Clinical presentation
Colonoscopic features
Clinical and colonoscopic features 1 M 30 Diarrhea, history of salmonella bacteremia 3
M
46
Abdominal pain
Within normal limits Mild inflammation of terminal ileum, cecal tumor, erosive colitis
9
M
36
Diarrhea
Severe hemorrhagic ileitis terminalis and colitis
10 11
F M
29 36
Lower abdominal pain Cytostatic therapy for Hodgkin's disease
Mild inflammation at all levels Erosion of transverse colon
12
M
41
AIDS
Within normal limits
16 23
F M
24 46
Weight loss, diarrhea Watery diarrhea
24
M
47
Diarrhea, hepatitis B
Within normal limits Cecal ulceration, inflammation of portions of ascending and transverse colon Anitis interna, colonic inflammation
26
M
47
Wasting, bloody diarrhea
Hemorrhagic colitis at all levels
29 34
M M
33 49
Diarrhea Anorectal pain
Hemorrhagic proctitis, rectal inflammation, terminal ileitis Erosive proctitis
37
M
30
Bloody diarrhea
Hemorrhagic colitis
38
M
32
Bloody diarrhea
Severe hemorrhagic colitis
Patient
CD4 count
CT features
Patient outcome
CT features and patient outcome 1 3 Within normal limits 3
18
9 10
8 1057
11
46
12 16
1 3
Alive on day 258
Bowel wall thickening
Concomitant enteric KS, alive on day 3
Small bowel wall thickening, perienteric stranding, mesenterial fat proliferation Within normal limits
Died on day 199 Alive on day 117
Within normal limits
Died on day 792
Proliferation of mesenterial fat Small bowel wall thickening/edema and contrast enhancement
Alive on day 277 Alive on day 801
23
86
Perienteric stranding, mesenterial lymphadenopathy and fat proliferation
Died on day 155
24
n. a.
Small bowel and cecal wall thickening, mesenterial lymphadenopathy and mucosal contrast enhancement
Lost to follow-up
26
104
Within normal limits
Wasting syndrome, died on day 468
29 34
14 176
Cecal bowel wall thickening, mesenterial lymphadenopathy Within normal limits
Died on day 272 Alive on day 359
37
149
Cecal bowel wall thickening
Alive on day 303
38
648
Small bowel and colonic wall thickening, periintestinal stranding, retroperitoneal lymphadenopathy, mucosal contrast enhancement
Alive on day 312
Correlation of colonoscopy with histology Thirty-two patients displayed abnormal findings in both colonoscopy and histology, 6 abnormal colonoscopic reports found no biopsy correlate, and 4 biopsies proving abnormal were missed by colonoscopy. In 4 patients neither colonoscopy nor laboratory workup revealed any abnormality. There was a trend for abnormal colonoscopic results to parallel pathologic biopsy findings (p = 0.13). Upon subgroup analysis of colonoscopic results, signs of colitis were associated with histologic proof of colitis (p = 0.02) as 26 of 29 patients with colonoscopic signs of colitis were confirmed by biopsy. There was also a trend for CMV colitis to parallel colonoscopic signs of colitis (p = 0.19), whereas there was
no such trend for nonspecific colitis. Of the 15 CMV colitis cases, 12 were positive for colonoscopic signs of colitis. Kaposi's sarcoma was found in 5 of 6 tumors detected by colonoscopy (p < 0.0001). One case of enteric non-Hodgkin's lymphoma was found. Common combinations of CT features Some typical combinations of abnormal CT findings were noted. Wall thickening was associated with intestinal wall edema (p = 0.002) and mucosal contrast enhancement (p = 0.002), and tended to occur more often with proliferation of mesenteric fat (p = 0.12) and ascites (p = 0.19).
F. D. Knollmann et al.: Intestinal disease in acquired immunodeficiency syndrome
1425
Relation to CD4 + T-lymphocyte counts No statistically significant association of CD4 + T-lymphocyte counts with any colonoscopic, histologic, or CT parameter could be established in the group studied with both colonoscopy and CT, whereas CT signs of enteric disease were more commonly found at a low CD4 count (p = 0.14; Fig. 7). Of histologic diagnoses, unspecific colitis was more commonly found at higher CD4 counts than CMV colitis (p = 0.19). CD4 counts of patients with unspecific colitis were higher than those of patients without unspecific colitis in this cohort of 45 patients with both CT and colonoscopic studies (p = 0.06). Prognostic impact Fig. 6. Correlation of CT features with histology and cultures of colonoscopic biopsies. Whereas some differences in the relative incidence of CT features for various etiologies can be found, no truly specific criteria apply, especially in light of the variable incidence of pathogens (n, first column). CT pathol. signs of intestinal disease by CT; KS Kaposi's sarcoma; wallthick. wall thickening; TB tuberculosis; mes. nodes mesenterial lymphadenopathy; mes. fat proliferation of mesenterial fat; retr. nodes retroperitoneal lymphadenopathy; mucos. sig. mucosal contrast enhancement; AFL air±fluid levels
Stranding of mesenteric fat was observed in association with mesenteric lymphadenopathy (p = 0.001), proliferation of mesenteric fat (p = 0.0001), and retroperitoneal lymphadenopathy (p = 0.02), although this combination did not find any specific histological correlation. Bowel wall edema was also encountered in conjunction with mesenteric fat proliferation (p = 0.18), ascites (p = 0.08), mucosal contrast enhancement (p = 0.0007), and air±fluid levels (p = 0.01). This combination was more common in patients with abnormal colonoscopic results (p = 0.18) and in the absence of CMV colitis (p = 0.16). Discriminant analysis of CT features to predict histologic diagnosis None of the canonical discriminant functions reached the level of statistical significance. There was a trend for the discriminant function of CMV colitis to correctly classify cases (p = 0.16). Of the cases, 78 % were correctly classified regarding the presence of CMV colitis with the strongest predictors being the absence of ascites (p = 0.07) or mucosal contrast enhancement (p = 0.12). Sensitivity of discriminant functions varied between 50 % for mycobacteriosis and 93 % for unspecific colitis. Specificity varied between 55 % for unspecific colitis and 85 % for KS and mycobacteriosis. Including age, CD4 + T-lymphocyte count and colonoscopic diagnosis in the analysis resulted in a significant discriminant function for KS only (p < 0.0001) due to the high sensitivity and specificity of the colonoscopic appearance (p < 0.0001), whereas mesenteric lymph node swelling was the main CT indicator of KS (p = 0.004).
In terms of patient survival, a low CD4 count (p = 0.15), histologic evidence of enteritis (p = 0.16), and ascites (p = 0.09) predicted a poor clinical course in KaplanMeier analysis of our 45 patients, whereas none of the colonoscopic parameters allowed any predictions. Upon Cox regression analysis, only the presence of ascites (p = 0.09) and low CD4 counts (p = 0.15) displayed a trend for poor survival. In the cohort of 339 cases, bowel wall thickening was no independent predictor of survival. Prognosis was predicted by a low CD4 count and abnormal abdominal CT findings due to ascites and tumors. Discussion Abdominal CT detects intestinal disease in a significant percentage of HIV-infected patients. Symptomatic patients subjected to both colonoscopy and abdominal CT almost uniformly are in an advanced stage of immunodeficiency. Although CT detects an appreciable portion of biopsy proven cases of colitis, it does not offer any specific clue as to the etiology of enteric disease. Similarly, barium enemas detect abnormal findings in the majority of AIDS patients, although findings are nonspecific [11]. The most commonly recommended approach in the evaluation of abdominal disease in AIDS patients is preparation of stool cultures. Although this simple and inexpensive test detects a number of potential pathogens, the issue of whether or not the pathogens found actually cause the patient's distress often remains to be solved by a trial-and-error therapeutic approach. In persistent or more severe disease, colonoscopy and CT are potential choices in the further workup. Whereas inflammatory bowel disease can be confirmed by both CT and colonoscopy in most cases, colonoscopy-directed biopsies are currently the only method to confirm CMV colitis. Cytomegalovirus infection is regarded as the most common serious opportunistic pathogen in AIDS by some groups [15] and a major cause for inflammatory intestinal disease in HIV-infected patients. Detection of the virus in biopsy specimen is generally not regarded
1426
F. D. Knollmann et al.: Intestinal disease in acquired immunodeficiency syndrome
Table 3. Features of patients without biopsy or microsurgical proof of intestinal disease Patient
Gender
Age (years) Clinical presentation
Clinical and colonoscopic features 5 M 27 13 M 32 17 M 55 18 M 36 20 M 31 22 M 45 39 M 41 Patient
CD4 count
CT features and patient outcome 5 13 13 14 17 41 18 2 20 17 22 n. a. 39 29
Colonoscopic features
Diarrhea Painful defecation Suspected CMV colitis Cramp-like upper abdominal pain Hemorrhagic diarrhea Liver cirrhosis, elevated tumor markers CMV retinitis
Within normal limits Proctoanitis Within normal limits Within normal limits Colitis at all levels, most severe in sigmoid and rectum Within normal limits Within normal limits
CT features
Patient outcome
Small bowel and colonic wall thickening and edema Within normal limits Small bowel and cecal wall thickening/edema, ascites, enteric air-fluid levels Within normal limits Periintestinal stranding, mesenteric fat proliferation Periintestinal stranding, ascites Small bowel wall thickening, mucosal contrast enhancement
Died on day 192 Alive on day 235 Died on day 463 Died on day 334 Alive on day 781 Lost to follow-up Died on day 49
Table 4. Features of patients with intestinal KS Patient
Gender
Age (years)
Clinical and colonoscopic features 3 M 46
Clinical presentation
Colonoscopic features
Abdominal pain
Mild inflammation of terminal ileum, cecal tumor, erosive colitis
7
F
57
Diarrhea
Three tumors in descending and transverse colon and sigmoid
14
M
29
Diarrhea, painful defecation
Moderately inflamed bowel at all sites, rectal angioma at 10 cm ab ano
21
M
40
Systemic mycobacteriosis (MAI), KS of skin, pallate, tongue and esophagus
3-cm rectal angioma, mild colitis at all levels
40
M
46
Hemophilia B, liver cirrhosis, hepatitis A, history of colitis ulcerosa, perianal bleeding
0.5- × 2-cm anal angioma
Patient
CD4 count
CT features
Patient outcome
CT features and patient outcome 3 18 Bowel wall thickening 7
n. a.
14
3
21 40
3 168
Concomitant enteric KS, alive on day 3
Within normal limits
Lost to follow-up
Perienteric stranding, mesenterial lymphadenopathy and fat proliferation, retroperitoneal lymphadenopathy
Still alive on day 223, colonic KS
Mesenterial lymphadenopathy Colonic wall thickening, periintestinal stranding, mesenterial lymphadenopathy, ascites, mucosal contrast enhancement
Alive on day 255 Died on day 114
as proof of infection. Instead, cytopathologic effects, such as intracellular inclusion bodies or detection of CMV genomic material by in situ hybridization [16], need to be demonstrated to allow for a diagnosis of CMV infection [3]. A recent nonblinded study comprising biopsy proven cases of CMV colitis without any control group recommended to assume CMV colitis in AIDS patients with CT signs of enteritis at a CD4 + T-lymphocyte count of less than 200/mL [10]. Our results do not support this conclusion. Instead, CT signs of inflammatory bowel disease predict CMV colitis only in half the cases in our experience. Commonly, an unspecific inflammatory process is found. Other than hematochezia, no clinical feature differentiating CMV colitis from other causes
of diarrhea in AIDS could be established in a recent autopsy series [17]. Although CD4 + T-lymphocyte counts are somewhat lower in CMV colitis than in unspecific colitis, the threshold for assuming CMV colitis, in our experience, is lower than previously published [10]. The risk of CMV infection has been found to increase more than twofold at a CD4 count of less than 100/ml in a 2-year period before [16]. The lower CD4 + T-lymphocyte count in patients with normal laboratory workup was not statistically significant. The difference is related to the low number of cases (difference by chance) and the fact that some sick patients may have evaded laboratory proof of disease, which is suggested by the CT features (5 of 7 cases abnormal by CT). Of note, the CD4 count of patients with-
F. D. Knollmann et al.: Intestinal disease in acquired immunodeficiency syndrome
Fig. 7. Mean CD4 + T-lymphocyte counts for patients with normal CT findings and CT signs of inflammatory or neoplastic bowel disease and for histologic findings of normal bowel wall, unspecific colitis, and CMV colitis. In unspecific colitis, CD4 + T-lymphocyte counts are higher than in either normal or CMV-positive biopsy specimen. Error bars indicate standard errors
out laboratory proof of disease was not different from that of patients with CMV colitis. In the workup of enteric disease in AIDS patients, colonoscopy offers a higher sensitivity for both inflammatory and neoplastic bowel lesions as by biopsy proof. Still, this result is somewhat redundant in that the sites of biopsies were determined by the colonoscopic findings. Conversely, CT findings also include small bowel lesions which are not amenable to colonoscopic biopsy. The sensitivity of colonoscopically directed biopsies is further limited by the number [18] and depth of biopsies taken. Thus, histologic workup of biopsies sampled by colonoscopy may not be a generally valid external standard to evaluate test qualities of CT in AIDS-associated colitis. The high incidence of inflammatory bowel lesions with unspecific histologic correlation is an ominous finding. Anorectal inflammatory lesions of unknown etiology [19] is a well-known finding indicating some as yet unknown pathogenic background. A separate HIV enteropathy of the small bowel has been suggested as a cause for diarrhea in patients with negative stool cultures [20]. Evidence that both retinitis and colitis can be caused by HIV itself has been accumulated by others [21]. The cases of unspecific colitis in this study may thus be ascribable to bowel wall infection by HIV itself, some as yet unidentified pathogen or a notorious pathogen missed by biopsy. The relatively high CD4 + T-lymphocyte counts in these patients indicate, however, that some other process may be causative than in biopsy proven CMV colitis. The incidence of other pathogens found in our study is probably too low to allow for generalizations regarding the significance of single imaging features. Their incidence parallels that of other reports. Mycobacterium avium-intracellulare infection of the gastrointestinal (GI) tract has a prevalence of 15 % in HIV-infected patients and is considered to be a late-stage event [22]. The GI tract is supposedly the port of entry of this pathogen. The CT features of intestinal mycobacterial infection have been described before with a predilec-
1427
tion for asymmetric bowel wall thickening at the ileocecal valve [23]. In all 4 cases of intestinal mycobacteriosis included in this study these features were detected. Differentiation of infection by Mycobacterium tuberculosis from Mycobacterium avium-intracellulare may be accomplished by the presence of centrally hypodense lymph nodes due to lymph node necrosis in Mycobacterium tuberculosis infection, although this remains a tentative diagnosis with one third false-positive results [12]. Microsporidia, one of the less common causes for diarrhea in AIDS, has been found to occur in its Enterozyton bieneusi strain in asymptomatic patients as well, which calls its role as a pathogen into question [24]. Cryptosporidia have been found in over one third of AIDS patients with diarrhea or malabsorption [25]. Diagnosis is usually established by stool cultures and no distinctive imaging features apply. Of the neoplastic etiologies, KS and malignant lymphoma are the most common considerations. Evidence that hyperattenuating lymphadenopathy denotes infiltration by KS [26] is circumstantial at best with direct pathologic proof largely missing. Colonoscopy remains the method of choice in the detection of KS, although these lesions may escape biopsy when located in the submucosa or visceral organs [27], as they often escape detection on CT [28]. Most KS remain asymptomatic [26]. Their correlation with short survival has been ascribed to concomitant infections expressing advanced compromise of the immune system [26]. Treatment focuses on symptomatic lesions and attempts to prolong the decline of immunocompetence [26]. Thus, detection of all but bulky symptomatic variants is not decisive. Non-Hodgkin's lymphoma are the second most common abdominal neoplastic lesions. Extranodal involvement is a typical feature of such tumors in AIDS patients [26]. The detection of lymphomas is one of the most important goals in abdominal CT, as they imply a very poor prognosis and may be included in therapeutic trials. In a larger series, the GI tract was the most common abdominal site for lymphomas [29]. Other than implied by independent variables, such as CD4 + T-lymphocyte count and the presence of ascites, intestinal disease in AIDS has few prognostic implications, if treated properly. Although CMV infection has been shown to adversely affect survival in a cohort of over 1000 patients [30], only 8 patients in this group acquired CMV colitis, whereas over 85 % of CMV patients suffered from retinitis. In the analysis of this study, the CD4-count threshold of 100/mL might well have caused that the CMV infection appeared as an independent prognostic factor, although short survival in CMV infection might have merely reflected even lower CD4 counts. Often, a multifactorial etiology is responsible for intestinal symptoms in AIDS [4]. Thus, only a partial response to therapy directed at an identified pathogen occurs in a considerable number of cases. For an evaluation of the severity and extent of abdominal disease, CT is a sensitive and somewhat less specific method. As a noninvasive procedure, it is also applicable to severely immunodeficient patients with limited cooperation.
1428
F. D. Knollmann et al.: Intestinal disease in acquired immunodeficiency syndrome
Table 5. Features of patients with intestinal mycobacteriosis Patient
Gender
Age (years)
Clinical presentation
Colonoscopic features
Clinical and colonoscopic features 8 F 25 Tuberculous peritonitis
Severe exudative colitis at all levels
28
M
34
Generalized mycobacteriosis, diarrhea Disseminated colonic ulcerations, confluent cecal ulceration
31
M
33
Miliary tuberculosis
Inflamed terminal ileum. Tumor of transverse colon, suspected rupture into small bowel loops
33
M
61
Diarrhea
Nonulcerating inflammation at all levels
Patient
CD4 count
CT features
Patient outcome
CT features and patient outcome 8 452 Small bowel and colonic wall thickening/edema and contrast enhancement, ascites, enteric air-fluid levels
Mycobacterium africanum, alive on day 649
28
41
Mycobacterium avium-intracellulare, died on day 1076
31
36
Colonic wall thickening, periintestinal stranding, mesenterial and retroperitoneal lymphyadenopathy, and mucosal contrast enhancement Small bowel wall thickening, mucosal contrast enhancement, ascites, enteric air-fluid levels Small bowel wall thickening
Mycobacterium avium-intracellulare, alive on day 193
33
15
Mycobacterium avium-intracellulare, alive on day 58
Table 6. Features of patients with other intestinal disease Patient
Age (years)
Clinical presentation
Clinical and colonoscopic features 6 M
39
Watery diarrhea
Within normal limits
43
62
Severe watery diarrhea
Disseminated soft colonic polyps
Patient
Gender
M CD4 count
CT features
Patient outcome
CT features and patient outcome 6 18 Perienteric stranding, mesenterial lymphadenopathy and fat proliferation, retroperitoneal lymphadenopathy 43
1
Small bowel, colonic and rectal wall thickening, ascites, retroperitoneal lymphadenopathy
Conclusions Computed tomography detects signs of intestinal disease in a significant proportion of HIV-infected patients and in the vast majority of AIDS patients with enteric symptoms. Determination of a specific pathogen evades CT diagnosis. The most common causes of CT signs of colitis are CMV colitis and unspecific colitis as per histologic correlation by colonoscopy-directed biopsies. Because colonoscopic evaluation of abdominal disease in AIDS patients remains problematic itself, CT offers a useful adjunct method to detect enteric lesions and to define their extent. In addition, CT provides prognostically important insights into extraenteric disease which may be of a higher therapeutic relevance than the intestinal lesions themselves. References 1. World Health Organization (1995) The current global situation of the HIV/AIDS pandemic. WHO, Global Programme on AIDS 2. Centers for Disease Control and Prevention (1995) First 500 000 AIDS cases: United States, 1995. Morbidity and Mortality Weekly Report 44: 849±853
Colonoscopic features
Enterotoxic E. coli, alive on day 261 Cryptosporidiosis, necrotizing colitis, died on day 11
3. Friedman SL (1990) Approach to AIDS patients with gastrointestinal symptoms. In: Cohen PT, Sande MA, Volberding PA (eds) The AIDS knowledge base. The Medical Publishing Group, Waltham, Mass, pp 5.10.1±1±3 4. Smith PD, Quinn TC, Strober W, Janoff EN, Masur H (1992) Gastrointestinal infections in AIDS. Ann Intern Med 116: 63± 77 5. Kuhlman JE, Fishman EK (1990) Acute abdomen in AIDS: CT diagnosis and triage. Radiographics 10: 621±634 6. Jones B, Wall S (1992) Gastrointestinal disease in the immunocompromised host. Radiol Clin North Am 30: 555±577 7. Feczko PJ (1994) Gastrointestinal complications of human immunodeficiency virus (HIV) infection. Semin Roentgenol 29: 275±287 8. Jeffrey RB Jr (1988) Gastrointestinal imaging in AIDS: abdominal computed tomography and ultrasound. Gastroenterol Clin North Am 17: 507±520 9. Wall SD, Jones B (1992) Gastrointestinal tract in the immunocompromised host: opportunistic infections and other complications. Radiology 185: 327±335 10. Murray JG, Evans SJJ, Jeffrey PB, Halvorsen RA Jr (1995) Cyto-megalovirus colitis in AIDS: CT features. Am J Roentgenol 165: 67±71 11. Wall SD, Ominsky S, Altman DF, Perkins CL, Sollitto R, Goldberg HI, Margulis AR (1986) Multifocal abnormalities of the gastrointestinal tract in AIDS. Am J Roentgenol 146: 1±5 12. Radin DR (1991) Intraabdominal Mycobacterium tuberculosis vs mycobacterium avium-intracllulare infections in patients with AIDS. Am J Roentgenol 156: 487±491
F. D. Knollmann et al.: Intestinal disease in acquired immunodeficiency syndrome 13. Centers for Disease Control and Prevention (1992) 1993 Revised classification system for HIV infection and expanded case definition for AIDS among adolescents and adults. Morbidity and Mortality Weekly Report 41: RR-17, 1±19 14. Cox DR (1972) Regression models and life tables. J R Stat Soc B 34: 187±220 15. Jacobson MA, Mills J (1988) Serious cytomegalovirus disease in the acquired immunodeficiency syndrome (AIDS). Ann Intern Med 108: 585±94 16. Clayton F, Klein EB, Kotler DP (1990) Correlation of in situ hybridization with histology and viral culture in patients with acquired immunodeficiency syndrome with cytomegalovirus colitis. Arch Pathol Lab Med 114: 639 17. Rene E, Marche C, Chevalier T et al. (1988) Cytomegalovirus colitis in patients with acquired immunodeficiency syndrome. Dig Dis Sci 33: 741±750 18. Goodgame G, Genta RM, Estrada R, Demmler G, Buffone G (1993) Frequency of positive tests for cytomegalovirus in AIDS patients: endoscopic lesions compared with normal mucosa. Am J Gastroenterol 88: 338±343 19. Wilcox CM, Schwartz DA (1994) Idiopathic anorectal ulceration in patients with human immunodeficiency virus infection. Am J Gastroenterol 89: 599±604 20. Ulrich R, Zeitz M, Heise W, LaÂge M, HoÈffken G, Riecken EO (1989) Small intestinal structure and function in patients infected with human immunodeficiency virus (HIV): evidence for HIV-induced enteropathy. Ann Intern Med 111: 15±21 21. Morris DJ (1990) Is human immunodeficiency virus (HIV) rather than cytomegalovirus the cause of retinitis and colitis in HIV-infected patients? Rev Infect Dis 12: 557±559
1429
22. Hellyer TJ, Brown IN, Taylor MB, Allen BW, Easmon CS (1993) Gastrointestinal involvement in Mycobacterium aviumintracellulare infection of patients with HIV. J Infect 26: 55±66 23. Balthazar EJ, Gordon R, Hulnick D (1990) Ileocecal tuberculosis: CT and radiologic evaluation. Am J Roentgenol 154: 499± 503 24. Rabeneck L, Gyorkey F, Genta RM, Gyorkey P, Foote LW, Risser JM (1993) The role of microsporidia in the pathogenesis of HIV-related chronic diarrhea. Ann Intern Med 119: 895±899 25. Cotte L, Rabodonirina M, Piens MA, Perreard M, Mojon M, Trepo C (1993) Prevalence of Protozoans in French patients infected with HIV. J Acquir Immune Defic Syndr Hum Retrovirol 6: 102±109 26. Herts BR, Megibow AJ, Birnbaum BA, Kanzer GK, Noz ME (1992) High-attenuation lymphadenopathy in AIDS patients: significance of findings at CT. Radiology 185: 777±781 27. Friedman SL (1989) Gastrointestinal and hepatobiliary neoplasms in AIDS. Gastroenterol Clin North Am 17: 465±486 28. Moon KL, Federle MP, Abrams DL et al. (1984) Kaposi sarcoma and lymphadenopathy syndrome: limitations of abdominal CT in acquired immunodeficiency syndrome. Radiology 150: 479 29. Radin DR, Esplin JA, Levine AM, Ralls PW (1993) AIDS-related non-Hodgkin's lymphoma: abdominal CT findings in 112 patients. Am J Roentgenol 160: 1133±1139 30. Gallant JE, Moore RD, Richman DD, Keruly J, Chaisson RE, and the Zidovudine Epidemiology Study Group (1992) Incidence and natural history of cytomegalovirus disease in patients with advanced human immunodeficiency virus disease treated with zidovudine. J Infect Dis 166: 1223±1227
Notice to authors European Radiology encourages all authors to submit manuscripts printed on both sides of the page. This will save paper and reduce the cost of postage.