Mycobacteria are uncommon causes of ocular disease, but proper recognition of the organism is essential if the disease is to be treated appropriately. There are more than 50 species. Mycobacterium tuberculosis, which causes tuberculosis (TB), and Mycobacterium leprae, which causes leprosy (Hansen disease), are for several reasons usually discussed separately from the remaining species, which are commonly referred to as “atypical,” “nontuberculous,” or “mycobacteria other than tubercle bacilli.” TB and leprosy have been recognized clinically for hundreds of years, whereas nontuberculous mycobacteria were considered saprophytic organisms until the middle of the 20th century. TB and leprosy are systemic diseases, with ocular involvement (most commonly uveitis) resulting from hematogenous dissemination to the eye. Nontuber-culous mycobacteria rarely cause intraocular disease, but they may cause keratitits and other external infections after trauma or surgery. TB and leprosy are spread by human-to-human contact, whereas nontuberculous my-cobacteria are ubiquitous and can be found in soil, water, and other sources. The resurgence of TB in the 1980s led to renewed interest in mycobacterial diseases in general. Recent advances in mycobacteriology have expanded our understanding of the clinical syndromes associated with these organisms.

This chapter reviews the epidemiology, pathogenesis, and animal models of mycobacterial eye diseases and summarizes the clinical feature, laboratory diagnosis, and treatment of these diverse diseases.

Mycobacteria are obligate, aerobic, gram-positive bacilli measuring 0.2 to 0.4 µm in diameter and 2 to 10 µm in length. They are nonmotile and non-spore-forming. There are more than 50 mycobacteria species in the genus Mycobacterium, family Mycobacteriacea, order Actinomy-cetales. Although they are generally killed by heat, these organisms are tolerant of drying and disinfecting agents such as chlorine, 2% formaldehyde, and gluteraldehyde.¹ Many of the species will survive the concentration of chlorine contained in tap water (0.05 to 0.2 µg per mL of free chlorine).²

M. leprae was discovered in 1873 by G. H. Armauer Hansen, and was the first bacterium to be identified as causing disease in man. M. tuberculosis was discovered in 1882 by Robert Koch who later received the Nobel prize for his work. Owing to the difficulty in culturing M. leprae in vitro, Koch was credited with the germ theory of disease rather than Hansen. Mycobacterium fortuitum and Mycobacterium chelonae were isolated in 1938 and 1953, respectively. M. fortuitum was first reported to cause ocular disease in 1965 when Turner and Stinson noted M. fortuitum keratitis following removal of a corneal foreign body.

Mycobacterial cell walls contain type A peptidoglycan, in which m-diaminopimelic acid and d-alanine are directly cross-linked. The peptidoglycan is directly linked to a polysaccharide composed of arabinose and galactose. The presence of arabinogalactan results in immunologic cross-reactivity between Mycobacteria, Nocardia, and Corynebacterium.

Mycobacteria are characterized by large amounts of lipid and true waxes in their cell walls. These lipids and waxes account for approximately 60% of the total dry weight of mycobacteria. In comparison, gram-negative bacteria cell walls contain only 20%, and gram-positive bacteria 1% to 4%. Owing to this high lipid content, my-cobacteria have unusual growth and staining characteristics. Mycobacteria retain carbolfuschin dye after a wash with an acid-alcohol decolorizer (“acid-fast”), forming the basis for the Ziehl-Neelsen and Kinyoun stains used in direct detection methods.

Mycosides, a class of lipids formed from mycolic acids bound to carbohydrates as glycolipids are unique to
acid-fast organisms.³ Mycolic acid is a large α-branched, β-hydroxy fatty acid that varies in size from one species of Mycobacterium to another. Wax D is a large mycoside in which 15 to 20 molecules of mycolic acid are esterified to a large polysaccharide composed of arabinose, galactose, mannose, glucosamine, and galactosamine. Wax D, the active ingredient in Freund adjuvant, acts to increase the antibody response to an antigen. As a result, antigens that are themselves poorly immunogenic, such as tuber-culoprotein, induce host delayed-type hypersensitivity.

Unlike some bacteria, the virulence of pathogenic mycobacteria is not related to the secretion of toxins or the presence of a capsule. Cord factor is a mycoside in which two molecules of mycolic acid are esterified to disaccharide 6,6′-dimycolytrehalose. Cord factor allows virulent mycobacteria to grow parallel to each other forming long serpentine cords. In vitro, cord factor has been shown to inhibit migration of polymorphonuclear leukocytes. When stimulated with 6,6′-dimycolytrehalose, macrophages produce cytokines that induce inflammation and type 1 helper T cells: tumor necrosis factor alpha (TNF-α), interleukin-1 (IL-1), chemotactic factors, and IL-12. Inflammatory cytokines from host inflammatory cells that are stimulated by mycobacterial cord factor induce granuloma formation and may promote neovascularization through stimulation of vascular endothelial growth factor.⁴ Other lipid components of mycobacteria have also been associated with toxicity to macrophages.

M. leprae cell walls contain an antigenic glycolipid termed phenolic glycolic lipid 1 (PGL-1) that is unique to that species. Derived from phthiocerol dimycocerosate (PDIM) with the addition of O-methoxylated sugars, PGL-1 plays an important role in peripheral nerve-bacillus interactions.⁵ Both PGL-1 and the glycoprotein lipoarabinomannan may induce immunologic unresponsiveness of lymphocytes and macrophages in lepromatous leprosy.⁶

Mycobacteria are slow-growing organisms, with doubling times on the order of days to weeks. For instance, unlike Escheria coli (generation time of 20 minutes), M. leprae has a generation time of 12 days. The slow growth accounts for the long incubation periods for TB and leprosy (as long as 8 to 10 years) as well as the indolent course of most nontuberculous mycobacterial (NTM) keratitis. Most mycobacteria grow best at 37ºC, although several species, including the rapidly growing NTM, may grow best at lower temperatures. M. tuberculosis and NTM can be grown on artificial media, but no in vitro method of culturing M. leprae exists.

Mycobacterial species have traditionally been categorized on the basis of growth rate, colony morphology, and pigment production in relation to stimulation. Species referred to as rapid growers take less than 7 days to grow on solid media. This category contains the species most responsible for nontuberculous mycobacterial ocular infection. Slowly growing mycobacteria often require 2 to 3 weeks for growth, and are subdivided by their ability to produce pigment in relation to stimulation with light: photochromogens, scotochromogens, and non-photocromogens. Photochromogens produce carotenoid pigments when exposed to light, while scotochromogens produce a yellow-orange pigment in either light or dark. Nonphotochromogens are rarely pigmented, but may produce white, tan, or pale-yellow pigment. The three categories of slowly growing mycobacteria correspond to groups I (photochromogens), II (scotochromogens), and III (nonphotochromogens) proposed by Runyon to classify nontuberculous mycobacteria. Runyon group IV corresponds to the rapidly growing mycobacteria. Runyon numerical designations are not generally used today.

Differentiation of species by morphologic, physiologic, and biochemical characteristics is listed in Table 56-1. Mycobacterium avium and Mycobacterium intracellulare are not distinguishable by these tests and are therefore referred to together as M. avium complex.

Infection with M. tuberculosis occurs primarily by inhalation of airborne bacilli. The number of bacilli necessary to establish infection depends on host immunity, the ability of the organisms to multiply within macrophages, and the genetic resistance of the recipient. In the early stages of infection, alveolar macrophages nonspecifically ingest and either kill or suppress growth of bacilli. After several weeks, delayed-type hypersensitivity to wax D and tuber-culoprotein causes tubercle formation, in which caseous necrosis surrounds and destroys the nonactivated macrophages harboring growing bacilli. CD4⁺ T lymphocytes predominate in delayed-type hypersensitivity. Both TNF-α and transforming growth factor β (TGF-β) are involved and contribute to the associated tissue damage.⁷

Tubercle bacilli can survive but cannot multiply within the anoxic and acidic caseating necrosis. Cell-mediated immunity, driven primarily by interferon-γ (IFN-γ) and possibly other cytokines secreted by activated lymphocytes, helps to destroy any bacilli that escape areas of caseation. Both CD4⁺ T lymphocytes and CD8⁺ T lymphocytes are involved.⁸

Nontuberculous mycobacteria are found in soil, water, and dust. They have been identified as normal flora of skin, sputum, and gastric contents.⁹ They are opportunistic pathogens, infecting tissue when local resistance is

compromised. In most cases, there is a history of antecedent trauma, surgery (e.g., penetrating keratoplasty, laser in situ keratomileusis, radial keratotomy, or cataract surgery), corneal foreign body, or contact lens wear.¹⁰, ¹¹ and ¹² In one study, 91% of patients with NTM keratitis had a history of prior trauma or surgery.¹³ Pulmonary infections have been reported, but person-to-person transmission does not occur. Inoculation of the corneal stroma stimulates both acute and chronic granulomatous inflammation, which limits the spread of infection.¹⁴ The use of topical corticosteroids suppresses the granulomatous response believed necessary to limit the spread of my-cobacteria and may prolong or worsen the course of the disease.¹⁴

The multifaceted spectrum of disease caused by M. tuberculosis in humans is a challenge to accurately model in animals. Still, a wide variety of models are available owing to broad susceptibility to infection across a number of species. Several models have been developed to study M. tuberculosis pathogenesis, to identify active drug treatments, and to evaluate potential vaccine efficacy. The two most commonly used models involve the guinea pig and the mouse.

Susceptibility of the guinea pig to infection with the human tubercle bacilli is comparable to that of human infants or immunodeficient adults. A model using low-dose aerosol inoculation to more closely approximate the conditions for human infection has been developed.¹⁵ Virulent tubercle bacilli are delivered via the respiratory route to the lungs. Twenty-eight days after infection, lesions are visible by X-ray, and organisms can be recovered from lesions. The guinea pigs develop high tuberculin sensitivity and show considerable caseous necrosis on histopathology, which parallels the condition in immunocompetent humans. Infection invariably progresses toward fatal disease in months to years. This animal model has been used extensively to evaluate the response to chemotherapy and to investigate the protective effect of potential TB vaccine candidates.¹⁶

The mouse has a resistance to TB similar to that of immunocompetent humans. Unlike humans, mice develop only low degrees of tuberculin sensitivity and show little caseous necrosis. The mouse model is useful in studying mutants of M. tuberculosis, immunopathology associated with TB, drug efficacy, and for evaluating vaccine candidates. Some models use immunocompetent animals; others use immunodeficient animals such as the nude mouse, the severe combined immunodeficiency mouse, or animals rendered immunodeficient by thymectomy followed by CD4⁺ T-cell depletion by means of purified anti-CD4 monoclonal antibody.¹⁷, ¹⁸ Both aerosol and intravenous inoculations have been used. Drawbacks include the fact that murine models of latent infection may not accurately reflect latent infection in humans and different pathology with regard to granulomas. Latent infection is difficult to model in mice because bacterial numbers remain high even when the infection is controlled. With persistent infection, the lungs may show increased pathology. The disease is slowly progressive in mice.¹⁹, ²⁰ On histopathologic examination, granulomas in mice are not as well organized as those in humans and seldom exhibit necrosis. Several important observations about the pathogenesis of M. tuberculosis have come about from mouse models: CD4⁺ T cells are an important component of the immune response in M. tuberculosis infection, CD8⁺ T cells are important in protection against M. tuberculosis infection, IFN-γ is not the sole effector mechanism for CD4⁺ T cells, TNF-α is important in host immune response, the level of IFN-γ or IL-12 is also important in host defense against mycobacterial infection, and even robust T-cell and IFN-γ response may not be adequate for protection against TB infection.²⁰

Because adult reactivation TB accounts for the majority of clinical TB cases in the world, an animal model that leads to the reactivation of dormant bacteria followed by the development of TB is desirable. The best-documented model is the Cornell model, in which mice are infected intravenously and then treated with isoniazid and pyrazinamide.²¹

After 12 weeks, the spleen and lungs are sterile. If the animals are left untreated, the disease will reactivate in 60%, resulting in TB 3 to 4 months later. This model may be used to evaluate potential therapeutic vaccines that target persons who are already infected with M. tuberculosis but do not have clinical disease.

Other animal models include goldfish and zebrafish infected with Mycobacterium marinum used to study virulence factors of mycobacteria.

With the recognition of the impact of M. avium complex infections in patients with AIDS, an animal model for this disease has been desirable. The beige mouse is naturally immune deficient and develops uncontrolled infection when intravenously inoculated with M. avium complex organisms, but it is not clear whether the efficacy of drug therapy in this model accurately predicts the therapeutic response in humans.²²

Animal models of M. fortuitum and M. chelonae keratitis have been described that closely mimic human disease in their clinical and histopathologic features.¹⁴, ²⁰, ²³ These models are useful because of the limited correlation between in vitro drug sensitivities and clinical efficacy,
and have demonstrated that different species of nontuberculous mycobacteria have different susceptibilities to various antimicrobial agents.

The prolonged growth cycle and preference for low temperatures of M. leprae were challenges to the development of animal models of leprosy. In 1960, Shepard developed the mouse footpad model by successfully demonstrating multiplication of M. leprae in the footpads of Carworth Farms white mice. In immunocompetent mice, the infection is localized to the footpads without much histopathologic change (i.e., only a few small granulomas are formed). In immunodeficient mice, the entire footpad becomes a foreign body-type macrophage granuloma and disseminated disease is possible. From this model, the importance of T lymphocytes in host defense against M. leprae was elucidated. Moreover, with athymic nu/nu mice, routine culture of large numbers of M. leprae for investigational use was possible. Gene knockout mice have been used for immunologic studies and have recently substantiated the role of cytokines in the variable immune response to M. leprae infection.²¹ The nine-banded armadillo is the most widely used model of leprosy. It has a low basal temperature (30ºC to 35ºC) that permits replication of M. leprae, and regularly develops disseminated infection in immunocompetent hosts, unlike the mouse model. The nine-banded armadillo may be inoculated via intraperitoneal, percutaneous, or respiratory instillation and intravenously. The most rapid and severe infections are achieved via intravenous inoculation.²⁴ The granu-lomatous response to M. leprae is the same as in human hosts. This animal model of leprosy is also unique in that effective protection against M. leprae has been achieved with BCG vaccination. The nine-banded armadillo model is useful for studying susceptibility and resistance to various therapies as well as vaccine candidates.