Mycobacterium haemophilum is fastidious, grows slowly, and requires a lower incubation temperature for growth than most other mycobacteria. Isolation of M. haemophilum also requires iron-supplemented medium. Because of these requirements, M. haemophilum would not be isolated using standard isolation techniques for tubercle bacilli. Failure to recognize these microbiologic considerations probably has led to under-reporting of M. haemophilum infections.

Little is known about the spectrum of illness caused by M. haemophilum infection and its response to therapy. In addition, the ecology of the organism is poorly understood and the reservoir and modes of transmission must still be elucidated. After reports of several cases of M. haemophilum infection in New York City in early 1991, we initiated an investigation to characterize further this infection. We describe the clinical and epidemiologic features of 13 persons identified in the investigation and compare the findings in these patients with those reported in other cases.

Methods

Top Methods Results Discussion Author & Article Info References

Case Definition

We defined a case as isolation of M. haemophilum from any resident of the New York City metropolitan area between 1 January 1989 and 1 September 1991. When possible, isolates were confirmed as M. haemophilum at the Centers for Disease Control and Prevention by growth characteristics and determination of typical mycolic acid patterns using high-performance liquid chromatography [21]. We did drug susceptibility tests using a modified proportional method [22], in which 7H10 agar was supplemented with hemin (39 µg/L). Inoculated plates were incubated at 32 °C for 21 days.

Case Finding

In April 1990, we asked infectious disease consultants in the New York metropolitan area to notify us about any suspected or confirmed infections resulting from M. haemophilum. We did more intensive case finding at 12 New York hospitals serving many immunocompromised persons. We contacted infectious disease physicians and clinical microbiologists at these hospitals and asked about any patients identified with M. haemophilum infection or any isolates identified as M. haemophilum since January 1989. We asked to review laboratory records for isolation of organisms identified as M. haemophilum and reports of acid-fast bacillus smear-positive, culture-negative specimens obtained from cutaneous or soft tissue lesions.

For all identified cases, information was collected on patient demographics; underlying illness; indicators of immune status; immunosuppressive and respiratory therapy; hospitalization; recreational drug, alcohol, and tobacco use; exposure to water from various sources; exposure to animals and soil; travel; and dietary habits. Data were also collected on each patient's course of illness caused by M. haemophilum, concurrent illnesses, culture results, type of therapy, and clinical response to therapy. Data sources included medical records, physician history, and, when possible, patient interviews.

Case-Control Study

A case–control study was done to examine risk factors for M. haemophilum infection and potential modes of transmission. Three control patients were selected for each of the nine patients who were available for inclusion in this study. Control patients were matched to case patients by age (±5 years), underlying reason for immunosuppression (AIDS or bone marrow transplant), and type of medical provider (private or public). Potential controls were identified among the patients at two clinics providing care to patients with M. haemophilum infection. Comparisons between case patients and control patients were based on medical records and interviews that focused on possible hospital and geographic exposure opportunities. Case patients were questioned about exposure during the 3 months before the onset of M. haemophilum-related symptoms, and matched control patients were asked about exposure during the same time period.

Data Analysis

Data were recorded in EPI-INFO (version 5.01, Centers for Disease Control and Prevention, Atlanta, Georgia). Analyses were done using EPI-INFO and the PECAN conditional logistic program (Statistical Analysis Systems 5.18, SAS Institute, Cary, North Carolina). Previous studies were identified by MEDLINE search followed by manual bibliographic review of retrieved articles.

Results

Top Methods Results Discussion Author & Article Info References

Thirteen patients at seven New York-area hospitals were found to be infected with M. haemophilum. Six of these patients have now been described elsewhere [23, 24]. Thirteen strains from 12 of the patients were confirmed as M. haemophilum by high-performance liquid chromatography at the Centers for Disease Control and Prevention.

The earliest onset of symptoms among the 13 patients occurred in August 1989. Four cases were diagnosed in 1990, and nine in 1991. Nine patients (69%) were men. Twelve were white, 1 was black, and 1 was Hispanic. The mean age of patients at the time of diagnosis was 34 years (range, 27 to 51 years). Eleven patients had AIDS with risk factors including homosexuality (6 of 8 men) and heterosexual exposure to high-risk persons (all 3 women). The exposure risk could not be determined in the other 2 men with AIDS. Among these 11 patients, the mean duration of AIDS before onset of symptoms attributable to M. haemophilum infection was 15.9 months (range, 0 to 43.4 months). The other 2 patients received bone marrow transplant, 1 for aplastic anemia and the other for acute myelocytic leukemia. Infection with M. haemophilum developed in the patient with aplastic anemia 6 months after his second bone marrow transplant, whereas infection occurred 4 months after bone marrow transplant in the other patient.

Cutaneous ulcerating lesions Figure 1 and Figure 2 were the initial presentation of disease due to M. haemophilum in 8 patients. Most commonly, multiple lesions were present. Lesions ultimately developed in 12 patients (Table 1). Cutaneous lesions were found most frequently on the extremities, often overlying joints, and less commonly on the trunk and face. The distribution of lesions did not follow lymphatic drainage (it was not sporotrichoid in appearance). They were typically raised, violaceous, and fluctuant and ranged in diameter from 0.5 to 2 cm. They then became enlarged and pustular, with an erythematous zone at the periphery and a light central area that eventually became necrotic and ruptured to drain small quantities of serosanguinous material. Mature lesions were often extremely painful. Alternative diagnoses considered included Kaposi sarcoma, cutaneous Mycobacterium avium complex and Mycobacterium marinum infections, and nonspecific bacterial infections.

View larger version (15K): [in this window] [in a new window]   Figure 1. Dates of symptom onset, definitive diagnostic sample collection, and diagnosis of Mycobacterium haemophilum infection in patients in New York City.

View larger version (133K): [in this window] [in a new window]   Figure 2. Disseminated Mycobacterium haemophilum lesions in a 37-year-old man with AIDS.

Respiratory complaints were the presenting symptoms in 3 patients, whereas 2 presented with arthritis (see Table 1). Respiratory symptoms attributed to M. haemophilum developed in 3 other case patients, whereas joint symptoms, including tenderness and swelling, eventually developed in most patients [9 of 13]. Six patients reported night sweats (Table 1).

Once the diagnosis was established, M. haemophilum was frequently isolated from multiple sites. Overall, the organism was isolated from cutaneous lesions (9 patients), bone (6 patients), sputum (4 patients), synovial fluid (3 patients), blood (1 patient), and from lung biopsy specimens (1 patient) (Table 2).

Diagnosis

Among the 13 patients, the time from initiation of laboratory evaluation to definitive identification of M. haemophilum infection ranged from 14 to 495 days (mean, 124 days) (Figure 3). During the study period, the interval improved, decreasing from a mean of 314 days for the first two cases to 44 days for the last two cases.

View larger version (120K): [in this window] [in a new window]   Figure 3. Mycobacterium haemophilum lesions resolving on the arm of the patient in Figure 2 .

Risk Factor Studies

Case patients had greater immunosuppression than did control patients. Of the 8 case patients for whom CD4 lymphocyte counts were available, 7 (88%) had counts less than 25 cells/mm³ compared with 5 (25%) of 20 control patients (matched odds ratio, 9.5; 95% CI, 1.1 to 81.7; P = 0.04).

None of the patients reported knowledge of other persons with M. haemophilum infection or were aware of patients with clinical syndromes similar to their own. The patients' residences and work places were geographically dispersed throughout the New York metropolitan area. Investigation of possible nosocomial transmission at one hospital that had reported four of the patients showed no common clinic visit dates or providers, but two bone marrow transplant recipients had been hospitalized on the same ward during a single 24-hour period in rooms at opposite ends of the ward that were served by independent ventilation systems. One of the patients was interviewed and denied contact with the other patient, who was in isolation at the time. The two patients had no common nursing staff or physicians.

Case patients did not differ statistically from control patients with respect to the presence of underlying diseases (including those with cutaneous or pulmonary manifestations) or previous use of medications, including antimycobacterial therapy and prophylactic therapy for opportunistic infections. Two (22%) of the case patients were receiving zidovudine, compared with 44% of controls, but the difference was not significant (P > 0.05). Case patients and control patients were similar with regard to reported alcohol and substance abuse and contact with animals, soil, and water sources. The proportion of case patients and control patients reporting a cutaneous injury or dermatologic condition that preceded the onset of M. haemophilum infection was similar (78% compared with 85%).

Clinical Course

Follow-up information was collected on the patients for a 1-year period after the investigation. Nine of the patients died. In one of the bone marrow transplant recipients, cavitary pulmonary disease due to M. haemophilum infection was considered the primary cause of death. Mycobacterium haemophilum infection may also have contributed to the death of a second patient, a 37-year-old man with AIDS in whom disseminated disease was diagnosed 2 months before his death. The patient had a lobar infiltrate and pleural effusion, but no autopsy was done. Patient 4 had presented with multiple cutaneous lesions in August 1990. Mycobacterium avium complex was isolated from his blood and bronchoalveolar lavage fluid, and M. haemophilum was subsequently isolated from multiple cutaneous lesions. The lesions regressed during treatment with doxycycline, rifampin, and amikacin; however, 10 months later hepatotoxicity developed, which required discontinuation of therapy. The lesions recurred in June 1991, and he showed improvement during therapy with ciprofloxacin, amikacin, and rifampin. In August 1991, he died after several weeks of increasing respiratory failure, during which M. haemophilum was isolated repeatedly from his sputum. Although death was attributed to lymphoma because foci were identified in lungs and esophagus, M. haemophilum was isolated from an adrenal gland, a lymph node, and the spleen. The deaths of the other six patients appeared to be unrelated to their M. haemophilum infection, which had improved or resolved in five. Of four patients surviving after the end of 1 year, M. haemophilum infection resolved in three and persisted in one.

Therapy

Once M. haemophilum infection was diagnosed, various treatment regimens were used, including combinations of isoniazid, rifampin, ethambutol, minocycline, doxycycline, clarithromycin, ciprofloxacin, amikacin, clofazimine, streptomycin, and pyrazinamide. Clinical responses to therapy are shown in Table 3. Drug susceptibilities were determined for nine isolates of M. haemophilum (Table 4). All isolates were susceptible to ciprofloxacin, cycloserine, and rifabutin. Complete resistance to ethambutol and pyrazinamide was reported. Although susceptibility was concentration dependent for isoniazid and streptomycin, fewer than 25% of isolates were susceptible to the highest drug concentration tested. Isolates from four patients were tested for susceptibility to clarithromycin, and all isolates had mean inhibition concentrations 0.5 µg/mL.

Discussion

Top Methods Results Discussion Author & Article Info References

Our findings in New York City indicate that disease due to M. haemophilum occurs more frequently than the medical literature suggests. Thirteen apparently sporadic cases of disease were identified during a 20-month period, raising by almost one third the total number of reported cases. Because the strict growth conditions required for isolation are not widely applied, the actual number of M. haemophilum infections in New York City during the investigation period was probably much higher.

The findings in this case series, combined with other published reports, provide the basis for separating patients with M. haemophilum infections into two broad categories. The principal risk group appears to be those who are severely immunocompromised and in whom M. haemophilum occurs as an opportunistic infection. The earliest reports documented infections in persons with either lymphoma or renal transplants [1, 4, 18]. Persons with AIDS have now become the largest group to be reported with this infection, and bone marrow transplant recipients can also be added to the list of persons at risk. Indeed, any condition resulting in marked suppression of cell-mediated immunity is likely to predispose patients to M. haemophilum infection, and this pathogen thus should be considered in the differential diagnosis of any skin lesion in patients with these underlying conditions.

Immunocompetent children compose the second population group in which M. haemophilum infection has been reported. Mycobacterium haemophilum infection produces cervical and perihilar lymphadenitis in children that is clinically similar to that produced by infection with M. avium complex, M. tuberculosis, and M. scrofulaceum. Mycobacterium haemophilum adenitis has been reported in children in Australia, Canada, and the United States [6, 15, 16, 20]. These children generally responded favorably to chemotherapy, surgical excision of lesions, or both.

Two recent reviews suggested that M. haemophilum infection may occur in immunocompetent adults [10, 25]. In 1974, a report described 29 immunocompetent patients living in the north central United States and in contiguous areas of Canada in whom a syndrome developed that included mycobacterial adenitis and typically solitary skin lesions [26]. These cases were documented during a period of several years and preceded the description of M. haemophilum in 1978. Clinical specimens were not cultured in media that would support growth of M. haemophilum, and no definitive diagnosis was ever established. No other suspected cases of M. haemophilum infection in immunocompetent adults have been reported since this case cluster.

In the New York cases, M. haemophilum infection was associated with a wide spectrum of disease among immunocompromised hosts. Tender cutaneous lesions were the most common manifestation of M. haemophilum infection, but the organism was also isolated from synovial fluid, blood, lymph nodes, bone, and sputum. Bone and lung tissue isolates represent the first reported instances of M. haemophilum infection in these tissues. In addition, the organism has been identified in skin, joint fluid, blood, sputum, lymph node, and wound specimens. The constellation of multiple diffuse, cutaneous lesions and isolation of the Mycobacterium organism from blood and other normally sterile sites suggests hematogenous spread of the organisms. This mode of dissemination is consistent with other mycobacterial infections in persons with AIDS, particularly infection with M. avium complex [27] and M. tuberculosis [28]. Our findings also indicate that among severely immunocompromised patients, M. haemophilum infection should be considered when assessing arthritis, osteomyelitis, and pneumonitis of uncertain cause.

Cutaneous lesions tend to cluster on the extremities and, as has been noted by Dever and colleagues [7], frequently overlie the joints. This distribution of lesions is consistent with the finding that lower temperature favors growth of M. haemophilum because skin temperature is cooler over joints than on the head and trunk [29, 30]. These lesions could be misdiagnosed as Kaposi sarcoma, which commonly presents as painless erythematous macular lesions located on the extremities [31]. Biopsy or acid-fast bacillus smear should allow differentiation between these two conditions. It may be more difficult to distinguish M. haemophilum from other mycobacteria that produce cutaneous manifestations in persons infected with human immunodeficiency virus (HIV). Cutaneous M. avium complex and M. tuberculosis have been reported rarely in persons with AIDS [32, 33]. Mycobacterium marinum infection has occurred with multiple sores and boils on the face, chest, and extremities in a man infected with HIV [34], and M. fortuitum complex has been reported to cause multiple subcutaneous necrotizing nodules in a person with HIV infection who was an injection drug user [33]. Because treatment regimens vary depending on the mycobacterial species, culture for the causative agent is required.

The causes of monarticular or oligoarticular arthralgia and swollen joints in persons with AIDS are numerous [35]. Because most of our patients had joint manifestations, M. haemophilum infection should be included in the differential diagnosis, especially if cutaneous lesions are present. Arthrocentesis is required to distinguish this organism from other infective causes of arthralgia, and the diagnosis should be considered in immunocompromised patients if acid-fast bacilli are found in joint fluid.

Three of the patients presented with upper respiratory complaints. The characteristic cutaneous lesions subsequently developed in only two of these patients. The differential diagnosis for respiratory complaints in immunocompromised hosts is extensive [36], and M. haemophilum infection is an unlikely cause. Nonetheless, M. haemophilum infection should be considered in patients with sputum that is smear positive for acid-fast bacilli when cultures are negative for M. tuberculosis and M. avium complex, particularly in the presence of unexplained joint or cutaneous signs.

The mode of transmission of M. haemophilum is unknown. The New York cases appeared to be sporadic because no common source was suggested by the case–control study. Infected patients were not more likely to have contact with other persons with M. haemophilum-like disease. These findings do not support person-to-person spread. Potential routes of transmission include percutaneous inoculation, inhalation, or ingestion. Because most infections were cutaneous, direct inoculation after environmental exposure is possible. However, case patients were no more likely than control patients to recall injuries or skin conditions before symptoms attributed to M. haemophilum infection developed, and previously described patients have not reported cutaneous trauma before illness. The isolation of M. haemophilum from sputum and lung tissue suggests the possibility of respiratory transmission. Although there was no association with previous respiratory infection, respiratory therapy, or exposure to respiratory irritants, this mode of transmission is also biologically plausible and can occur in other mycobacterial infections. It has been suggested that M. avium complex infection in AIDS is acquired by ingestion [32]; however, the dietary patterns of case patients and control patients in this study were similar. The inability to identify common risk factors for disease in this series may be related to limitations in statistical power imposed by the small sample size. Prospective studies on newly diagnosed cases of disease are indicated to further explore potential modes of transmission.

Nontuberculous mycobacteria are widespread in the environment, particularly in aquatic reservoirs. In one environmental survey, 321 water specimens from natural, treated, and animal-contact sources were collected; 6% to 67% contained mycobacteria. Slowly growing mycobacteria were predominant, and 5 of the 149 strains could not be identified [37]. Nontuberculous mycobacteria are also commonly found in soil. Wolinsky and Rynearson [38] collected 72 soil samples from four states and identified at least one mycobacterial species from 86% of the samples. Mycobacterium fortuitum (64%), scotochromogens (54%), and slowly growing species (42%) were most commonly identified. Neither of these surveys used culture methods suitable for the growth of M. haemophilum, and the organism has never been isolated from a nonhuman source. Slow growth, minimal biochemical activity, and fastidious growth requirements make environmental isolation of M. haemophilum particularly challenging. Development of more sensitive detection techniques, such as DNA amplification to identify M. haemophilum-specific genetic sequences, may be helpful. Environmental sampling may be appropriate only after epidemiologic investigations have suggested possible reservoirs.

Patients in this study with M. haemophilum infection were profoundly immunosuppressed, as shown by CD4 lymphocyte counts. This finding may in part account for the relatively small numbers of infections previously reported and the large increase of reported infections in this investigation. As the population of severely immunosuppressed persons has grown because of AIDS and the increasing application of organ transplant procedures, the number of persons at greatest risk for M. haemophilum infection has increased in parallel. Among those infected with HIV, greater use of zidovudine and improved treatment and prophylaxis of opportunistic infections have increased survival of severely immunosuppressed patients. The growing population of persons with CD4 lymphocyte counts less than 50 cells/mm³ has provided a host niche for pathogens such as M. avium complex and M. haemophilum.

The incidence of M. haemophilum infection is unknown. The relative rarity of its isolation from clinical specimens and the small number of reported cases may be attributable, in part, to its specific growth requirements. Among the New York City cases, M. haemophilum was isolated fortuitously when clinicians were searching for other pathogens. In other instances, clinical specimens that were acid-fast smear positive and culture negative prompted additional testing for M. haemophilum.

Mycobacterium haemophilum grows on chocolate agar, egg-based media containing 15 to 25 µg/mL of ferric ammonium citrate, agar-based media supplemented with 0.4% hemoglobin, 60 µmol/L hemin, or when X-factor (hemin) strip or disc is present [39, 40]. The temperature growth range is 25 to 35 °C, but the optimal incubation temperature is reported to be 32 °C [39]. As the number of immunocompromised hosts increases, it will be increasingly important that laboratories providing services for these patients be able to isolate and identify the organism.

Several factors complicate determination of optimal strategies to treat infection caused by M. haemophilum. Because only a few cases have been reported, different therapeutic modalities have not been thoroughly evaluated. Previous reports have described clinical response to various therapeutic regimens, including isoniazid, rifampin, and ethambutol [1]; trimethoprim-sulfamethoxazole [12]; minocycline [8]; erythromycin [8, 20]; rifampin and minocycline [4]; rifampin and p-aminosalicylic acid [10]; and rifampin and erythromycin [9]. Several of the present authors have recommended a regimen including rifampin, amikacin, and ciprofloxacin [24]. It is difficult to draw conclusions based on these reports because host immune function varied during therapy and may have affected clinical response.

There are no standardized susceptibility tests for M. haemophilum. Isolates in this study did not show a uniform response in susceptibility tests. In addition, the relation between in vitro drug susceptibilities and clinical response is not known for this organism. In vitro drug susceptibility results are useful in predicting the clinical response of tuberculosis, but their utility in other nontuberculous mycobacterial infections, particularly those caused by M. avium complex, is unclear.

It is difficult to recommend therapy on a strictly empiric basis. In this study, antimycobacterial therapy was not supervised for most patients, and compliance with prescribed regimens is consequently uncertain. The issue is yet more complicated, because most of the study patients received several drug regimens over time, making the response to a single regimen difficult to discern. Nevertheless, efforts to identify M. haemophilum infections efficiently are worthwhile because the infection appears to be treatable. Of nine previously reported M. haemophilum infections in persons with AIDS for whom data are available, lesions and symptoms regressed in two thirds of them after treatment. In persons immunosuppressed for other reasons, lesions responded to therapy in all 10 previously described patients, with regression occurring as immune function improved.

In summary, M. haemophilum can produce a severe and sometimes fatal multisystem infection in severely immunocompromised hosts. Infections due to M. haemophilum are likely to be more frequently recognized as the population of susceptible immunocompromised hosts increases and as appropriate laboratory procedures are adopted. Accurate diagnosis is important because infections frequently respond to therapy. Increased awareness of the clinical manifestations of M. haemophilum infection should help clinicians identify additional cases so that therapeutic approaches can be optimized, epidemiologic studies can be done, and prevention strategies can be developed.