Early Manifestations of Disseminated Mycobacterium avium Complex Disease: A Prospective Evaluation

Transcription

1 126 Early Manifestations of Disseminated Mycobacterium avium Complex Disease: A Prospective Evaluation Fred M. Gordin, David L. Cohn, Paul M. Sullam, John R. Schoenfelder, Beverley A. Wynne,* and C. Robert Horsburgh, Jr. Division of Infectious Diseases, Department of Veterans Affairs Medical Center, and Georgetown University, Washington, DC; Denver Diseases Control Services, Denver Health and Hospitals and Division of Infectious Diseases, University of Colorado Health Sciences Center, Denver; Division of Infectious Diseases, Department of Veterans Affairs Medical Center, and University of California, San Francisco; Pharmacia & Upjohn, Columbus, Ohio; Division of Infectious Diseases, Department of Medicine, Emory University, Atlanta, Georgia A nested case-control study was conducted in two trials of prophylaxis for Mycobacterium avium complex (MAC) infection to describe the specific signs, symptoms, and laboratory abnormalities of MAC disease in AIDS. Patients had 200/mm 3 CD4 cells and a prior AIDS-defining illness. Of 571 patients, 102 (17.9%) developed MAC bacteremia during a mean follow-up of 256 days. Among cases of MAC disease, 90 were compared with 180 matched controls. Patients with MAC disease were more likely than controls to have lower weights (66.3 vs kg, P Å.001) and Karnofsky scores (74.3 vs. 84.4, P õ.001); a higher proportion had fever (48% vs. 26%, P Å.003), abdominal pain (23% vs. 13%, P Å.05), decreased hemoglobin levels (10.9 vs g/dl, P õ.001), and elevated alkaline phosphatase (203 vs. 138 U/L, PÅ.04) and lactate dehydrogenase (334 vs. 280 U/L, P Å.02) levels. Characteristics of MAC disease that occurred before bacteremia were weight loss (3 months prior), fever (2 months), and anemia and elevated lactate dehydrogenase (1 month). These data suggest that patients have symptomatic MAC disease for several months prior to the occurrence of bacteremia. Disseminated Mycobacterium avium complex disease is now function values have all been reported as manifestations of the most common bacterial infection among persons with advanced disseminated M. avium complex infection. However, all of AIDS [1 4], and it has been estimated that up to one- these signs and symptoms have also been associated with ad- fourth of all AIDS patients will acquire this infection during vanced HIV infection itself, as well as with other opportunistic their lifetimes [5]. Despite the occurrence of M. avium complex infections and malignancies. disease in large numbers of patients, its clinical manifestations Prior studies of disseminated M. avium complex infection and natural history have not been well delineated. While retro- have been limited by the analysis of data from patients in a spective and descriptive studies have reported the signs and single geographic region [6 8], the lack of uniform methods symptoms present in persons with M. avium complex disease for the detection of M. avium complex or standardized defini- [1, 3 10], it is not clear which clinical features are due to M. tions of disseminated disease [9, 10], and the lack of a wellavium complex and which are related to the underlying human matched cohort of patients with similar severity of HIV disease immunodeficiency virus (HIV) infection or other concomitant but without M. avium complex infection [7, 8]. To overcome illnesses. Fever, weight loss, night sweats, fatigue, diarrhea, these limitations, we evaluated the occurrence of disseminated lymphadenopathy, organomegaly, anemia, and elevated liver M. avium complex disease in patients who were enrolled in the placebo arms of two identical studies of M. avium complex prophylaxis [11]. These patients were from many geographic sites in the United States and Canada, fulfilled standardized Received 2 August 1996; revised 31 January eligibility criteria, were followed using uniform definitions of Presented in part: 35th Interscience Conference on Antimicrobial Agents and Chemotherapy [abstract I-92], San Francisco, 18 September clinical status, and had blood drawn monthly for mycobacterial Written, informed consent was obtained from all patients, and human experi- culture. We were able to evaluate this group of high-risk pamentation guidelines of the US Department of Health and Human Services tients to determine risk factors for disseminated M. avium comand those of the participating institutions were followed in the conduct of this clinical research. plex disease. We then created a matched cohort of patients J.R.S. and B.A.W. were employed by Pharmacia & Upjohn, Inc., at the time with a similar level of immunosuppression but without dissemof the study. inated M. avium complex disease in order to differentiate the Financial support: Pharmacia & Upjohn. Reprints or correspondence: Dr. Fred M. Gordin, Infectious Diseases, VA signs and symptoms associated with M. avium complex disease Medical Center (151B), 50 Irving St., NW, Washington, DC from those related to HIV infection. The time of onset of these * Present affiliation: Roxanne Laboratories, Columbus, Ohio. signs and symptoms in relation to the detection of M. avium The Journal of Infectious Diseases 1997;176: complex bacteremia and the impact of disseminated M. avium 1997 by The University of Chicago. All rights reserved /97/ $02.00 complex disease on survival were evaluated.

2 JID 1997;176 (July) Disseminated M. avium Infection 127 Methods Comparisons were made at enrollment and at the time of M. avium complex disease (or the equivalent time for the controls). Patients were enrolled in two identical double-blind, placebo- In order to assess a relationship to the time of M. avium complex controlled trials of rifabutin for prophylaxis of disseminated M. bacteremia, additional comparisons were made at each of the 5 avium complex disease, as previously reported [11]. For the pur- months preceding the time of M. avium complex bacteremia (or pose of this analysis, only patients who were randomly assigned the equivalent time for the controls). Survival was assessed in to receive placebo were included. At enrollment, patients were terms of days following the date of the first blood culture positive required to be at least 18 years old, have 200/mm 3 CD4 lympho- for M. avium complex. All statistical tests were two-sided and cytes, and have a prior AIDS-defining illness [12]. All patients were conducted at a.05 level of significance. were required to be receiving antiretroviral therapy and prophylaxis against Pneumocystis carinii at study entry. Patients were excluded if either of 2 baseline blood cultures or a baseline stool culture was positive for M. avium complex. They were prohibited Results from taking any drugs with antimycobacterial activity during the Patients were recruited from February 1990 through January trials. Patients were evaluated every 4 weeks using standardized 1992 at 73 centers in the United States and Canada; 580 were forms for recording changes in symptoms or physical findings. At randomized to receive placebo within the 2 studies. These paeach visit, patients had complete blood cell counts and selected tients were predominantly male (97%) and white (88%), and chemistry tests performed. Lymphocyte subset evaluations were done every 3 months. they had a mean age of 37.7 years ({8.1). The average patient Patients had a single blood culture for mycobacteria done at had been diagnosed with AIDS for slightly over 1 year, and each monthly visit. All cultures were processed at one of two the mean and median CD4 cell counts of the population were central laboratories using broth radiometric techniques [13]. Patients 56 and 36/mm 3, respectively. who were found to have M. avium complex bacteremia Of the 580 patients, 9 had M. avium complex bacteremia at were classified as having disseminated disease. These patients were enrollment and were removed from further analysis, leaving reevaluated every 1 3 months thereafter and received antimyco- 571 who were evaluated monthly. During a mean length of bacterial therapy at the discretion of their care provider. follow-up of 256 days, 102 (17.9%) of the patients developed To determine the impact of disseminated M. avium complex M. avium complex bacteremia, while 469 (82.1%) did not. disease on various clinical and laboratory parameters, we devel- Using Kaplan-Meier methodology, the 1-year incidence of M. oped a matched cohort of patients who were participating in the avium complex bacteremia was 24.6% for the entire cohort study but who did not develop disseminated M. avium complex (controls) disease. As an initial step in developing this control [11]. When determined by CD4 cell subsets at baseline, the 1- cohort, a proximate CD4 cell count was determined for patients year incidence of disseminated M. avium complex disease was with disseminated M. avium complex disease (cases) and was defor 33.6% for persons with CD4 cell counts of 0 25/mm 3, 22.8% fined as the CD4 cell count nearest to, but not later than, the persons with CD4 cell counts of 26 50/mm 3, 35.6% for time of disseminated M. avium complex disease. Furthermore, the persons with CD4 cell counts of 51 75/mm 3, 2.9% for persons proximate CD4 cell count had to be taken no longer than 90 days with CD4 cell counts of /mm 3, and 9.4% for persons before the time M. avium complex bacteremia was first detected. with CD4 cell counts ú100/mm 3. For each case, the proximate CD4 cell count and the number of The enrollment characteristics of the patients who did and days from the proximate CD4 cell count to the detection of M. did not develop disseminated M. avium complex disease are avium complex bacteremia ( M. avium complex lag ) were calcushown in table 1. Patients who developed disseminated M. lated. Controls were chosen for each case by finding a patient at avium complex bacteremia had substantially lower mean CD4 the same investigational site who did not develop disseminated M. avium complex disease but who had a CD4 cell count similar cell counts at baseline than those who did not (35 vs. 61/mm 3 ). ({10) to that of the case. For the control, the time of equivalent In addition, those who developed M. avium complex bacteremia M. avium complex was the date of the matching CD4 cell count had statistically significantly lower weights and total white plus the M. avium complex lag. Therefore, the time of equivalent blood cell counts, although the absolute differences were small. M. avium complex would be the same as the baseline time or time In addition, patients who developed M. avium complex disease the blood was drawn for culture from patients with M. avium had small but significantly higher values for serum glutamicbacteremia. This process was repeated to create 2 controls for each oxaloacetic transaminase (SGOT), bilirubin, and alkaline phoscase. phatase at baseline. There were no differences between persons The focus of the statistical analyses was to compare cohorts of who acquired M. avium complex disease and those who did patients who did and did not develop M. avium complex bacteremia not with respect to age, race, or sex, although relatively few with respect to characteristics that were possibly associated with women or nonwhite persons were enrolled (data not shown). M. avium complex bacteremia. Comparisons with respect to quantitative variables were made using the Wilcoxon rank sum test or In order to differentiate the clinical and laboratory abnormal- a t test. Categorical variables were compared using either the x 2 - ities of the patients who developed disseminated M. avium square or Fisher s exact test. Kaplan-Meier methodology with a complex disease from changes due to underlying HIV infection log-rank test was used to compare the groups with respect to or other concomitant diseases, a matched cohort was developed survival. as described in Methods. Of the 102 patients with disseminated

3 128 Gordin et al. JID 1997;176 (July) Table 1. Risk factors for disseminated M. avium complex (MAC) shown in table 3. Differences in weight between the 2 groups bacteremia. were apparent as early as 3 months before the onset of M. avium complex bacteremia. Fever became associated with M. Enrollment characteristic Bacteremia No bacteremia (means) (n Å 102) (n Å 469) P avium complex bacteremia Ç2 months prior to the first positive blood culture. The presence of more marked anemia and ele- Age (years) NS vated LDH levels occurred 1 month prior to the detection of Weight (kg) M. avium complex disease; abdominal pain, fatigue, a lower Karnofsky score (%) NS Karnofsky score, and an elevated alkaline phosphatase level CD4 cell count (/mm 3 ) õ.001 Hemoglobin (g/dl) only became statistically more common in patients with dis- WBCs (110 9 /L) õ.001 seminated M. avium complex disease at the time the first blood Platelets (110 9 /L) NS culture was positive. Night sweats and diarrhea appeared to SGOT (U/L) become associated with M. avium complex disease at the time SGPT (U/L) NS of bacteremia. Bilirubin (mg/dl) Alkaline phosphatase (U/L) The impact of M. avium complex bacteremia on survival was evaluated using individuals in the disseminated M. avium NOTE. WBCs, white blood cells; SGOT, serum glutamic-oxaloacetic complex disease and control cohorts (figure 1). The median transaminase; SGPT, serum glutamate-pyruvate transaminase; NS, not signifisurvival from the time of the initial positive blood culture was cant. 264 days for those with disseminated M. avium complex and 592 days from the time of equivalent M. avium for controls M. avium complex disease, 90 had a CD4 cell count taken 90 who did not develop disseminated M. avium complex disease days before the development of M. avium complex bacteremia. (P õ.01). Each of the 90 patients with disseminated M. avium complex (cases) was matched by CD4 cell count at the time of M. avium complex bacteremia with 2 culture-negative controls, for a total Discussion of 180 patients in the control cohort. The mean CD4 cell count This prospective evaluation provided a unique opportunity at the time of M. avium complex bacteremia was 20 for the to define the early manifestations of M. avium complex disease cases and 21 at the time of comparison for the controls. in individuals with HIV infection. A large number of high- The clinical and laboratory characteristics for the cases at risk patients were followed with monthly mycobacterial blood the time of M. avium complex bacteremia and the controls at the cultures in multiple geographic sites throughout the United equivalent time are shown in table 2. At the time of bacteremia, States and Canada. Even more important, the presence of a patients with M. avium complex disease had significantly lower weights and Karnofsky scores, and a higher proportion had fever, night sweats, and abdominal pain than patients in the Table 2. Clinical and laboratory characteristics of cases at time of matched cohort. In addition, patients with disseminated M. M. avium complex bacteremia and of matched controls. avium complex disease had lower values for hemoglobin and higher values for alkaline phosphatase and lactate dehydroge- Cases Controls nase (LDH) than did the controls. Night sweats and diarrhea Characteristic (n Å 90) (n Å 180) P appeared to be more common in patients with disseminated M. Weight* (kg) avium complex disease, although these differences were not Karnofsky score* õ.001 statistically significant. Six (7%) of 90 individuals with dissem- Fever (%) inated M. avium complex disease had no associated signs, Night sweats (%) symptoms, or laboratory abnormalities. Abdominal pain (%) Diarrhea (%) In order to determine the relation of clinical and laboratory Fatigue (%) abnormalities with time of detection of M. avium complex Hemoglobin* (g/dl) õ.001 bacteremia, specific measured parameters were compared for WBCs (110 9 /L) the case and control cohorts for each of the 5 months prior to Platelets* (110 9 /L) the first blood culture positive for M. avium complex or match- SGOT* (U/L) SGPT* (U/L) ing date (for controls). No significant differences were observed Bilirubin* (mg/dl) at 4 or 5 months before the development of disseminated M. Alkaline phosphatase* (U/L) avium complex disease. LDH* (U/L) In the 3 months prior to disease, parameters that became statistically significant (P õ.05) between the case and control NOTE. WBCs, white blood cells; SGOT, serum glutamic-oxaloacetic cohorts and the time of onset of these differences in relation to the first positive blood culture for M. avium complex are transaminase; SGPT, serum glutamate-pyruvate transaminase; LDH, lactate dehydrogenase. * Mean values.

4 JID 1997;176 (July) Disseminated M. avium Infection 129 Table 3. Time course of signs, symptoms, and laboratory abnormalities associated with disseminated M. avium complex (MAC) disease of cases versus controls. At time of MAC infection 1 month prior 2 months prior 3 months prior Cases Controls P Cases Controls P Cases Controls P Cases Controls P Weight* (kg) Karnofsky score* õ NS NS NS Abdominal pain (%) NS NS NS Fever (%) NS Night sweats (%) NS NS NS Diarrhea (%) NS NS NS Hemoglobin* (g/dl) õ õ NS NS Alkaline phosphatase* (U/L) NS NS NS LDH* (U/L) NS NS NOTE. NS, not significant (P ú.10); LDH, lactate dehydrogenase. * Mean values. matched cohort of patients with equally severe immunosup- were diagnosed early in their disease because of the monthly pression allowed the impact of disseminated M. avium complex culture of blood for mycobacteria. This is supported by the disease to be distinguished from other changes related to ad- high frequency of asymptomatic patients. The abnormalities in vanced HIV infection. Overall, the 1-year incidence of M. signs and symptoms shown in table 2 may more accurately avium complex bacteremia was 24.6%, with the highest 1-year reflect the abnormalities that present at the onset of disseminated rates for individuals with õ75 CD4 cells/mm 3. This incidence M. avium complex disease than in series reporting abnorrates is similar to that reported by others [7, 9, 14]. malities in persons who may have had M. avium complex The only clinically important predictor of the development infection for some time prior to diagnosis. Another indication of M. avium complex bacteria was the CD4 lymphocyte count that our patients may have been diagnosed earlier in their disease at baseline. Patients who developed disseminated M. avium is that we found the hemoglobin levels to be substantially complex disease had significantly lower mean CD4 cell counts higher in our patients than in patients in other studies. Our (35 vs. 61/mm 3 ) than persons who did not (table 1). Several patients had a mean hemoglobin concentration of 10.9 g/dl at other characteristics were also correlated with the development the time of initial diagnosis of M. avium complex bacteremia, of M. avium complex disease; however, the small differences while others have reported that 85% of patients have a hemo- in values are not likely to be clinically useful. As a group, globin concentration of õ8.5 g/dl [4]. Other abnormalities persons who became infected with M. avium complex weighed reported as being associated with M. avium complex disease less at baseline (68.8 vs kg). In addition, there were include weight loss and fatigue, reported for ú60% of our statistically significant elevations in liver function values in patients, and abdominal abnormalities (including pain and diar- persons who developed M. avium complex disease, although rhea), found in about one-fourth to one-third of our patients. the absolute differences in levels of SGOT, bilirubin, and alkaline To differentiate the symptoms associated with M. avium phosphatase were small. In other series, anemia has been complex disease from symptoms due to other HIV-related ab- associated with an increased risk of developing M. avium complex normalities in patients with equally advanced HIV disease, we bacteremia; however, in our study, the baseline hemoglo- matched cases of M. avium complex disease with controls by bin level was only marginally lower in patients who developed CD4 cell count. CD4 cell counts were used as the most power- M. avium complex bacteremia (12.2 vs g/dl, P Å.06). ful single predictor of HIV disease progression, as virus load At the onset of M. avium complex bacteremia, there were measurements were not available for this cohort. The data in fewer signs and symptoms than have been reported by others table 2 demonstrate that many of the signs and symptoms (table 2). For example, in our group, only 48% of patients had associated with M. avium complex disease occur in individuals fever, and 24% had night sweats. In a study of 114 patients at the same stage of HIV disease but without M. avium complex with disseminated M. avium complex disease reported from infection. Despite the overlap of abnormalities between the Atlanta, 84% had fever and 44% had night sweats at diagnosis 2 groups, this study prospectively confirmed the association [15]. In a compilation of symptoms reported from 10 reported between disseminated M. avium complex disease and weight studies, Benson and Ellner [4] found that 87% of persons with loss, abdominal pain, fever, night sweats, anemia, a low Karnofsky disseminated M. avium complex disease had fever and 78% score, and an increase in level of serum alkaline phosdisseminated had night sweats [4]. The fact that fewer patients in our cohort phatase. Such associations have previously been reported only had fever and night sweats suggests that patients in our study from cross-sectional and cohort studies and are assumed to

5 130 Gordin et al. JID 1997;176 (July) Figure 1. Probability of survival following diagnosis of M. avium complex (MAC) bacteremia. Patients with MAC infection (dashed line) had shorter survival time (median, 264 days) than controls (solid line; median, 591 days) (P õ.01). soon becomes persistent, and the number of mycobacteria in the blood increases over time [17 19]. Thus, M. avium com- plex disease in the absence of bacteremia or associated with bacteremia of low titer is associated with early disease, while multiorgan system involvement and high-titer bacteremia are associated with long-standing disease. Such long-standing dis- ease is associated with greater morbidity, poorer response to antimycobacterial agents, and shorter survival [15]. To examine these time-dependent changes, we compared the clinical and laboratory characteristics of the patients with disseminated M. avium complex disease with those of the control cohort at each month, from 5 months before to the initial diagnosis of M. avium complex bacteremia (table 3). Using this technique, we were able to define the time at which significant differences occurred between the cases and controls for each individual characteristic. Patients with disseminated M. avium complex disease had a clinical prodrome that was identifiable 3 months before the first positive blood culture. Such patients had a significant loss of weight compared with the controls at this point, which con- tinued to the time of dissemination. One to 2 months prior to the first detection of M. avium complex in the blood, patients represent manifestations of disseminated M. avium complex disease because they have not been reported to be predictors of disease. Moreover, those studies did not differentiate the symptoms of M. avium complex infection from those of other HIV-associated illnesses and may have combined persons with early- and late-stage M. avium complex disease. The present study, where patient blood was cultured monthly, identified disseminated M. avium complex disease and its associated abnormalities at the onset of dissemination. An elevated serum LDH level was also associated with disseminated M. avium complex disease. Although not previously reported in association with M. avium complex disease, LDH level has been noted to be a harbinger of P. carinii pneumonia in patients with AIDS [16]. The clinical relevance of LDH as a marker for M. avium complex disease remains to be elucidated. The pathophysiology of M. avium complex disease has not been well studied. It appears that this disease is acquired from the environment through the gastrointestinal tract or, less commonly, the lung [4, 5]. Foci of localized infection in the gastrointestinal or respiratory tract eventually produce bacteremia, which then spreads disease to internal organs, such as the spleen and bone marrow. Bacteremia may be transient at first, but

6 JID 1997;176 (July) Disseminated M. avium Infection 131 reported more fever, and lower hemoglobin and higher LDH M. avium complex disease and should receive prophylactic levels could be detected. Many of the other abnormalities associated antibiotics. Second, because such patients have fever and with disseminated M. avium complex disease, including weight loss from M. avium complex infection prior to the occur- a lower Karnofsky score, abdominal pain, night sweats, diarrhea, rence of bacteremia, examination of tissues for evidence of M. and an elevated level of alkaline phosphatase could be avium complex may be warranted when blood cultures are detected only at the time of dissemination. Thus, it appears that negative; biopsy of the gut or liver may be useful, but stool most of the manifestations of disseminated M. avium complex culture is not [25]. Third, when M. avium complex infection disease appear closely before or at the time of initial M. avium is suspected, blood should be cultured monthly to detect myco- complex bacteremia. However, the occurrence of fevers and bacteremia at the earliest possible time because early detection weight loss 2 3 months before the first positive blood culture of M. avium complex disease is associated with fewer symp- suggests that localized disease is present well before bacter- toms and longer survival. Fourth, because fever and weight emia. This is consistent with animal models of disseminated loss from other causes are common in persons with advanced M. avium complex disease [20, 21] and observations at autopsy AIDS, we do not recommend presumptive treatment for M. that M. avium complex is found in intestines, liver, and spleen avium complex disease. Improved methods for early diagnosis in most patients who die of disseminated M. avium complex of M. avium complex disease are needed. disease. An autopsy study failed to identify sites of localized disease in some patients, but this failure may have been due to a treatment effect with antimycobacterial agents [17]. Acknowledgments Early in the AIDS epidemic, some believed M. avium com- We thank Dean Margeson for his help with the statistical analyplex to be predominately a marker of severe immunosuppres- sis and Cybil Richardson for manuscript preparation. sion, rather than a disease with substantial attributable mortality [22]. In part, this belief was founded on the observation that survival time of those with and without M. avium complex References infection was quite short. We found survival to be approxi- 1. Hawkins CC, Gold JWM, Whimbey E, et al. Mycobacterium avium complex mately half as long in patients with disseminated M. avium infections in patients with the acquired immunodeficiency synmately drome. Ann Intern Med 1986;105: complex (median, 264 days) as in controls (median, 592 days). 2. Horsburgh CR Jr, Selik RM. The epidemiology of disseminated nontuber- Because many of our patients with M. avium complex infection culous mycobacterial infection in the acquired immunodeficiency synwere treated with antimycobacterial agents, the true survival drome (AIDS). Am Rev Respir Dis 1989;139:4 7. of patients with untreated M. avium complex infections may 3. Ellner JJ, Goldberger MJ, Parenti DM. Mycobacterium avium infection in fact be shorter, as treatment of this disease has been shown and AIDS: a therapeutic dilemma in rapid evolution. J Infect Dis 1991; 163: to lengthen survival [14, 23, 24]. The survival of our patients 4. Benson CA, Ellner JJ. Mycobacterium avium complex infection and AIDS: with M. avium complex disease was substantially longer than advances in theory and practice. Clin Infect Dis 1993;17:7 20. that of largely untreated patients in a prior observational study 5. Horsburgh CR Jr. Mycobacterium avium complex infection in the acquired [7] (median survival, 134 days); such longer survival is likely immunodeficiency syndrome. N Engl J Med 1991;324: due to both earlier detection and antimicrobial therapy, as well 6. Havlik JA Jr, Horsburgh CR Jr, Metchock B, Williams PP, Fann SA, Thompson SE III. Disseminated Mycobacterium avium complex infecas other improvements in HIV-related care. tion: clinical identification and epidemiologic trends. J Infect Dis 1992; This large, multisite study demonstrates that disseminated M. 165: avium complex disease occurs in a high percentage of patients 7. Nightingale SD, Byrd LT, Southern PM, Jockusch JD, Cal SX, Wynne with severe immunosuppression who do not receive prophylaxis. BA. Incidence of Mycobacterium avium intracellulare complex bacter- There were no useful clinical predictors of dissemination other emia in human immunodeficiency virus positive patients. J Infect Dis 1992;165: than a low CD4 lymphocyte count. Prior history of an opportunis- 8. Chin DP, Hopewell PC, Yajko DM, et al. Mycobacterium avium complex tic infection is also an important predictor [7], but we were unable in the respiratory or gastrointestinal tract and the risk of M. avium to examine this factor because all patients in our study had a prior complex bacteremia in patients with human immunodeficiency virus AIDS-defining illness. Many of the signs and symptoms that have infection. J Infect Dis 1994;169: been associated with M. avium complex infection are present in 9. Hoover DR, Graham NMH, Bacellar H, et al. An epidemiologic analysis of Mycobacterium avium complex diseases in homosexual men infected patients with equally advanced HIV disease without M. avium with human immunodeficiency virus type 1. Clin Infect Dis 1995;20: complex infection. Weight loss, anemia, and fever are, however, harbingers of M. avium complex bacteremia. Night sweats, diar- 10. Chaisson RE, Moore RD, Richman DD, Keruly J, Creagh T, the Zidovudine rhea, abdominal pain, and elevated levels of alkaline phosphatase, Epidemiology Study Group. Incidence and natural history of Mycorhea, which are frequently associated with disseminated M. avium comimmunodeficiency bacterium avium complex infections in patients with advanced human virus disease treated with zidovudine. Am Rev Re- plex in the literature, occur coincident with the detection of mycospir Dis 1992;146: bacteria in the blood. 11. Nightingale, SD, Cameron DW, Gordin FM, et al. Two controlled trials These results have several important implications for patients of rifabutin prophylaxis against Mycobacterium avium complex infection with advanced AIDS. First, such patients have a high risk for in AIDS. N Engl J Med 1993;329:

7 132 Gordin et al. JID 1997;176 (July) 12. Revision of the CDC Surveillance case definition for acquired immunode- 20. Brown ST, Edwards FF, Bernard EM, Niki Y, Armstrong D. Progressive ficiency syndrome. MMWR Morb Mortal Wkly Rep 1987;36:suppl 1: disseminated infection with Mycobacterium avium complex after intra- 1S 15S. venous and oral challenge in cyclosporine-treated rats. J Infect Dis 1991; 13. Salfinger M, Stool EW, Piot D, Heifets L. Comparison of three methods 164: for recovery of Mycobacterium avium complex from blood specimens. 21. Bermudez LE, Petrofsky M, Kolonoski P, Young L. An animal model of J Clin Microbiol 1988;26: Mycobacterium avium complex disseminated infection after coloniza- 14. Chin DP, Reingold AL, Stone EN, et al. The impact of Mycobacterium tion of the intestinal tract. J Infect Dis 1992;165:75 9. avium complex bacteremia and its treatment on survival of aids pa- 22. Chaisson RE, Hopewell PC. Mycobacteria and AIDS mortality. Am Rev tients a prospective study. J Infect Dis 1994;170: Respir Dis 1989;139: Horsburgh CR Jr, Metchock B, Gordon SM, Havlik JA Jr, McGowan JE 23. Horsburgh CR Jr, Havlik JA, Ellis DA, et al. Survival of patients with Jr, Thompson SE III. Predictors of survival in patients with aids and acquired immune deficiency syndrome and disseminated Mycobactedisseminated Mycobacterium avium complex disease. J Infect Dis 1994; rium avium complex infection with and without antimycobacterial che- 170: motherapy. Am Rev Respir Dis 1991;144: Zaman MK, White DA. Serum lactate dehydrogenase levels and Pneumo- 24. Shafran SD, Singer J, Phillips P, Canadian MAC Study Group. The Canacystis carinii pneumonia. Am Rev Respir Dis 1988;137: dian randomized open-label trial of combination therapy for MAC bac- 17. Torriani FJ, McCutchan JA, Bozzette SA, Grafe MR, Havlir DV. Autopsy teremia: final results [abstract LB-20]. In: Program and abstracts of the findings in AIDS patients with Mycobacterium avium complex bacteremia. J Infect Dis 1994;170: th Interscience Conference on Antimicrobial Agents and Chemother- 18. Kemper CA, Havlir D, Bartok AE, et al. Transient bacteremia due to Mycobacteogy, apy (San Francisco). Washington, DC: American Society for Microbiolrium avium complex in patients with AIDS. J Infect Dis 1994;170: Torriani FJ, Behling CA, McCutchan JA, Haubrich RH, Havlir DV. Disnation 25. Yajko DM, Nassos PS, Sanders CA, et al. Comparison of four decontamiseminated Mycobacterium avium complex: correlation between blood methods for recovery of Mycobacterium avium complex from and tissue burden. J Infect Dis 1996;173: stools. J Clin Microbiol 1993;31:302 6.