INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The human immunodeficiency virus (HIV) epidemic has changed the epidemiology of nontuberculous mycobacterial (NTM) disease. Patients with advanced HIV infection living in the United States and Europe have a high incidence of Mycobacterium avium complex disease (1) and a smaller increase in the risk of disease caused by other NTM (2). HIV infection also affects the clinical presentation and treatment outcome of NTM disease. For HIV-associated M. avium complex the typical picture is a disseminated infection in a patient with far advanced immunosuppression and little or no focal pulmonary disease (1). The clinical and radiological features of HIV-associated NTM disease caused by other mycobacterial species are more heterogeneous and include a higher proportion of patients with apparently limited pulmonary disease (3). Less is known about the effect of HIV on NTM disease in developing countries. Disseminated disease due to M. avium complex does occur in HIV-positive Africans, but is identified in life (12, 13) and at autopsy examination (14) less frequently than in the United States or Europe.

The aims of this retrospective study were to investigate the impact of HIV infection on the clinical and radiological characteristics of Mycobacterium kansasii disease in South African gold miners. Miners have a high prevalence of chronic chest disease (15), and pulmonary NTM disease was common before the HIV epidemic (16). The HIV epidemic has only recently reached South Africa, but is spreading rapidly (17). HIV infection is a risk factor for pulmonary NTM disease in miners (15), and HIV prevalence among miners treated for NTM disease has risen from 4% in 1993 to 35% in 1997 (E. L. Corbett, unpublished observation).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Identification and Microbiology

HIV-tested male mineworkers from whom a sputum specimen had grown NTM between January 1, 1993 and June 30, 1996 were identified from the tuberculosis (TB) clinic database of the Ernest Oppenheimer Hospital, Welkom, South Africa. This hospital is the sole source of tertiary care for miners working for a single large mining company.

Miners presenting either with symptoms suggestive of mycobacterial disease or who were noted to have new radiological changes at their annual medical examination were routinely screened for mycobacterial disease with sputum microscopy and culture. Concentrated sputum samples were examined for mycobacteria using fluorescent microscopy. Positive slides were confirmed with Ziehl-Neelsen staining. Initial culture of all sputum samples was onto Lowenstein-Jensen (LJ) slopes. The number of colonies on positive LJ slopes was recorded as: scanty (1 to 19); 1+ (20 to 99); 2+ (100 to 200); 3+ (more than 200). Positive LJ slopes were sent to the South African Institute of Medical Research (SAIMR), Johannesburg, for species identification and drug sensitivity testing. The hospital policy was to send only a single LJ slope for identification per disease episode, unless further positive cultures were obtained more than 6 mo later. The SAIMR laboratory used standard biochemical tests to distinguish different NTM species. All miners found to have a positive mycobacterial isolate were referred to the TB clinic and recorded on the database regardless of the species of the isolate. HIV testing has been offered to TB and NTM patients from 1993, with pre- and post-test counseling. CD4⁺ lymphocyte measurement (CD4 count) was carried out in some but not all HIV-positive miners with mycobacterial disease.

Ethical approval for the study was obtained from the ethics committees of the Ernest Oppenheimer Hospital and the London School of Hygiene and Tropical Medicine.

Radiology

A total of 108 HIV-negative and 36 HIV-positive M. kansasii patients was selected for radiological review of both presenting and premorbid radiographs. Patients with relapsed disease and those who had been treated for TB within the previous 12 mo were not considered for inclusion because of the unavailability of a stable premorbid radiograph in such cases. All HIV-positive patients who did not have either of these exclusion criteria were included. The HIV-negative patients were selected by including the next three HIV-negative patients presenting after each HIV-positive patient.

For each patient a premorbid radiograph taken at least 12 mo before diagnosis was identified either from the annual screening mini-chest radiographs or from standard-sized radiographs when available. Serial radiographs were screened and only radiographs that had been static for 12 mo were included as premorbid radiographs, in order to avoid including radiographs with active slowly progressive mycobacterial disease. Where this was suspected an earlier radiograph was used instead. Presenting radiographs, taken within 2 wk of mycobacterial disease diagnosis, were read together with the premorbid radiograph by two independent readers using a standardized scoring system that included an assessment of the presenting radiograph, the premorbid radiograph, and new changes apparent from the differences between the two radiographs. Thin-walled cavities were defined as cavities where the thickness of part or all of the cavity wall was 1 mm or less. Abnormalities were defined as being either upper or lower field by their position relative to the ipsilateral hilum. Each lung was divided into three zones of equal vertical height at the midclavicular line. The number of zones that contained an abnormality was summed as a zone score. Decisions were reached by consensus in cases of initial disagreement.

Additional Risk Factors for NTM Disease

The job at the time of diagnosis was ascertained from company records. Jobs based at the rock face or involving the underground transport of ore were classed as dusty. Records of previous TB treatment were obtained from the TB clinic and hospital case notes.

Treatment Outcomes

The duration of treatment and regimen used were at the discretion of the attending physicians. The most common regimen used was a combination of rifampicin, ethionamide, and ethambutol. Patients no longer taking treatment were classified into the following groups: cureda negative sputum culture within 3 mo of stopping treatment; completedcompleted treatment but no sputum cultures taken within 3 mo of stopping treatment; failurepositive sputum culture within 3 mo of stopping treatment; defaultedlost to follow-up before completion of treatment; transferred outleft the mines before completion; died on treatmentdied of any cause before completion.

Statistical Analysis

Data were analyzed using STATA 5.0 (Stata Corporation, College Station, TX) and EPI-INFO 6.02 (Centers for Disease Control, Atlanta, GA) statistical software. Odds ratios (OR) and exact 95% confidence intervals (95% CI) were used to compare radiological features between patients grouped according to HIV status. A chi-squared test for trend was used to compare the number of NTM risk factors per patient and the extent of radiological disease in HIV-positive and HIV-negative men. In order to avoid including cases of nonpathogenic infection with a test for disease severity, patients who had no new radiological changes were excluded from the latter analysis.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

There were 406 miners from whom a single fully identified NTM had been cultured during the specified time period. Of these, 366 (90%) had been tested for HIV, and 69 (19%) of those tested for HIV were positive. The NTM species distribution did not vary significantly according to HIV status (p = 0.34): the most frequent species was M. kansasii, isolated from 202 (68%) HIV-negative and 43 (62%) HIV-positive men, followed by Mycobacterium scrofulaceum, which was isolated from 41 (14%) HIV-negative and eight (12%) HIV-positive men. M. avium complex was isolated from 18 (6%) and four (6%) HIV-negative and HIV-positive men, respectively, and there were 36 (12%) and 14 (20%) other NTM isolates of species not normally considered pathogenic isolated from HIV-negative and HIV-positive men, respectively.

The patient characteristics, sputum smear results, and nonradiological NTM risk factors for HIV-positive and HIV-negative men with M. kansasii isolates are shown in Table 1. All men with M. kansasii isolates, apart from one HIV-negative man, were treated for M. kansasii disease. There was a lower proportion of smear-positive patients in the HIV-positive group, but this was not significant (OR, 0.60; 95% CI, 0.27 to 1.41). The proportion of patients with a past history of TB treatment was nonsignificantly higher in HIV-negative than in HIV-positive men, and the majority of patients in both groups were working in dusty jobs at the time of their diagnosis.

CD4 counts are not routinely requested on HIV-positive NTM patients, but 14 of the HIV-positive M. kansasii patients (33%) had a CD4 count estimated within 6 mo of diagnosis. The mean CD4 count was 490 × 10⁶/L (SD, 388) with a range of 1,431 to 73 × 10⁶/L. Only two patients had a CD4 count below 100 × 10⁶/L.

Treatment outcomes for patients who were no longer on treatment by January 1997 are shown in Table 2. The number of patients dying on treatment was small. This outcome was more common in the HIV-positive group, but not significantly so (OR, 5.9; 95% CI, 0.74 to 45.3). Drug choice and treatment duration did not differ by HIV status (median duration, 17 mo for HIV-positive and HIV-negative patients). A higher proportion of the HIV-positive patients were still on treatment at the time of review, but this was because more were diagnosed in the later years of the study period.

Either the presenting or a suitable premorbid radiograph could not be found for five (13%) of the HIV-positive and 16 (15%) of the HIV-negative men selected for radiological review. The radiological changes associated with M. kansasii isolates are shown for both HIV-positive and HIV-negative men in Table 3. Significant differences between the two HIV groups were that the HIV-positive men were more likely to have a normal premorbid radiograph (OR, 2.62; 95% CI, 1.01 to 6.67) and less likely to have focal scarring suggestive of previous healed TB (OR, 0.27; 95% CI, 0.10 to 0.68). New cavitation was significantly less common (OR, 0.34; 95% CI, 0.13 to 0.88) and new hilar adenopathy was significantly more common (OR, 5.07; 95% CI, 1.24 to 21.9) in the HIV-positive group. Men with new cavities were more commonly smear-positive than men without new cavities in both HIV groups (OR, 3.70 for HIV-negative men and 4.17 for HIV-positive men). The association between new cavities and smear positivity was significant for the HIV-negative men (CI, 1.04 to 12.6) but not for the smaller HIV-positive group (CI, 0.52 to 50.1).

The number of identifiable NTM risk factors per patient, other than HIV infection, were calculated for the M. kansasii patients who were included in the radiological study. The four risk factors considered were: dusty job, past history of TB treatment, silicosis, and premorbid focal radiological scarring. These data are shown for men grouped according to HIV status in Table 4. There were fewer risk factors per man for HIV-positive than for HIV-negative patients (² test for trend, p = 0.01), but only 10% of HIV-positive men had no other risk factor for NTM disease.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Most of the patients included in this study were treated for M. kansasii disease on the basis of suggestive clinical and radiological features in association with a single sputum NTM isolate. The diagnosis of NTM disease in individual cases cannot be made with confidence on the basis of a single NTM isolate from nonsterile sites because of the alternative possibilities of nonpathogenic colonization and specimen contamination. However, it is likely that most of the isolates from patients in this study did represent NTM disease as the majority from both HIV groups were associated with other features of active mycobacterial disease and M. kansasii sputum isolates have previously been shown to have a high likelihood of representing NTM disease in both HIV-negative and HIV-positive individuals (4, 18).

The HIV epidemic in the United States and Europe has been accompanied by both an increasing incidence and a changing spectrum of NTM disease. Before the HIV epidemic, NTM disease occurred with a low incidence, mainly in older individuals with underlying chronic lung disease. The predominant disease-associated NTM organism varied from one geographical location to another and the typical presentation of NTM disease in adults was as a localized pulmonary infection that resembled tuberculosis radiologically. This contrasts with HIV-associated NTM disease in the United States and Europe, which occurs with a high incidence and where the majority of patients present with a disseminated M. avium infection with no pulmonary focus and no predisposing factors apart from advanced HIV infection (1). HIV-associated M. kansasii disease is less common, but still occurs with an increased incidence, mainly in patients with advanced immunosuppression. Previous reports of the radiological manifestations of HIV-associated M. kansasii have identified focal pulmonary disease in most patients, with apparently limited pulmonary disease in a sizeable minority, but have found cavitation to be atypical and present in the minority (4). The prognosis after a diagnosis of disseminated NTM disease has been poor, mainly reflecting the advanced degree of immunosuppression although NTM disease itself contributes to the high mortality (4, 19). Before the availability of highly active antiretroviral regimens, the median survival time after a diagnosis of disseminated M. avium disease was in the order of 6 mo (19) and was similar for M. kansasii disease (4). NTM disease, with a similar clinical picture and predominance of M. avium complex, has also been identified in patients with acquired immunodeficiency syndrome (AIDS) from several African countries, but with a much lower prevalence (12).

The effect of HIV infection on NTM disease in the South African miners described here is different from that found elsewhere. There was no significant effect of HIV infection on the bacteriological spectrum of NTM disease, with the predominant isolates being M. kansasii followed by M. scrofulaceum in both HIV status groups. The majority of HIV-positive M. kansasii patients presented with typical upper field cavitary disease and 72% of the HIV-positive men not still on treatment were known to have survived the full duration of their treatment course. The available CD4 counts were considerably higher than those reported from HIV-positive M. kansasii patients in the United States (2, 4), suggesting that M. kansasii disease is occurring earlier in the course of HIV infection. This is in keeping with the low mortality during treatment in the HIV-positive group.

There were, however, differences in the manifestations of M. kansasii disease between the two HIV status groups. HIV-positive men were less likely to have new cavitation. There was a higher proportion of patients with hilar adenopathy and a less clear-cut upper field predominance in the radiographs from HIV-positive compared with HIV-negative M. kansasii patients. There was also a higher mortality during treatment in the HIV-positive patients, although numbers were small and the difference was not significant.

The clinical and radiological appearances of HIV-associated M. kansasii disease at this location are reminiscent of patients with early HIV infection and Mycobacterium tuberculosis disease (20, 21). The impact of HIV infection on the radiological presentation of M. tuberculosis disease becomes more marked with increasing immunosuppression (20, 21). The limited impact of HIV infection found in this study, compared with previous reports of a marked effect in patients with more advanced HIV disease, supports a similar relationship between the degree of HIV-associated immunosuppression and the radiological presentation of M. kansasii disease.

The question as to why HIV-associated M. kansasii disease is occurring at such an early stage of HIV infection in miners may be best answered by considering other risk factors for NTM disease. NTM disease was common in this workforce before the HIV epidemic (16), probably attributable to the high prevalence of other risk factors such as a dusty job, silicosis, and post-tuberculous lung disease, each of which has been shown to be significantly associated with NTM disease in miners (15). HIV infection is also a significant risk factor for NTM disease in miners, but is not currently the predominant one (15). Of note, only a small minority of the HIV-positive men described here had no additional NTM risk factor identified, although they did tend to have fewer other risk factors than the HIV-negative men.

Taken together, the results from this study suggest that the effects of different NTM risk factors in this population may be combining and so allowing HIV infection to result in M. kansasii disease that occurs at a considerably lesser degree of immunosuppression and more closely resembles disease in HIV-negative men than is the case with HIV-associated NTM disease in other settings. NTM disease incidence, morbidity, and mortality are likely to increase further among miners as the HIV epidemic progresses. Clinicians should be aware that HIV-infected individuals who also have other NTM risk factors may be at an increased risk of developing pulmonary NTM disease early in the course of their HIV disease. The good short-term prognosis found in this study for pulmonary NTM disease associated with early HIV infection may be relevant for similar patients elsewhere.