|
|
| |||||||||||||||||||||||||||||||
Background Diarrhea occurs frequently among persons with the acquired immunodeficiency syndrome, but the cause often remains unknown. We used a group of diagnostic assays to determine which viruses were etiologic agents of diarrhea in a group of persons infected with the human immunodeficiency virus (HIV).
Methods Stool and serum specimens were obtained from HIV-infected patients enrolled in a longitudinal study in Atlanta. Fecal specimens from patients with diarrhea and from control patients without diarrhea were screened by electron microscopy, polyacrylamide-gel electrophoresis, and enzyme immunoassays for rotaviruses, enteric adenoviruses, caliciviruses, picobirnaviruses, and astroviruses. Paired serum samples were tested for antibody responses to Norwalk virus and picobirnavirus.
Results Viruses were detected in 35 percent of 109 fecal specimens from patients with diarrhea but in only 12 percent of 113 specimens from those without diarrhea (P<0.001). Specimens from patients with diarrhea were more likely than those from patients without diarrhea to have astrovirus (12 percent vs. 2 percent, P = 0.003); picobirnavirus (9 percent vs. 2 percent, P = 0.017); caliciviruses, including small round structured viruses (6 percent vs. 1 percent, P = 0.062); and adenoviruses (9 percent vs. 3 percent, P = 0.047). They were also more likely to have a mixed viral infection (6 percent vs. 0 percent, P = 0.006). With the use of polyacrylamide-gel electrophoresis to analyze concentrated RNA extracts from stool, picobirnavirus was detected in fecal specimens from 6 of the 65 patients with diarrhea and was associated with prolonged viral shedding and chronic diarrhea. No rotaviruses, enteric adenoviruses, or instances of seroconversion to positivity for Norwalk virus were observed.
Conclusions Novel enteric viruses such as astrovirus and picobirnavirus may be more important etiologic agents of diarrhea in HIV-infected patients than previously recognized and may be more common than either bacterial or parasitic enteropathogens.
Several viruses have been implicated as enteric pathogens in patients with T-cell deficiency from other causes, such as chemotherapy before bone marrow transplantation or severe combined immunodeficiency syndrome,23,24 but studies in patients with AIDS have yielded inconsistent results8,25,26,27,28,29. The reported prevalence of viruses is generally low. Only the more common viruses (e.g., rotavirus and enteric adenovirus) have been sought, and these agents have not consistently been associated with diarrhea. Novel enteric viruses, such as astrovirus, caliciviruses, and non-group A rotaviruses, have not been investigated, because diagnostic methods are not widely available for these agents30. The identification of unusual serotypes of enteric adenoviruses in patients with AIDS suggests that other novel viruses may also be present31,32.
We have recently expanded our repertoire of assays for the novel agents of viral gastroenteritis to include polyacrylamide-gel electrophoresis (PAGE) to detect atypical rotaviruses33 and picobirnaviruses,34 a new enzyme immunoassay to detect astroviruses,35 and a concentration technique to increase the sensitivity of detecting viral agents in fecal specimens by direct electron microscopy25,36. To identify possible associations between the presence of these novel viruses and diarrhea, the stage of HIV infection, and the CD4 count, we applied these methods to the study of fecal specimens obtained from a group of HIV-infected patients, some of whom had diarrhea and some of whom did not.
Methods
Study Population
This investigation was conducted between January 1 and December 31, 1991, as part of an ongoing prospective study of enteric opportunistic infections in a cohort of HIV-infected patients followed at three Atlanta hospitals affiliated with the Emory University School of Medicine37. The study was approved by the appropriate institutional review boards, and informed consent was obtained from all participants. Outpatients who were HIV-positive and willing to be seen at monthly intervals for one year were eligible for enrollment. At recruitment and at monthly intervals thereafter, the patients were given a questionnaire, examined, and asked to provide blood and stool specimens. Between January 1 and December 31, 1991, 211 patients were enrolled in the study and followed for more than one month. Follow-up data on the 91 patients who did not have diarrhea at enrollment were used to calculate the incidence of new cases of diarrhea in this cohort.
Etiologic studies of viral diarrhea were conducted in the fall of 1991 with specimens that had been collected from the 110 patients enrolled during the first eight months of the study and followed for two months or more. To examine the association between viral agents and diarrhea, we compared the rates of viral detection in specimens obtained from patients who had diarrhea (case patients) with those in specimens obtained from patients who were without diarrhea (controls). This case-control design based on specimens avoided the problem of counting the number of patients who had diarrhea and the number who did not have diarrhea, since the duration of follow-up ranged widely from two to eight months. To avoid collecting more than one specimen per episode of diarrhea, we considered case specimens to be the first stools obtained from patients who had diarrhea at entry to the study or who had one or more new episodes of diarrhea by the cutoff date, August 31, 1991. For each case specimen, a control specimen was obtained from a patient who during the same month (or within one month) did not have diarrhea at the time of the clinic visit or in the previous or subsequent month. Since the period of viral shedding in gastroenteritis is usually short, we ensured the independence of the observations by obtaining only the first specimen per episode of diarrhea and requiring the control specimen to be collected after a diarrhea-free period of at least one month. Thus, any questions about whether prolonged shedding had occurred could be addressed at a later time by the screening of all serial specimens (i.e., those collected monthly) obtained from a single patient or group of patients. Of the 110 patients enrolled by August 31, 65 patients had a total of 109 diarrheal episodes; 113 control specimens were obtained from 65 patients, including 2 patients who reported having diarrhea at enrollment but who had formed stools. Fifty-nine patients, primarily those recruited late in the study, provided only one specimen for evaluation, whereas 51 provided two or more specimens.
Diarrhea was defined as the occurrence of three or more semiliquid or watery bowel movements within a 24-hour period and was considered acute if it lasted 28 days or less and chronic if it persisted for longer periods. Fecal specimens from patients with diarrhea at enrollment or with chronic diarrhea were usually liquid, and their collection coincided with the presence of symptoms of illness. Specimens from patients with acute diarrheal episodes were obtained as long as 4 weeks after the acute episode, although we found in practice that most were collected within 10 to 14 days of the illness, and a majority of the patients were ill on the day of collection. An episode was considered to be over when a one-month period that included a clinic visit passed without the occurrence of diarrhea.
Fecal Specimens
Specimens for viral diagnosis were stored at 4 °C and tested under standard conditions. To prepare the suspensions, 1.5 g of stool was mixed in 15 ml of phosphate-buffered saline (pH 7.0) and centrifuged at 2000 x g for 10 minutes36. The supernatant mixed with an equal volume of 1,1,2-trichlorotrifluoroethane (Genetron) was placed in a vortex for 1 minute, held at 4 °C for 60 minutes, placed in a vortex again, and then centrifuged at 3000 x g for 5 minutes.
The supernatant was tested for group A rotavirus, adenovirus hexon group antigen, and astrovirus by enzyme immunoassay35,38. Specimens positive for adenovirus hexon, the common group antigen, were tested for enteric serotypes 40 and 41 with a commercial assay (Cambridge BioSciences, Worcester, Mass.). Positive results were confirmed by analysis with the polymerase chain reaction (PCR) according to methods developed at the Centers for Disease Control and Prevention (CDC) (Monroe SS: unpublished data). In brief, viral RNA extracted from stool specimens39 was tested with astrovirus primers common to serotypes 140 and 241. Reverse transcription (42 °C for one hour) followed by 30 cycles of PCR (90 °C for one minute, 40 °C for two minutes, and 72 °C for one minute) yielded products that were analyzed by PAGE. An aliquot (400 microl) of the supernatant was extracted with phenol-chloroform, and the aqueous phase was tested by PAGE (7.5 percent acrylamide) and silver staining to identify atypical rotaviruses and picobirnaviruses42.
The remaining supernatant (10 ml) was ultracentrifuged at 100,000 x g for two hours. The pellet was incubated overnight at 4 °C in 100 microl of phosphate-buffered saline and then resuspended, and 20 microl was floated on a 400-microm-mesh carbon-coated grid for 1 minute (Formvar, Electron Microscopy Sciences, Fort Washington, N.J.), negatively stained with 2 percent phosphotungstic acid (pH 6.5) for 45 seconds, and examined in a Philips 201 electron microscope (Philips, Mahwah, N.J.)25,36. The remaining pellet (80 microl), representing a 15-fold to 30-fold concentration of virus from the stool, was tested by PAGE. Because the use of the pellet significantly increased the rate of detection of picobirnaviruses, all results for picobirnaviruses are based on this method.
In this study, all fecal specimens (from patients with diarrhea and from controls) were evaluated for potential parasitic pathogens, including cryptosporidium species and E. bieneusi, as previously described37. Routine bacteriologic analysis for species of salmonella, shigella, campylobacter, and yersinia and special studies for Clostridium difficile toxin and Mycobacterium avium complex were done at the discretion of the patients' physicians, and these results are not reported here. All available parasitologic and bacteriologic results were reviewed for patients whose diarrhea was associated with viruses.
Serum Specimens
In the six patients from whom picobirnaviruses were isolated, the immune response to infection was examined under standard conditions by immunoelectron microscopy43; each patient's paired serum samples and the corresponding stool specimens were examined. A sample of gamma globulin (dilution, 1:5) was also tested for antibody to picobirnavirus by immunoelectron microscopy. Paired serum samples from 25 patients with diarrhea in whom no enteropathogen was identified were tested for seroconversion to the Norwalk virus with a biotin-avidin assay44. For each patient the CD4 counts were determined at the participating hospitals with commercial methods.
Statistical Analysis
The rates of detection of viruses in the specimens from case patients and controls, as well as the number of patients contributing specimens in each group, were tested for significance with Fisher's exact test (two tailed).
Results
The patients in this cohort had a broad range of socioeconomic backgrounds; 25 percent were from minority groups, 85 percent had acquired their HIV infection by homosexual contact, and all but two were male. At entry, 79 of the 110 patients (72 percent) were in CDC clinical stage IV-C45 and had had some opportunistic infection. The 91 patients who did not have diarrhea on enrollment and who were followed for more than one month during the first year had 39 episodes of diarrhea during 348 person-months of follow-up, an incidence of 11.1 episodes per 100 person-months of observation (8.9 of acute diarrhea and 2.2 of chronic diarrhea). One third of the patients with acute diarrhea at admission went on to have chronic diarrhea, and this resolved in 37 percent of the patients regardless of whether treatment was administered.
Seven different viruses were identified in the 222 specimens examined (Table 1), and these were found significantly more often in the stools of patients with diarrhea (35 percent) than in the specimens from patients without diarrhea (12 percent). Astrovirus and adenovirus were significantly more common in the specimens from patients with diarrhea than in those from patients with no diarrhea. Human calicivirus and small round structured viruses (SRSVs) also tended to be more common in specimens from patients with diarrhea. Since the classification of SRSVs according to findings on electron microscopy includes the human caliciviruses46 and since recent cloning results indicate that two SRSVs, the Norwalk47 and Southampton48 viruses, are both caliciviruses, we combined these groups for analysis; SRSVs or caliciviruses were more often found in specimens from patients with diarrhea than in those from controls (P = 0.06). Finally, picobirnaviruses were the second most common virus detected and were also found significantly more often in patients with diarrhea. All the astroviruses detected by enzyme immunoassay were confirmed by electron microscopy or PCR. All the adenoviruses detected were confirmed by enzyme immunoassay, and none were of enteric serotype 40 or 41. No rotaviruses, herpesviruses, or cytomegaloviruses were detected. Finally, specimens from patients with diarrhea were more likely to be associated with a mixed viral infection than were specimens from patients without diarrhea (6 percent vs. 0 percent, P = 0.006).
|
To examine the duration of excretion of picobirnavirus, we screened all serial specimens (i.e., those obtained monthly) available from the six infected patients. One patient with chronic diarrhea shed picobirnavirus continuously for seven months, a pattern consistent with the length of his illness, and another asymptomatically shed picobirnavirus before and after an episode of acute picobirnavirus-associated diarrhea. In two other patients, picobirnavirus was detected in association with a history of chronic diarrhea, but serial stool specimens were not available for further testing.
We reviewed the parasitologic and bacteriologic results for diarrheal specimens from which viral agents were isolated. All specimens were tested for parasites, and four were positive: one specimen from which astrovirus was isolated was positive for Giardia lamblia, one from which an SRSV was isolated was positive for microsporidia, and two from which picobirnavirus was isolated (both were from the same patient) were positive for cryptosporidia. Half the specimens were also tested for bacteria, C. difficile, and mycobacterium. One specimen from which picobirnavirus was isolated was collected from a patient with diarrhea on entry to the study and was also positive for a mixed infection with Shigella flexneri and M. avium complex.
This low rate of mixed infections was consistent with the other etiologic findings in the cohort. Relatively few patients with diarrhea had an enteric bacterial pathogen identified in their stools. Intestinal parasites were assayed in specimens from 195 patients with diarrhea, 103 with acute diarrhea and 92 with chronic diarrhea; a parasite was detected in 3 percent of the specimens from patients with acute diarrhea and 22 percent of those from patients with chronic diarrhea (P<0.001). Cryptosporidium species, E. bieneusi, and Entamoeba histolytica were found more frequently in patients with chronic diarrhea than in those with acute diarrhea (11 vs. 2 percent, P = 0.02; 7 vs. 0 percent, P = 0.005; and 2 vs. 0 percent, P = 0.22, respectively). Routine bacterial pathogens were found with similar frequency (11 percent) in the two groups; there were low rates of detection (range, 3 to 6 percent) for species of salmonella, shigella, and campylobacter, and the rates were not significantly different between groups. Only 1 of 43 specimens from the patients with chronic diarrhea was positive for C. difficile toxin.
To search for an association between the CD4 counts and the presence of either diarrhea or a viral agent, the CD4 counts were examined in each group of patients by stratifying the patients according to the stage of HIV infection. Although the CD4 counts decreased as the stage of infection increased, they did not differ significantly between patients with enteric viruses or diarrhea and those without enteric viruses or diarrhea.
The electropherotypes of picobirnavirus from two patients who shed virus for several months were examined for strain differences over time (Figure 1). As compared with Patient 2, Patient 1 consistently shed large quantities of the virus, and the patterns did not change over time. To visualize the weak bands observed in the specimen from Patient 2 better, we prepared RNA from more concentrated stool suspensions: a 50 percent suspension and the concentrated pellet prepared for electron microscopy (Table 2). When the entire collection of 222 fecal specimens and selected serial specimens from picobirnavirus-infected patients were retested, the detection rate increased 400 percent; 4 picobirnaviruses were seen with the 10 percent suspension, 10 were observed with the 50 percent suspension, and 16 were detected with the concentrated pellet. On the basis of this finding, the results of PAGE with the concentrated pellets are presented for picobirnavirus. Electron microscopy was less sensitive than PAGE when either of the more concentrated stool suspensions was used. By electron microscopy, the viruses measured 35 to 40 nm in diameter, showed no distinct surface structure, and were always seen as individual particles, without grouping, clumping, or surface antibody. Immunoelectron microscopy of serial serum samples from patients recovering from diarrhea showed no aggregation of virus and no deposition of antibody on the surface, and the addition of gamma globulin did not cause the particles to aggregate or lead to the deposition of antibody.
|
|
Discussion
Diarrhea was a common problem in this HIV-infected cohort, affecting half the patients at entry and having an incidence of 11 percent per month among those without diarrhea at enrollment. The detection rates for viruses were higher than those for parasitic and bacterial agents, suggesting that viruses may be more common agents of diarrhea in HIV-infected patients than has been previously suspected. Viruses were found significantly more often in fecal specimens from patients with diarrhea than in specimens from patients without diarrhea, and two novel viruses -- astroviruses and picobirnavirus -- each contributed significantly to this association. Other serotypes of adenoviruses and calicivirus or SRSV were also more common in specimens from patients with diarrhea than in those from patients without diarrhea. Most patients had severe HIV-associated disease classified as CDC clinical stage IV-C, but the prevalence of enteric viruses was not associated with the CD4 counts. Our results show that astrovirus must be recognized as a potential agent of diarrhea in persons with AIDS. We also show that picobirnavirus can be found in fecal specimens from patients with diarrhea in the United States and in HIV-infected persons.
These results differ from those of earlier studies conducted in the United States and abroad (Table 3)8,25,26,27,28,29,49. Previous studies selected patients with different clinical presentations (e.g., gastroenteritis and colitis) and stages of HIV infection, used alternative diagnostic approaches (e.g., cultivation, histopathological analysis, and electron microscopy), and detected some herpesviruses not normally associated with diarrhea, which were then used in calculating the rates of viral detection. The inability to link diarrhea to specific viral agents may be a result of small samples, low rates of viral detection, and the variety of agents identified. Our study used case patients and control groups of comparable size, used more sensitive assays for astrovirus and picobirnavirus (enzyme immunoassay, stool concentrates for PAGE and electron microscopy, and PCR), and involved HIV-infected persons whose disease was already in the opportunistic-infection stage. One study in 1987 used PAGE to detect group A and atypical rotaviruses but did not detect picobirnaviruses, which had only recently been identified26.
|
This ongoing, preliminary study has several limitations. Given the relatively small study population, the observed associations, although interesting, require confirmation with larger samples and in varied clinical and epidemiologic settings. We chose the case-control design using fecal specimens as the unit of observation because we could separate diarrheal and non-diarrheal episodes clearly in time, did not anticipate the occurrence of prolonged viral shedding, and believed that comparing patients whose duration of follow-up varied greatly would be problematic. Consequently, the significant association of picobirnavirus with diarrheal specimens could be explained in part by a clustering of 12 such findings in six patients. The fact that some patients had separate episodes of picobirnavirus-associated diarrhea interspersed with periods without illness or shedding does not help clarify this association and leads one to question the independence of these events. Consequently, although we have detected a newly described virus in 6 of 65 HIV-infected adults with diarrhea and demonstrated that it was often associated with prolonged shedding and chronic diarrhea, its role as an etiologic agent of diarrhea in these patients remains to be established. Finally, although the rates of viral detection were high, all viruses detected were confirmed with alternative methods. We did look at associated bacterial and parasitic infections and could not attribute our findings to the occurrence of mixed infections.
In summary, we have found astrovirus, calicivirus, and adenovirus to be significantly associated with diarrhea in HIV-infected persons who, by virtue of their severe immunodeficiency (CDC HIV stage, IV-C), were already susceptible to other opportunistic infections. Astrovirus, an established agent of diarrhea in children, appears to affect HIV-infected patients as well38,56,57. The findings of picobirnaviruses in this population, their association with prolonged shedding, and the need for special detection methods are interesting, but the finding of a significant association with diarrheal specimens was confounded by clustering of the virus in several patients and remains to be clarified. Although preliminary, our results should encourage others to use the most sensitive assays available in a careful search for novel infectious agents (viral, parasitic, and bacterial) as potential causes of diarrhea in HIV-infected persons.
Supported by funds from the National Center for Infectious Diseases, Centers for Disease Control and Prevention, and the Public Health Service, Department of Health and Human Services, and by a special research grant (YO2-AI-90002-02) from the National Institute of Allergy and Infectious Diseases Intra-Agency Agreement to Dr. Glass.
We are indebted to Cesar Carcamo, Barbara Forrester, Sarah Stine, Jon Gentsch, Marion Koopmans, John O'Connor, and participants in the Enteric Opportunistic Infections Study for their help with this study.
Source Information
From the Viral Gastroenteritis Unit, Respiratory and Enteric Viruses Branch, Division of Viral and Rickettsial Diseases (G.S.G., R.I.G., H.G.P., S.S.M.), and the Parasitic Diseases Branch, Division of Parasitic Diseases (A.W.H., R.W., R.T.B.), National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta. The other members of the Enteric Opportunistic Infections Working Group are as follows: D.D. Juranek, D.G. Addiss, G.S. Visvesvara, J.M. Stewart, J.M. Roberts, S.P. Wahlquist, L. Gorelkin, C.R. Horsburgh, Jr., K.G. Castro, R.V. Tauxe, and D.J. Vugia, Centers for Disease Control and Prevention; S.E. Thompson, III, C.M. Wilcox, K.O. Hewan-Lowe, D.A. Schwartz, P.E. Kozarsky, J.P. Steinberg, J.A. Shulman, R.M. Dismukes, M.H. DuPuis, J.F. Nickerson, D. Rimland, S.E. Hogan, and A. Johnson, Emory University School of Medicine, Atlanta; and N. Elliott, Crawford Long Hospital, Atlanta.
Address reprint requests to Dr. Glass at the Viral Gastroenteritis Unit (GO4), Centers for Disease Control and Prevention, 1600 Clifton Rd., Atlanta, GA 30333.
References
| |||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
