Clinical and Experimental Advances in Treatment of Visceral Leishmaniasis

Visceral leishmaniaisis (kala-azar) is a disseminated protozoal infection, transmitted by sandfly bite, in which macrophages of the liver, spleen, and bone marrow are preferentially parasitized and support intracellular replication. Most human infections caused by visceralizing strains of Leishmania are probably subclinical (13, 101, 139), attesting to innate resistance or, more likely, to T (Th1)-cell-dependent immune responses which induce acquired resistance (33, 39, 79, 101, 102). While treatment is not given for subclinical infection, remote recrudescence still remains a possibility, especially if the host becomes T-cell deficient (62, 66, 76, 123). In contrast, if the initial Th1-cell-associated immune response fails to develop or its effector mechanisms are disabled or not properly maintained (122, 123), recently acquired (or reactivated) kala-azar evolves to full expression as a subacute or chronic illness for which treatment is required.

Visceral leishmaniaisis occurs in >80 countries in Asia and Africa (Leishmania donovani), southern Europe (L. infantum), and South America (L. chagasi). However, L. donovani is the principal pathogen, and 90% of the estimated 500,000 new symptomatic cases per year arise in just five countries—India, Sudan, Bangladesh, Nepal, and Brazil (10). Decades-old epidemics in northeast India and for the past decade in Sudan have largely maintained the disease (77, 139); India alone may contribute as many as 40 to 50% of the world's cases.

As recently as the early 1990s, treatment for kala-azar worldwide in both children and adults was essentially limited to pentavalent antimony (Sb), in use for 50 years. Sb therapy requires once-daily injections usually for 28 days, is not necessarily well tolerated, and is now ineffective in the region (Bihar State) which houses 90% of India's cases (i.e., ∼40% of the world's cases) (175). Fortunately, clinical trials carried out during the past decade have opened the door to a range of new treatments, including short-course (even single-dose) parenteral regimens and highly effective oral therapy (Table 1) (123). Novel experimental approaches also continue to be identified via testing of new antileishmanial compounds and macrophage-targeted drug delivery systems. Since T cells and immune pathways are closely linked with initial and/or long-term treatment efficacy, harnessing immunologic mechanisms to act with chemotherapy or even alone represents an additional experimental approach in the ongoing effort to optimize the host response.

TABLE 1.

Treatment alternatives in visceral leishmaniasis

Treatments available
1950–1990	1990–2000
Pentavalent Sb	Pentavalent Sb
Pentamidine	Pentamidine
	Amphotericin B
	Lipid-associated amphotericin B, as amphotericin B cholesterol dispersion, amphotericin B lipid complex, or liposomal amphotericin B
	Amphotericin B in fat emulsion
	Aminosidinea
	Combination chemotherapy with aminosidine and Sb
	Immunochemotherapy with interferon-γ and Sb
	Oral chemotherapy with miltefosine

^aNot currently available; new formulation is being manufactured and tested under WHO auspices. 

While drug management in kala-azar has evolved rapidly and with success (Table 1), obstacles continue to limit the impact of these advances in regions of endemicity. In developing countries, the rural settings in which infected patients typically live (making access to timely or proper medical care difficult), bare-bones national health expenditures, inconsistent drug availability, and poor nutrition are traditional obstacles (124). Relapse after seemingly successful therapy, even in immunologically intact-appearing patients, has also remained a chronic problem ever since active treatment (Sb) was first introduced. Three recently encountered obstacles include the prohibitive cost of an effective new class of agents (lipid formulations of amphotericin B), intersection with human immunodeficiency virus (HIV) with a predictable increase in treatment failures, and large-scale resistance to Sb in India. Nevertheless, developments in the treatment of visceral leishmaniasis represent clear-cut advances by any measure. This report highlights this clinical progress and then looks at new therapeutic approaches driven by continued experimental work in the research laboratory.

CURRENT INJECTABLE CHEMOTHERAPY

Pentavalent antimony (Sb).

Despite prolonged duration of therapy and adverse reactions (22), Sb remains the first-line treatment in all regions of the developing world (except Bihar, India) (22, 77, 175) because of decades of clinical experience, proven efficacy (>90% long-term cure rate), administration by injection (intravenous or intramuscular [i.m.]) rather than infusion, and what has been considered acceptable cost (see below). The recommended regimen consists of once-daily injections of full-dose drug (20 mg/kg of body weight) for 28 days. While active elsewhere in India, Sb is no longer useful in Bihar, where as many as 65% of previously untreated patients now fail to respond or promptly relapse (175); resistance to Sb in L. donovani isolates from Bihar has also been formally demonstrated in the laboratory (94).

Amphotericin B.

In experimental animal models, amphotericin B is one of the most active antileishmanial agents (20). Largely because of the decline and fall of Sb in India and the failure of pentamidine as a satisfactory substitute (22, 123), conventional amphotericin B deoxycholate has been rediscovered in kala-azar as an effective but arduous treatment. In India, infusions of 1 mg/kg given either daily for 20 days (181) or, more commonly, on alternate days (15 infusions over 30 days) (170) regularly induce long-term cure in >90 and up to 98% of both Sb-unresponsive and previously untreated patients, respectively (123, 181). Drawbacks to amphotericin B include the requirement for infusions, length of therapy, adverse reactions, close laboratory monitoring for potential toxicity and, to some extent, cost (see below).

Lipid formulations of amphotericin B and advent of short-course regimens.

The new formulations of amphotericin B provided the important opportunity to reduce duration of therapy in kala-azar while preserving efficacy (49, 53, 166); these formulations allow considerably higher daily doses of drug and simultaneously appear to target infected tissue macrophages via enhanced phagocytic uptake. Each of the three commercially available preparations, given once daily by infusion, is well tolerated and their usefulness in kala-azar has exceeded all clinical expectations (123). Provided sufficient total doses are administered, short-course regimens of as brief as 5 days (using daily infusions) or up to 10 days (during which five or six infusions are given) are remarkably active (49, 53, 169, 176). Efficacy has been documented worldwide in children and adults and in severely ill patients under appalling conditions in war-torn Sudan (23, 54, 154).

There are regional differences in responsiveness to the lipid formulations, which may relate to patient age, infecting Leishmania strain, and/or visceral parasite burden (123). Indian kala-azar, which occurs in adults and children, responds best as judged by trials in which amphotericin B lipid complex (Abelcet; The Liposome Company, Princeton, N.J.) and liposomal amphotericin B (AmBisome; NeXstar Pharmaceuticals, San Dimas, Calif.) were tested (23, 169, 176). Brazilian infection (L. chagasi), mostly in children, responds to amphotericin B cholesterol dispersion (Amphotec; Sequus Pharmaceuticals, Inc., Menlo Park, Calif.) but may be less responsive to AmBisome (23, 53, 64). In the Mediterranean region, where young children are often targets for L. infantum, higher total doses of AmBisome are required but the response rate is near 100% (54).

At the same time, these agents also introduced an insurmountable obstacle to wide deployment—prohibitively high cost (Table 2). This factor alone has frustrated clinical application of even short-course regimens in developing countries, reemphasizing cost as the primary determinant of whether the most promising treatments actually reach the field. Using low-dose or even more-compressed regimens (e.g., single-dose therapy) and testing generic preparations of lipid-mixed or -associated amphotericin B represent responses to the problem of cost. Although not yet borne out (171), it is possible that low-dose, short-course regimens might engender resistance to amphotericin B.

TABLE 2.

Estimated costs (per patient) for treating a 25-kg patient with Indian kala-azar by conventional or short-course regimens reported to induce ≥90% cure rate^a

Agent (reference)	Regimen	Duration (days)	Initial cost ($)			Treatment failures (%)	Final cost ($) (total)c
			Hospital	Drugb	Total
Pentavalent Sb (175)d	20 mg/kg/day	28	359	11	370	10	423
Amphotericin B (123, 182)	1 mg/kg/day	20	325	60	385	1	391
	1 mg/kg qode	30	400	48	448	8	518
Aminosidine (82)	16 mg/kg/day	21	244	50	294	7	331
Amphotericin B fat emulsion (177)	2 mg/kg qod	10	150	69	219	7	256
Amphotericin B lipid complex (169)	2 mg/kg/day	5	95	582	677	10	764
Liposomal amphotericin B (177a)	1 mg/kg/day	5	95	564	659	7	720
	5 mg/kg once	1	65	564	629	9	707
	7.5 mg/kg once	1	29	752	781	10	868

^aBased upon intention-to-treat analyses of cure and cost assumptions summarized elsewhere (123) or discussed in text. 

^bRetail drug costs in India, except for amphotericin B lipid complex and liposomal amphotericin B (U.S. average wholesale price [9]). 

^cIncludes retreatment of nonresponders with arbitrarily selected, 100% effective short-course regimen (amphotericin B lipid complex at 3 mg/kg/day for 5 days [169]; $871 per treatment course). 

^dIn areas outside of Bihar State. 

^eqod = alternate-day regimen. 

^fSundar et al., submitted for publication. 

Single-dose regimens.

In addition to the expectation that treatment could be made more affordable by maximally reducing hospital stay and related costs, three other factors led to testing single-dose lipid-associated amphotericin B (Abelcet, AmBisome) in Indian kala-azar: (i) the observation that giving total doses of as low as 3.75 mg/kg (AmBisome) or 5 mg/kg (Abelcet) over a 5-day period (i.e., 0.75 or 1 mg/kg/day, respectively) induced reasonable cure responses of ∼85 to 90% (169, 176); (ii) prior experimental data indicating high-level efficacy for single-dose liposomal amphotericin B (20); and (iii) data indicating that administering the total dose of drug (Sb) as a single injection is as or more effective than delivering the same total dose divided into daily injections (21, 112).

The initial trial in India, carried out with single-dose Abelcet (5 mg/kg), yielded a 70% long-term cure rate (171); however, single-dose AmBisome at 5 mg/kg (177a) or 7.5 mg/kg (S. Sundar, T. Jha, C. Thakur, M. Mishra, and R. Buffels, Abstr. 40th Intersci. Conf. Antimicrob. Agents Chemother., p. 23, 2000) induced cure rates of 91 and 90%, respectively. In the latter trial, all 203 subjects treated with 7.5 mg/kg were routinely and safely discharged within 24 h after treatment, indicating high-level efficiency as well. The stable, long-circulating nature of AmBisome (I. Bekersky, D. Dressler, R. M. Fielding, D. N. Buell, and T. J. Walsh, Abstr. 40th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 856, 2000) may explain its enhanced efficacy in this single-dose setting. While such regimens have not yet been tested outside of India, the clinical appeal is obvious. Paradoxically, single-dose AmBisome is not cost effective because of the price of the drug (Table 2).

Generic preparations of lipid-mixed and -associated amphotericin B.

Simply hand-mixing of amphotericin B deoxycholate with a commercially available lipid emulsion appears unlikely to meaningfully associate drug and lipid, and the use of such preparations is controversial (159). Nevertheless, amphotericin B diluted in lipid emulsion was tested for kala-azar in an attempt to duplicate beneficial clinical effects but at a much reduced cost. Sixty-five of 70 Indian patients (93%) were cured after receiving five alternate-day infusions of 2 mg of amphotericin B/kg mixed with 20% lipid emulsion (177). Although this 10-day treatment is the most cost-effective parenteral regimen for Indian kala-azar (Table 2), additional trials, including testing a 5-day regimen, have not been carried out because of perceived difficulties in standardizing such preparations. Local manufacture of a lipid formulation with more affordable pricing represents a separate approach; one such generic preparation of liposomal amphotericin B is being tested in India (26).

Aminosidine.

This aminoglycoside, identical to paramomycin sulfate and given once daily usually by i.m. injection, has been combined with Sb to successfully reduce duration of therapy (153, 182). Aminosidine also appears to be active in India when used alone (16 to 21 mg/kg/day for 21 days) in a region of high-level Sb resistance (83, 183). The World Health Organization (WHO) has proposed $50 as the drug cost for aminosidine treatment in an adult (123). Drawbacks to this agent include the length of treatment, potential for oto- or nephrotoxicity, and insufficient recent clinical experience.

Cost.

The cost of Sb plus the overall expense of 28 days of treatment represents the worldwide benchmark against which newly introduced agents and regimens are measured. For example, in India, 28 days of treatment with a locally manufactured Sb preparation at 20 mg/kg/day costs approximately $11 for drug in an adult (123); $423 is a fair calculation for total cost (Table 2). Not surprisingly, drug costs for the same regimen using internationally available Sb formulations are higher: approximately $100 to $125 for meglumine antimoniate and $150 to $200 for sodium stibogluconate (123). In Bihar, India, where Sb is no longer useful (175), the cost of amphotericin B treatment currently represents the new benchmark (Table 2). Drug cost for an alternate-day regimen of 15 alternate-day infusions of 1 mg of locally obtained amphotericin B/kg is $48 ($6 per 50-mg vial) (123).

If WHO expectations prove accurate, the proposed $50 cost for aminosidine and overall costs should not deter its use (Table 2). However, just the opposite is true for the lipid formulations of amphotericin B, for which U.S. average wholesale prices are as follows: AmBisome ($188 per 50 mg), Amphotec ($93 per 50 mg), and Abelcet ($194 per 100 mg) (9). If such prices were applied in developing countries, none of these agents would ever be used. However, (i) a different local price scale may be offered, (ii) actual acquisition costs may be ∼30 to 40% lower, and (iii) short-course regimens reduce hospital-associated expenses (even though modest to begin with in regions such as India and Africa) (123). Thus, in a previous pharmacoeconomic analysis in which the three preceding factors and overall management were considered in India (drug plus hospitalization), the cost of 5 days of low-dose Abelcet treatment (1 mg/kg/day) could be brought into line with that of 15 alternate-day infusions of amphotericin B (1 mg/kg) given over 30 days (123, 169). In contrast, although hospital-related cost reductions are guaranteed, neither the total cost of the 5-mg/kg single-dose AmBisome regimen (with a 5-day hospitalization initially proposed to verify a clinical response [177a]) nor that of the subsequently tested 7.5-mg/kg single-dose regimen (with its ≤24-h hospital stay) is sufficiently low to permit their use (Table 2).

The preceding cost considerations, drawn from experience in India, are likely to be relevant to other developing countries. In regions of southern Europe where kala-azar is endemic, hospital-associated expenses are, of course, many fold higher. Under these conditions, the cost of a lipid formulation of amphotericin B would almost certainly be offset by savings resulting from reduced hospital stay.

ORAL CHEMOTHERAPY

Numerous oral agents have been tested and discarded for kala-azar (22, 123, 138); an 8-aminoquinoline, first reported in 1994 as having promise (158), is currently being restudied. However, in the past 3 years, the membrane-active phospholipid derivative hexadecylphosphocholine (miltefosine) has been identified as the first effective oral treatment in visceral infection (123). Originally developed as an antineoplastic agent, miltefosine also demonstrated experimental antileishmanial activity (43, 46, 90, 92), providing the rationale for testing in India (172). Results from a large, recently completed phase III study in Bihar have confirmed the >95% cure rate documented in four prior trials (84, 172–174; J. Engel, personal communication). Thus, the long-sought objective of oral therapy for kala-azar has likely been achieved and clearly represents a major therapeutic advance. The recommended regimen for miltefosine in adults (≥12 years old) is projected to be 28 days, with dose based upon body weight: 50 mg twice daily for adults ≥25 kg and 50 mg once daily for those <25 kg. Preliminary data from an initial phase I/II study in Indian children also indicate safety, satisfactory tolerance, and efficacy (J. Engel, personal communication). Limitations of miltefosine include adverse reactions (primarily self-limited gastrointestinal reactions), duration of therapy, the absence of experience with kala-azar outside of India, and teratogenicity in animals, which precludes its use in pregnant women (123). Miltefosine has not yet been approved for use nor priced for sale; thus, its cost is an important unknown.

HOST IMMUNE MECHANISMS AND RESPONSE TO TREATMENT

Two aims of the effort to understand the host immune response in kala-azar relate specifically to treatment and are intertwined. The first aim is to determine how immune mechanisms regulate the efficacy of chemotherapy, both at the time drug is given and in prevention of posttreatment relapse. The second aim is to identify specific immunologic components or interventions which can be translated into treatment and either used alone or, more plausibly, in combination with chemotherapy. The use of activating, prohost defense cytokines on the one hand and inhibition of deactivating cytokines on the other represent two such therapeutic interventions.

Acquired resistance.

The complex cell-mediated immune response in visceral infection is likely modulated by a range of innate and environmental factors (e.g., nutrition) (25, 37) and also effector cells (natural killer cells, CD8⁺ cells, and perhaps neutrophils [160]) (122). However, most evidence (primarily generated experimentally) points to a mechanism of resistance which (i) is T (CD4⁺)-cell dependent and involves T-cell costimulatory pathways (71, 107, 162); (ii) requires secretion of regulatory, activating cytokines (primarily Th1-cell associated, including interleukin 12 [IL-12] and gamma interferon (IFN-γ) (18, 34, 57, 114, 116, 125, 126, 152, 179, 187); (iii) induces adhesion molecule- and chemokine-mediated recruitment of inflammatory mononuclear cells into infected tissue (41, 127) and within assembled granulomas; and (iv) culminates in activation of leishmanicidal mechanisms in parasitized resident macrophages and influxing blood monocytes (16, 121, 122). If this response develops fully, the likely outcome is killing of most intracellular parasites, induction of quiescence in residual organisms, and maintenance of low-level infection in a life-long, asymptomatic state (113, 119).

Activating cytokines.

In experimental visceral infection, at least five pleiotropic cytokines, IL-12, IL-2, IFN-γ, granulocyte-macrophage colony-stimulating factor (GM-CSF), and tumor necrosis factor (TNF), interdigitate to regulate key endogenous mechanisms of acquired resistance: generation and maintenance of the Th1-cell response (IL-12, IFN-γ), induction of IFN-γ secretion (IL-12, IL-2), mobilization of blood monocytes (GM-CSF), granuloma assembly (all five cytokines), and macrophage activation (IFN-γ, TNF, GM-CSF) (16, 57, 114, 116, 122, 125, 126, 152, 179, 187). These five cytokines are also expressed (or likely expressed) in human kala-azar (18, 34, 36, 42, 68, 69, 86, 88, 99, 100, 142, 168) and are thus presumably poised to interact in endogenous form with chemotherapy. Except for TNF, each activating cytokine is also in clinical use and therefore potentially available for testing in exogenous form in combination with antileishmanial drugs (immunochemotherapy).

Deactivating cytokines.

The failure to spontaneously control visceral infection may reflect an intrinsically inert or unstable Th1-cell-inducing mechanism (87); however, most information points instead to active suppression. Factors implicated experimentally and/or clinically in deactivating Th1-cell responses and favoring visceral infection include Th2-cell-associated cytokines (IL-4, IL-10, IL-13) (11, 12, 18, 34, 51, 56, 61, 68, 80, 81, 86, 88, 100, 120, 150, 168, 193) and transforming growth factor-β (TGF-β) (71, 148, 187, 188). Nevertheless, if the Th1-cell response evolves and predominates, the parallel generation of regulatory cytokines recognized as suppressive appears of little consequence (102) and probably beneficially limits tissue inflammation.

Fully established kala-azar, however, may well represent the clinical expression of a net suppressor response, for example, an imbalanced (rather than overtly polarized) Th2 > Th1 cell state. While popularized in the pathogenesis of experimental L. major infection (143), the potential role of IL-4 and IL-4-driven differentiation to the Th2-cell phenotype is much less clear in both human and experimental visceral infection. While some reports implicate IL-4 in progressive infection (11, 120, 150, 168, 193), others do not (12, 18, 56, 68, 86, 87, 88, 99, 100, 102, 151, 187). In contrast, the broadly immunodeactivating cytokine, IL-10, appears more likely to act as a primary suppressive factor, particularly in human infection (12, 67, 68, 86, 120, 122, 131, 168).

Although the presence of downregulating cytokines such as IL-4 and IL-10 does not directly impair the experimental in vivo efficacy of antileishmanial chemotherapy (131; H. Murray and R. Coffman, unpublished observations), these factors (i) promote intracellular infection and increase the parasite burden against which drug is required to act, (ii) reduce the secretion of or target cell response to activating cytokines which regulate or optimize the efficacy of chemotherapy (Sb) (see below), and (iii) may foster relapse after apparently effective therapy (12, 67, 68, 86, 120, 122, 131, 168). Suppressive endogenous factors such as IL-10 are relevant here since they represent viable targets for therapeutic intervention (123, 131). Thus, rather than administering (or inducing endogenous production of) a Th1-cell cytokine to enhance the response to chemotherapy (110, 125), treatment designed to neutralize IL-10 (34), for example, or block its receptor might instead be given prior to chemotherapy.

HOST DETERMINANTS OF RESPONSIVENESS TO CHEMOTHERAPY

Sb.

When applied in vitro to unstimulated macrophages, Sb induces leishmanicidal effects in the absence of additional cells or added cytokines (111). In experimental visceral infection, however, Sb's activity is not direct but is strictly dependent upon host T cells, influxing blood monocytes, and an intact Th1-cell cytokine response (IL-12, IFN-γ) (4, 111, 125, 127, 128); endogenous TNF plays a role as well (126). Paradoxically, Sb is also less active in L. donovani-infected IL-4 knockout (KO) mice (4), perhaps related to enhanced IL-10 production and/or the absence of IL-4's less well-appreciated potential to prime for a Th1 response (4).

IFN-γ upregulates mononuclear phagocyte accumulation of Sb (110), an effect which may be important in intracellular infection. Nevertheless, requirements for both endogenous IFN-γ and TNF in Sb responsiveness also suggested that simultaneous induction of macrophage activation might provide a two-hit mechanism (drug plus macrophage-derived toxic products) to achieve optimal parasite killing. However, L. donovani-infected mice deficient in the activated macrophage's primary leishmanicidal pathways, governed by inducible nitric oxide synthase (iNOS) and phagocyte oxidase (121), respond normally to Sb (128).

An alternative role for T cells and cytokines in Sb's in vivo efficacy may relate to granuloma assembly—the mechanism that encloses L. donovani within a structure rich in recruited cells, soluble mediators, and inflammatory signals which together may enable drug-induced killing in the tissues (122). Mice overtly deficient in granuloma development (T-cell-deficient athymic [nude], IFN-γ and IL-12 KO, and intracellular adhesion molecule-1-deficient mice with defective monocyte influx [111, 125, 127, 128]) are also experimental hosts which fail to respond to Sb. Thus, the capacity of IL-12, IFN-γ, and TNF to attract, activate, and retain influxing blood monocytes and T cells within the parasitized tissue focus may explain regulation of Sb efficacy in the intact host. The granuloma-remodelling action of treatment with exogenous IL-2 or GM-CSF, which deliver mononuclear (IL-2) or myelomonocytic (GM-CSF) cells to infected tissue (114, 116), might represent a separate approach to enhance the effect of chemotherapy.

Amphotericin B and miltefosine.

Amphotericin B and miltefosine are also directly microbicidal towards intracellular L. donovani amastigotes in vitro (123). However, and in clear-cut contrast to Sb, both agents act independently of the immune response and retain full leishmanicidal activity in animals devoid of T cells, IL-12, and IFN-γ, influxing blood monocytes, activated macrophages, and granulomas (115, 125, 127–129). Nevertheless, maintaining intracellular L. donovani in a long-term, quiescent state requires T cells (119) and, not surprisingly, infection in nude mice relapses once either amphotericin B or miltefosine treatment is discontinued (119, 130).

Relapse after chemotherapy.

Posttreatment relapse has long been recognized in visceral leishmaniasis in otherwise healthy individuals and is a predictable result in T-cell-deficient patients (123). Indeed, despite an apparently complete clinical and parasitologic response to initial therapy, no treated patient with kala-azar is considered cured until at least six additional months have passed uneventfully.

The mechanisms which maintain the drug-treated host relapse-free have received relatively little experimental attention. Logic suggests that the same responses (T-cell dependent, cytokine induced, activated-macrophage mediated) which induce initial acquired resistance also maintain residual, posttreatment parasites in a latent state. Yet, except for a uniform requirement for T cells (119, 130), experimental results are at odds with this assumption. For example, (i) the broad IL-12-driven Th1-cell response, (ii) specific IL-12-induced IFN-γ secretion, and (iii) activated macrophage generation of iNOS-derived reactive nitrogen intermediates are all required (and likely act together) in normal mice to control and resolve initial L. donovani infection (57, 121, 125, 152, 179). Since no antileishmanial agent would be expected to kill 100% of intracellular visceral amastigotes, one would predict that infection in a host lacking IL-12, IFN-γ, or iNOS would relapse after drug treatment was stopped. However, (i) following comparable >95% microbicidal responses to amphotericin B therapy (125, 128), IFN-γ but not IL-12 KO mice relapse (H. Murray, unpublished observations), and (ii) after similar killing induced by amphotericin B or brief treatment with Sb, residual visceral infection does not reactivate in iNOS KOs (128).

The preceding observations refocus attention on endogenous IFN-γ but also provoke the question of how drug treatment permits otherwise critical host defense mechanisms, IL-12 secretion and iNOS expression (121, 125), to become dispensable. Thus, more should be learned about the compensatory mechanism(s) for prevention of relapse, since such information has clinical potential. If a discrete regulatory mechanism is deficient in a relapse-prone patient population, it might be amenable to therapeutic modulation or some form of reconstitution if demonstrated to be absent.

Response to treatment in AIDS-related kala-azar.

Treatment results in T-cell- and cytokine-deficient animals suggested that CD4⁺ T-cell-depleted patients with AIDS-associated visceral leishmaniasis would (i) respond poorly to Sb but satisfactorily to amphotericin B and (ii) likely relapse if initial treatment successfully induced an apparent clinical response and drug was then discontinued. Taken together (but with variability in treatment regimens and definitions of efficacy), most reports from southern Europe, where coinfection has been best demonstrated (6, 8), appear to confirm the following: (i) overall, approximately 50% of patients fail to initially respond to Sb in a region where 0 to 5% of otherwise healthy individuals are Sb unresponsive (72); (ii) of a total of approximately 50 coinfected patients treated with some form of amphotericin B, >90% showed initial responses; and (iii) relapse rates in HIV-related kala-azar after any treatment is discontinued are >50% and up to 90 to 100% (reviewed in references 6 and 123).

Results from Spain, however, in the only randomized controlled study in HIV-associated kala-azar (91), provided a different finding in that the initial efficacies of both Sb (66% response) and amphotericin B (62% response) were reduced. Since this study did not include secondary prophylaxis, the majority of initial responders to either treatment relapsed. However, while once-monthly injections of Sb may prevent symptomatic recurrences (144), no consensus has been reached about what constitutes optimal maintenance treatment in such patients. If satisfactorily tolerated in AIDS-related kala-azar, miltefosine may have a dual future role as initial and maintenance oral therapy, since its experimental activity is T-cell independent and once-weekly treatment prevents relapse in T-cell-deficient mice (129, 130).

Application of cytokine immunochemotherapy: IFN-γ and GM-CSF.

Experimental definition of IFN-γ's capacity to activate macrophages to kill L. donovani and act synergistically with Sb both in vitro and in vivo (108, 110, 112) led to its clinical application in kala-azar. Results from limited or uncontrolled pilot trials, which included several patients coinfected with HIV, suggested that combining IFN-γ injections with Sb accelerated and enhanced overall efficacy (reviewed in reference 123). In Sb-refractory patients in Brazil and India, retreatment with Sb plus IFN-γ also induced long-term responses in about two-thirds of subjects (14, 164). Nevertheless, in a controlled study in India (Bihar) of previously untreated patients, there was no meaningful difference in long-term cure rates induced by IFN-γ plus Sb versus Sb alone (49% versus 36%, respectively) (167). The emergence of higher-level Sb resistance in Bihar likely influenced low overall response rates, and the majority of subjects receiving the combination may have in effect been receiving IFN-γ alone—a treatment active in experimental infection (112) but only modestly so in humans (123). If IFN-γ's capacity to accelerate the response to Sb and/or reduce the duration of treatment (165) is reexamined, it should be tested in a region where Sb resistance is low.

In experimental infection, GM-CSF treatment also induces macrophage antileishmanial activity against L. donovani associated with peripheral blood leukocytosis and striking accumulation of monocytes and granulocytes at developing tissue granulomas surrounding infected macrophages (116). Although not tested with Sb experimentally, GM-CSF was used in Sb-treated Brazilian children to successfully ameliorate the characteristic leukopenia of kala-azar and reduce susceptibility to secondary infections (15). This trial was not intended to detect a beneficial effect on the antileishmanial response to Sb (15); however, results in cutaneous infection suggest that intralesional GM-CSF and parenteral Sb can be combined to produce superior clinical effects (5).

LABORATORY SYSTEMS FOR TESTING ANTILEISHMANIAL TREATMENT

In vitro models.

Determining antileishmanial effect and mechanism of action in controlled in vitro systems remain basic to identification of new, viable therapeutic approaches. Cell-free systems provide information about direct antiparasitic effects (30); however, the intracellular (phagolysosomal) location of Leishmania within the infected host makes testing in cell culture more meaningful. Appropriate target cells include those which support parasite replication: human or mouse macrophages and macrophage- or monocyte-like cell lines (43, 96, 108, 110, 156). These same cells can be used to screen for overt drug-induced toxicity, and their effector functions can be manipulated to explore how macrophage antileishmanial mechanisms may interact with chemotherapy. Further, at least for Sb, intracellular testing can also demonstrate relevant modifications in drug metabolism and effect (82, 145) and confirm parasite susceptibility versus resistance, providing reasonable correlations with clinical observations (94).

The logical form of Leishmania to test is the amastigote, to which sandfly-inoculated promastigotes transform within host cells and which causes intracellular infection in vivo. Axenically grown amastigotes, free of host cellular factors, are now available for use, particularly in cell-free test systems (29, 58, 156). To initiate infection of cultivated cells in vitro, either amastigotes (obtained from infected animals or axenically prepared) or culture-maintained promastigotes can be used; the latter assume the amastigote form after ingestion. In such models, drug is typically added to the culture medium 24 h after parasite challenge; thereafter, measurements made daily indicate no effect (continued intracellular replication), static activity (inhibition of replication), or microbicidal effects (amastigote destruction and digestion).

In vivo models.

Studying visceral infection induced in laboratory animals, primarily mice and hamsters but also dogs and monkeys (16, 20, 21, 46, 90, 92, 122, 141, 187), has repeatedly generated basic information and experimental stepping stones pointing to treatments worth clinical pursuit. Work in mouse models, for instance, amply demonstrated the high-level efficacy of oral miltefosine (43, 46, 90, 92). Some investigators lean towards infection in hamsters as a more stringent test for drug efficacy since, unlike mice, hamsters injected with visceralizing Leishmania strains develop fatal infection. However, well-characterized models in normal mice with varying innate susceptibility and in immunodeficient and in gene-modified mice provide for considerable experimental flexibility and innovation. More importantly, these models have a history of yielding interpretable results with satisfactory correlation with potential activity in human infection (20, 46, 90, 92, 110, 123, 133, 134). Differences in drug metabolism and pharmacokinetics in animals versus humans must also be considered.

Testing treatment in established visceral infection is intuitively more relevant than examining effects of prophylaxis or therapy given within a day or two after parasite challenge. Thus, injecting susceptible animals with a visceralizing strain, leaving them undisturbed for 10 to 14 days (or sometimes longer) to allow for progressive infection, and then initiating treatment seems most reasonable. Depending upon experimental intent, additional variables in in vivo models include length of treatment, use of optimal or suboptimal drug doses, coadministration of a second agent, and timing of observations.

In addition, while of unclear clinical relevance since human infection in spleen and bone marrow responds to Sb treatment, there are organ-specific differences in Sb's efficacy in parasitized mice. Infection in mouse liver responds appreciably better to free Sb than does infection in spleen or bone marrow, leading some investigators to prefer judging treatment results by effects in the spleen (28, 31). It is possible that the liver better expresses the T-cell-dependent responses required for Sb's efficacy; however, failure to achieve sufficient intracellular drug levels in spleen and bone marrow appears equally likely, since encapsulated Sb is active in both organs (17). Under any circumstances, organ-specific differences are also Sb specific, since amphotericin B, aminosidine, and miltefosine are active in all three target organs (28, 90, 106).

In most models in which immunologically intact mice (or hamsters) are used and optimal treatment is given, recrudescence of infection does not occur once the simultaneously developing Th1-cell immune response is fully expressed. However, successful posttreatment host defense responses can be undermined experimentally. Recurrent infection or late relapse can be provoked (i) by deficient T-cell number or function or absence of one particular endogenous cytokine (IFN-γ) after amphotericin B or miltefosine treatment (119, 130; H. Murray, unpublished observations) or (ii) as demonstrated in Sb-treated BALB/c mice with cutaneous L. major infection, by unremitting deactivation induced by an IL-4-driven, Th2-cell-associated response (131). There are as yet no satisfactory models of relapse after Sb therapy in visceral infection, since the animals (immunodeficient) likely to show recrudescence are also the ones which fail to respond initially to Sb (111, 119, 125, 128). However, both amphotericin B and miltefosine can be used to probe the mechanisms which prevent posttreatment relapse, since their initial efficacy is independent of the antileishmanial immune response (115, 119, 129, 130).

EXPERIMENTAL TREATMENT APPROACHES

The past decade's experimental efforts to identify new antileishmanial treatments has taken four basic directions, aimed at drug delivery, novel antiprotozoal agents or new applications of existing drugs, and immunotherapy alone or in combination with chemotherapy. Seeking better synergy, some work has also bridged more than one research area. This concluding discussion therefore focuses on (i) recent studies carried out with agents or formulations not yet tested in human kala-azar, (ii) results primarily derived models of visceral infection, and (iii) new treatments tested in vivo rather than in vitro.

Drug delivery systems.

As demonstrated >20 years ago (7, 136) and without any experimental doubt, encapsulating drug is effective in delivering treatment to parasitized macrophages within Leishmania-targeted organs. Liver, spleen, and bone marrow are particularly rich in sinusoidal and resident macrophages capable of phagocytizing circulating material; thus, encapsulation or carrier vehicles facilitate rapid, high-level tissue uptake and favor intracellular drug accumulation. The remarkable clinical efficacy and good tolerability of the lipid formulations of amphotericin B certainly attest to this approach, and well illustrate interdigitating benefits suggested by early experimental studies of targeted agents (7, 20, 32, 44, 136), including (i) use of lower total drug doses with comparable or greater efficacy, (ii) selective tissue uptake and reduced systemic toxicity, (iii) improved tolerability permitting higher daily doses and, in turn, short-course therapy, and (iv) likely persistence of drug in targeted tissues and/or within parasitized macrophages themselves.

The experimental appeal of encapsulation vehicles, including a spectrum of liposomes and vesicles (niosomes), remains undiminished and the goal of optimizing drug delivery continues to be revisited in models of visceral infection. In addition to Sb (32), paromomycin (aminosidine) (186), atovaquone (35), and IFN-γ (59, 78), which are agents currently in use or previously studied as free drug in human kala-azar (123), have been tested in encapsulated form. Liposomes have also been used to deliver other drugs alone (89, 98, 105) or in combination (59, 78). In virtually all instances, encapsulated drug is appreciably more active in vivo than free drug, usually by more than 5-fold and sometimes by more than 100-fold. Macrophages can also be targeted by other drug delivery techniques, including (i) receptor-mediated methods (39, 135, 146, 157), (ii) nanoparticle-bound techniques (55, 147), or (iii) by promoting drug aggregation to increase phagocytic cell uptake (140).

Delivery systems have been devised for immunostimulating cytokine genes as well. In visceral infection, IFN-γ and IL-12 genes can be transferred via liposomes (180) or transfected dendritic cells (1). In cutaneous infection, the IL-12 gene has been delivered by gene gun or an adenovirus vector (65, 149); single or multiple cytokine genes can also be transferred by modified Salmonella organisms administered orally (189).

NEW ANTILEISHMANIAL AGENTS OR APPLICATION OF EXISTING DRUGS

Agents directed at the parasite.

In addition to the preceding agents (tested in targeted and free form), review of the 1990–2000 literature indicates more than a dozen other agents with in vivo activity in experimentally infected animals (2, 3, 38, 47, 52, 63, 65, 66, 104, 149, 191). The selection of study drugs has been eclectic and largely empiric, and their broad array defies useful classification. However, rational or structure-based drug design has also begun to be applied (40, 155), with appropriate concern for validating putative therapeutic targets (19). Results for a number of the new antileishmanial compounds have been reported in just the past several years, making it difficult to judge their potential usefulness.

Combining Sb with aminosidine proved useful in human kala-azar (153); thus, combination chemotherapy also continues to be examined experimentally. When given with Sb, atovaquone (118), ketoconazole or metronidazole (65), or an iridoid glycoside plant extract (picroliv) (103) produce additive and possibly synergistic effects. The first three of these drugs may be potential candidates to combine with Sb since they have long been in clinical use and each is administered orally.

Agents directed at the macrophage.

In addition to cytokine treatment (e.g., IFN-γ or GM-CSF [112, 116]), in vivo parasite killing can also be induced by other approaches which appear to primarily target the macrophage. New lipopeptides (188), for example, may activate membrane-associated mechanisms similar to those of synthetic phospholipids such as miltefosine (129). Intracellular events, including signal transduction, have also been targeted in vitro and in vivo. Buthionine sulphoximine, an inhibitor of γ-glutamylcysteine synthetase, depletes gluthathione in parasitized macrophages, leading to enhanced cellular nitric oxide production, possibly increased H₂O₂ secretion, and parasite killing (85). Applied in vitro in L. donovani-infected macrophages and topically in mice with cutaneous L. major infection, imidazoquinolines (e.g., imiquimod) appear to directly activate macrophages and stimulate iNOS expression via signal transduction pathways (27). Imiquimod also shows broader immunomodulatory action, triggering Th1-cell-associated cytokine secretion (IL-12, IL-12-stimulated IFN-γ) and downregulating Th2-cell-associated cytokine secretion (184). Phosphotyrosine phosphatases, induced within macrophages by ingested L. donovani organisms, have also been identified as novel intracellular targets for chemotherapy (97, 132). These enzymes inhibit macrophage signaling pathways, including those which support cytokine induction of iNOS (97, 132). Treatment with peroxovanadium compounds, which inactivate these phosphatases, shows in vivo antileishmanial effects which are iNOS dependent and may also involve induction of Th1-cell-type cytokines (97).

Immunoenhancement.

Identifying agents which directly stimulate the macrophage to kill intracellular amastigotes represents one approach to immunomodulation. Of perhaps more physiologic interest, however, are interventions which induce (in the inert host) or free up (in the actively suppressed host) expression of the basic Th1-cell antileishmanial immune response, thus producing a broader range of effector mechanisms, including macrophage activation. The notion that immunoenhancement might stand alone as treatment in human visceral infection, as it can experimentally (1,78, 107, 112, 114, 116, 117), remains appealing and provides a rationale for defining the effects of immunoactivating agents used by themselves. However, from a practical standpoint, clinical experience with the only cytokine well-studied in kala-azar, IFN-γ (123), suggests that such experimental testing should also be carried out in combination with chemotherapy.

Stimulation of Th1-cell responses and/or induction of endogenous IFN-γ.

Irrespective of the stimulus used to induce or optimize Th1-cell activation in visceral infection, this state should generate a spectrum of effector cells (most importantly, leishmanicidal macrophages), unless overshadowed by a simultaneously triggered, deactivating mechanism (120, 188) (see below). Given required initiating and activating roles (57, 125, 152, 179), both IL-12 and IFN-γ continue to be viewed as primary therapeutic candidates either to be injected or induced endogenously in the parasitized host. The results of such treatment in established visceral infection may reflect acceleration of already developing Th1-cell events and/or actions of supraphysiologic cytokine levels, since by day 10 (and as early as day 3 [56]) after experimental challenge, most immunologically intact animals show evidence of IL-12 and/or IFN-γ expression (56, 99, 100, 102, 187).

In normal mice, IL-12- or IFN-γ-induced killing or control of visceral parasite replication can be achieved by treatment with (i) exogenous recombinant cytokine (112, 117, 179), (ii) cytokine gene transfer (1, 180), and (iii) inducers of endogenous IFN-γ (24, 114, 117). Similar experimental approaches in cutaneous L. major infection, worth testing in visceral disease models, include injecting (i) pertussis toxin to induce IL-12 (74), (ii) IL-18 to stimulate IFN-γ secretion (137), and (iii) synthetic oligodeoxynucleotides containing nonmethylated CpG dinucleotides to preferentially promote Th1-cell responses (185, 192).

T-cell costimulation.

CpG oligodeoxynucleotides may induce antileishmanial effects by triggering antigen-presenting cell–T-cell costimulatory pathways leading, in particular, to an IL-12-driven Th1-cell response (185, 192). Transfer of dendritic cells, pulsed ex vivo with specific antigen and transfected with the IL-12 gene (1), likely acts in a similar fashion. Evidence in L. donovani infection also clearly points to the therapeutic potential of enhancing the CD28-B7 costimulatory mechanism by a single injection of monoclonal antibody (MAb) directed at cytotoxic T-lymphocyte antigen-4 (CTLA-4), the negative-signaling component of this pathway (107). The finding that CTLA-4 engagement induces TGF-β, which in turn suppresses IFN-γ secretion and promotes L. chagasi replication (71), strengthens the rationale for testing anti-TGF-β treatment. In the L. major model, prophylactic costimulation, including engaging the CD40 ligand and CD40 pathway by anti-CD40 injection, promotes Th1-cell responses and control over infection (61). The effect of the oral agent tucaresol in visceral infection also probably reflects T-cell costimulation (161).

Inhibition of suppressive mechanisms.

Approaches to immunoenhancement may be derailed, however, if cytokine-mediated deactivating mechanisms are already in place. For example, once the polarized, IL-4-driven Th2-cell mechanism develops in L. major-infected BALB/c mice, these animals are rendered refractory to the activating effects of exogenous IL-12 and IFN-γ (143, 178). Nevertheless, despite ongoing L. major infection, Th1-cell responses can still be released and cure induced by other immunotherapeutic interventions: by injection of anti-TGF-β (93), anti-IL-4 (131), or CpG dinucleotides (193), or by sequentially deconstructing the basic Th2-cell mechanism and then injecting IL-12 (74).

Since a Th2-cell-type response, in which IL-10 figures most prominently, is also likely relevant in human kala-azar (12, 18, 51, 67, 68, 86, 168), testing similar inteventions as treatment in established, experimental visceral infection has appeal. However, in contrast to L. major infection in BALB/c mice (143), conventional models of visceral infection have not readily lent themselves to studying this mechanism's interaction with immunotherapy (or chemotherapy), thus making it more difficult to firmly identify suppressive targets for neutralization. For instance, (i) an early vigorous Th2-cell-type response is not a feature of infection induced by L. donovani, L. infantum, or L. chagasi in hamsters or in susceptible animals including BALB/c mice (4, 56, 81, 87, 99, 100, 102, 187); (ii) IL-4 KO mice are not more resistant (151); and (iii) in certain models, IL-4 and/or IL-10 are either not provoked (87) or are induced in a seemingly inconsequential fashion in normal mice since resistance is ultimately acquired (56, 102, 187, 188). Nevertheless, progressive infection in hamsters and, to some extent, initial visceral replication in normal BALB/c mice do correlate with the presence of IL-10 (99, 100, 187), and BALB/c IL-10 KO mice control L. donovani organisms in a strikingly accelerated fashion (H. Murray and R. Coffman, unpublished observations).

Two models of visceral L. donovani infection, however, have been specifically developed for testing treatment in a predominant Th2-cell-type environment. BALB/c mice, preimmunized with heat-killed L. major promastigotes, cross-react to challenge with viable L. donovani with a response mediated by both IL-4 and IL-10, which inhibits acquired resistance and induces the noncure phenotype (120). In these mice, treatment with exogenous IL-12 and IFN-γ readily induces leishmanicidal activity (120), contrasting directly with observations made in the L. major-BALB/c mouse-Th2-cell response model (143, 178).

The preceding L. donovani model (effects related to IL-4 plus IL-10 [120]) and a second model of progressive visceral infection in IL-10 transgenic BALB/c mice (effects related to sustained IL-10 alone) have also been used to gauge whether a Th2-cell-type response impairs the efficacy of chemotherapy. In both groups of mice, the initial leishmanicidal response to Sb treatment is entirely preserved (H. Murray and R. Coffman, unpublished observations). Thus, despite a disease-exacerbating Th2-cell-type response, neither the efficacy of Th1 cytokine treatment nor chemotherapy is overtly influenced during the time treatment is given. These experimental results, then, suggest no major endogenous obstacle to cytokine-based, combination immunochemotherapy in visceral infection. At the same time, whether neutralizing suppressive factors may still enhance the extent of initial treatment efficacy and whether such factors may influence the durability of the posttreatment response remain open questions.

Immunochemotherapy and future approaches.

The experimental notion that immunostimulation could be joined with chemotherapy to produce an improved effect in visceral infection took root in the mid-1980s (73, 110) and has since moved in two directions, combining antileishmanial drug therapy (i) with agents which enhance the Th1-cell-associated response and/or directly stimulate macrophages or (ii) with interventions aimed at extinguishing the effects of suppressive mechanisms. Both of these approaches might even be combined in a sequential regimen of first neutralizing deactivation, then providing Th1-cell stimulation, and finally administering chemotherapy. Together, such strategies reflect the sense by investigators in this field that there is an immunologic mechanism, which once unlocked will open the door to optimal initial and long-term responses to antileishmanial therapy.

Thus far, the one approach to proceed through experimental study to reach clinical testing has been the use of IFN-γ plus Sb treatment (123). However, when combined with Sb, experimental effects in visceral infection can also be induced by other cytokines (IL-12 [125]), cytokine inducers (24), and agents which appear to target the macrophage (70). Other drugs have also been successfully combined with IFN-γ (89), and even though amphotericin B acts independently of host T-cell mechanisms (115), its visceral antileishmanial efficacy can readily be enhanced by cotreatment with IL-12 (H. Murray, unpublished observations).

Successful manipulation of the deactivating Th2-cell-associated mechanism, relevant as an immunotherapeutic strategy in visceral infection, has shown clear-cut benefits in L. major-infected BALB/c mice treated with Sb plus either anti-IL-4 MAb or intralesional IL-12 (131). The net effect of this treatment was prevention of progressive infection after Sb treatment was discontinued and eventual resolution. Sb's initial antileishmanial activity likely helped to partially curtail ongoing Th2-cell action by reducing the antigenic stimulus of high parasite load; parallel treatment with anti-IL-4 or IL-12 presumaby reduced Th2-cell effects and either released the deactivated Th1-cell response or directly promoted it (131). The actions of endogenous IL-10 have been similarly targeted by using a single injection of anti-IL-10 receptor MAb in normal BALB/c mice with established L. donovani infection. Preliminary studies indicate that this approach enhances the efficacy of subsequently administered Sb (H. Murray and R. Coffman, unpublished observations). In view of the apparent suppressive role of IL-10 in human kala-azar (reviewed above), this effect raises the possibility of a transient IL-10 receptor blockade as an intervention to improve chemotherapy's overall efficacy and/or to reduce the required dose or duration of drug treatment.

There are additional immunotherapeutic strategies also currently being tested experimentally or worth considering in the future for combination with chemotherapy: inducing or enhancing endogenous IL-12 (65, 74, 149) or IFN-γ (1, 65, 114, 137, 149, 180); triggering T-cell costimulatory pathways to strengthen the Th1-cell response (1, 61, 107, 185, 192); injecting anti-TGF-β (71, 93); using indomethacin (50, 60, 109) or cyclooxygenase-2 inhibitors (L. donovani infection triggers cyclooxygenase-2 expression [H. Murray and A. Dannenberg, unpublished observations]) to reverse prostaglandin-mediated upregulation of IL-10 and downregulation of IL-12 (163); and deactivating suppressive intracellular pathways within the macrophage itself (97).

A final immunochemotherapeutic strategy, also cytokine based, relates to granuloma remodeling: increasing the number and size of tissue granulomas by exogenous IL-1 treatment (48) or delivering selected effector and responder cells to parasitized foci. Depending upon the cell population selected for delivery, experimental results in established visceral infection indicate that exogenous treatment with IL-2, GM-CSF, or granulocyte-CSF can encase parasitized macrophages within the developing L. donovani-induced granuloma with activated mononuclear cells, myelomonocytic cells, or granulocytes, respectively (114, 116). The appeal of testing antileishmanial chemotherapy in the presence of enhanced granuloma assembly, perhaps induced in the future by cotreatment with specific chemokines such as IFN-γ-inducible protein 10 (41, 95), is also clear.