Systematic Review of Biomarkers To Monitor Therapeutic Response in Leishmaniasis

Potential biomarkers for VL, CL, and PKDL were identified by a primary-literature search performed using PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, querying the PubMed database, restricted to the English language, as follows: “(((Leishmaniasis[title]) or Kala-azar[title]) or PKDL[title]) and (((((((((biomarker) or biomarkers) or marker) or markers) or level) or levels) or concentration) or activity) or profile).” This electronic search was performed from November 2013 to August 2014, and the date last searched was 19 August 2014. Results were screened manually to identify relevant publications based on title and/or abstract. Publications that did not focus on the identification or evaluation of biomarkers were excluded. Selected publications were then evaluated according to the exclusion criteria as described in Table 1. If the abstract did not clearly indicate whether a study met the initial inclusion criteria, the entire publication was assessed. Secondary literature was subsequently identified using references from the identified primary literature and related publications on PubMed and by specifically querying PubMed using the term of the identified biomarker in combination with “Leishmaniasis” and/or “Kala-azar.”

The biological and clinical pharmacodynamic potential of biomarkers was evaluated based on five criteria: (i) time to normalcy, i.e., the time needed for the biomarker level to regress to healthy/control levels; (ii) specificity, in relation to concomitant (infectious) diseases, such as malaria and HIV; (iii) sensitivity, the marker's quantitation in (treated) patients compared to that in healthy controls and its association with treatment cure or failure; (iv) additional sensitivity, i.e., further assessment of sensitivity by more in-depth association of the marker's quantitation to standard clinical markers of disease, such as spleen and lesion size; and (v) geographical applicability. Biomarkers were given a score (−/+/++/?) for each criterion as further explained in Table 2.

Assessing the viable parasite load within a patient is probably the most direct marker of disease status for leishmaniasis, and assessing the reduction of the viable parasite biomass would allow for exact monitoring of the therapeutic response. Several target genes have been identified and used for the molecular identification and quantification of Leishmania in clinical samples, including kinetoplast DNA (kDNA, both mini- and maxicircles), small-subunit (SSU) RNA, such as 18S rRNA, and 7SL RNA. For VL patients, the measurement of the Leishmania parasite load in blood using quantitative PCR (qPCR) has been evaluated mainly for diagnosis but also as a proxy value of the overall parasite load and clinical response during and after treatment (14–26). The parasite load in blood rapidly decreases upon initiation of treatment, in parallel with clinical improvement (14–17). qPCR of blood of East African VL patients reflected differences in treatment responses to different AmBisome dosages (27); however, the sensitivity of the assay was lower than for Indian VL patients (28).

For CL, the parasite burden is localized and confined to the upper layer of the dermis, in which it is probably homogenously spread in the inflammatory zone that surrounds the necrotic ulcer (29). Confirmation of parasites by microscopy or, if available, PCR-based techniques from lesion biopsy specimens or scrapings is currently the diagnostic practice for CL (30–36). Quantitation of parasite RNA in repeated lesion biopsy specimens has been demonstrated as a technique to assess the parasite burden in CL lesions (35, 37). Treatment response was quantified in CL patients, demonstrating declines in Leishmania major parasite loads of ∼1 log/week after initiation of miltefosine treatment, which paralleled clinical improvement (29). Swabbing of lesions, which is less invasive than biopsy, was performed to determine whether parasite DNA/RNA loads were diagnostic for CL, and the sensitivity was around 90% (38–40). The pharmacodynamic use of repeated swabbing has not yet been reported. Interestingly, the presence of parasites in CL has also been shown at (unaffected) extralesional sites (38, 40), opening up other possibilities for less invasive sampling procedures. For PKDL, Leishmania DNA was also detected in lesion material before treatment; a significantly higher parasite burden was found in chronic lesions than in transient lesions, with burdens reduced to nondetectable levels posttreatment (17). The pharmacodynamic use of newer molecular tools (e.g., loop-mediated isothermal amplification [LAMP]) (41, 42) has not yet been investigated.

Disease-specific antigen detection is regularly used as a predictive biomarker, e.g., for various cancer types (43), and is potentially useful for infectious diseases as well. For leishmaniasis, however, the application of antigen tests has been limited mainly to diagnostics making use of a urine-based latex agglutination assay (KAtex), which detects a heat-stable low-molecular-weight carbohydrate antigen found in the urine of VL patients (26, 44–46). The method has been successfully evaluated and compared to other methods for diagnosis of VL patients in various geographical areas, ranging from East Africa to South Asia (3, 26, 47–55). Even though specificity was consistently high (98% to 100%) in these studies, sensitivity appeared to be very low to moderate (48% to 95.2%), with a high discrepancy between studies. Studies from India and Sudan indicated that the urine antigen detection test became negative in cured VL patients at least 30 days after the end of treatment (48, 49), indicating a possible pharmacodynamic use of this assay.

Leishmania parasites reside and replicate inside the phagocytic cells of the reticuloendothelial system, mainly macrophages, increasing the overall macrophage biomass in the host. Since the macrophage load also decreases again with waning parasitic infection, soluble macrophage-related markers—specifically when produced by infected macrophages—are potential semidirect biomarkers. Neopterin is a heterocyclic pteridine compound which is synthesized by macrophages after gamma interferon (IFN-γ) stimulation (56). It is considered an indicator of activation of cellular immunity. Increased neopterin production is found in a broad range of diseases, e.g., in viral infections (HIV, hepatitis B and C) and infections due to intracellular bacteria (tuberculosis, malaria). Serum neopterin concentrations were elevated in VL patients compared to levels in controls and decreased to normal concentrations at the end of treatment in cured patients, whereas they were still significantly increased in refractory patients (57). Serum neopterin concentrations were not found to be elevated in CL patients (58).

Adenosine deaminase (ADA), found particularly in macrophages and lymphocytes, is a key enzyme in the breakdown of adenosine, a purine nucleoside that suppresses the inflammatory responses. For VL, serum ADA activity was increased at diagnosis and returned to almost normal concentrations at the end of therapy (day 30) in Nepalese and Indian patients (59–62). At diagnosis, activity appeared higher in VL patients than in malaria, leprosy, or tuberculosis patients (60). Also, in Turkish CL patients, adenosine deaminase was increased at the time of diagnosis (63).

Recovery from VL is linked mainly to the CD4⁺ T-cell-mediated cellular immune response. More specifically, the Th1-mediated response is generally associated with macrophage activation, host resistance, and protection against Leishmania parasites, leading to control of disease. Conversely, the Th2-mediated response is associated with downregulation of macrophage activation and eventually progression of disease. Unfortunately, this distinction between Th1 and Th2 activation is a simplified model, and many patients demonstrate a nonspecific immune response profile.

The most studied cytokines are the proinflammatory cytokines IFN-γ and tumor necrosis factor alpha (TNF-α) and the regulatory cytokine interleukin-10 (IL-10). Plasma IL-10 was found to be increased in Indian patients with active VL and could be detected in the keratinocytes and sweat glands of patients who eventually developed PKDL (64). The increase of IL-10 concentrations in VL patients was later confirmed in a range of countries, including, among others, India, Brazil, and Ethiopia (65–77). IL-10 levels were found to drop significantly after successful treatment (66, 70, 73, 78) to near-control levels 5 to 7 days posttreatment (74). Ansari et al. found no difference in pretreatment IL-10 levels between responsive and unresponsive patients (74). For CL, IL-10 might be a possible pharmacodynamic marker indicating treatment failure, as IL-10 mRNA levels in lesion biopsy specimens were found to be positively associated with unresponsiveness to treatment (79, 80). Cured mucocutaneous leishmaniasis (MCL) patients demonstrated a higher percentage of IL-10-expressing cells pretreatment than relapsing patients (81). Interestingly, IL-10 was found to be positively correlated with the parasite load in the blood of VL patients (17, 70) and lesional tissue of PKDL patients (82).

TNF-α is a cytokine produced mainly by activated macrophages. TNF-α levels were found to be significantly increased in patients with active VL (77, 83–85); they declined during treatment (85–89) and returned to healthy-control levels at the end of treatment (84). Unresponsive patients retained elevated levels of TNF-α (85). In contrast, other studies found minimal levels of this cytokine in Indian VL patients (74, 90, 91). Moreover, TNF-α was also present in asymptomatic VL patients (83), complicating the interpretation of TNF-α in cases of VL. For CL patients, studies of TNF-α serum levels are contradictory; some studies found elevated levels of TNF-α in the plasma of CL patients that decreased posttreatment compared to healthy controls (92–95), but others could confirm this only for MCL patients (85, 96). TNF-α mRNA levels in lesion biopsy specimens were found to be positively correlated with lesion size (80).

IFN-γ is a critical soluble cytokine for innate and adaptive immunity against intracellular infections and is involved in the activation of macrophages. IFN-γ levels were found to be significantly elevated in patients with active VL, which was confirmed in a wide range of countries: India (74, 75), Bangladesh (70), Brazil (70, 72, 73, 77), Ethiopia (67, 68), Sicily (66), and Iran (65). During and after successful treatment, IFN-γ levels were found to drop significantly but remained elevated compared to levels in healthy controls (66, 70, 73, 75, 78). In contrast, Cenini et al. (84) showed that IFN-γ levels returned to healthy-control levels at the end of treatment. Moreover, IFN-γ plasma concentrations appeared to be significantly higher after the end of treatment in patients unresponsive to therapy than in responsive patients treated with sodium stibogluconate (SSG) (74, 75). Discrepant results in asymptomatic VL patients indicated that IFN-γ was elevated in 48% of asymptomatic Brazilian patients but that it was undetectable in the vast majority of asymptomatic Ethiopian patients (67, 83). Additionally, a recent study of Brazilian pediatric VL patients showed that low levels of IFN-γ were associated with signs of severity, such as jaundice or hemorrhage (97). In CL lesion biopsy specimens, no significant difference in IFN-γ levels could be found between patients with favorable and unfavorable lesion evolutions (79).

For PKDL patients, the expression of the mRNA of the three cytokines IL-10, IFN-γ, and TNF-α in lesions was found to be significantly elevated compared to that in control tissues (74, 82). After treatment, these levels were restored to near-control levels (74). Ganguly et al. found that IL-10 and IFN-γ levels were significantly higher in patients with polymorphic PKDL than in patients with macular PKDL (98).

Concerning patients with HIV-VL coinfection, only TNF-α and IFN-γ serum levels were still significantly elevated in HIV patients when they developed VL coinfection, while IL-10 levels tended to decrease (99). Also, compared to Chagas disease, dengue fever, and tuberculosis patients, leishmaniasis patients showed high levels of TNF-α (70). TNF-α and IFN-γ levels increased significantly when malaria patients developed a VL coinfection (100).

The interleukins IL-6 (74, 75, 77, 84, 101) and IL-12 (65, 67, 69, 70) (often measured as the concentration of the subunit IL-12p40) were also found to be significantly increased in the sera of VL patients. In Sudanese and Ethiopian VL patients, IL-6 returned to normal concentrations within the treatment period (84, 101) and seemed indicative of relapse events (101). However, IL-6 was not correlated with spleen/liver size (73). Also IL-12 levels were found to drop significantly within 30 days of treatment (73) and was largely absent in cured and asymptomatic cases (67, 69). In contrast, in Bangladesh and Brazil, IL-12 was shown to be elevated in asymptomatic VL cases (83, 90). Both interleukins also showed pharmacodynamic potential in CL patients. IL-12 was correlated with unfavorable lesion evolution and lesion duration (80, 102). IL-6 mRNA from biopsy specimens was correlated with lesion size, and also IL-6 serum concentrations were found to be elevated in CL patients (80, 94).

IL-18 was also increased in patients with active VL compared to levels in healthy controls (67). Interestingly, a significant decrease in urinary IL-18 levels was detected during treatment (103). Urinary detection of biomarkers would be ideal due to its noninvasive collection method.

Soluble CD40 ligand (sCD40L) (also called sCD154) was significantly decreased in VL patients at diagnosis compared to levels in controls in areas of endemicity (70, 87). During treatment, sCD40L levels increased toward healthy control levels. However, similar CD40L levels were found for Chagas disease and VL patients, which might cause specificity issues (87).

Levels of circulating soluble cytokine receptors for IL-2 and IL-4 (sIL-2R and sIL-4R, respectively) were elevated in patients with active VL, with higher concentrations of sIL4R than in patients with other local and systemic parasitic infections (57, 104–106). Serum sIL-2R concentrations correlated with Leishmania DNA serum levels (70) and significantly decreased during treatment (57, 70) but returned to normal only after several months (105). At the start of treatment, sIL-2R levels were also significantly higher in patients developing PKDL than in patients not developing PKDL (64). Additionally, mRNA levels of the IL-2R α-chain were significantly elevated in lesions of PKDL patients before treatment and returned to control levels posttreatment (82).

Circulating sCD4 and sCD8 were increased at the start of treatment and returned to levels comparable to those in healthy controls within several months after treatment (57, 105). sCD8 was significantly decreased posttreatment in responders to therapy compared to levels in nonresponders, making it a possible suitable pharmacodynamic marker (57).

Serum levels of the soluble receptors for TNF (sTNFRs) were significantly elevated in patients with active VL compared to levels in controls in areas of endemicity and nonendemicity (91). Responding patients showed a steep decrease in sTNFR levels already at day 15 during treatment, in contrast to nonresponders (86, 91).

Acute-phase proteins widely used as clinical markers of inflammation and infection, which increase during many (non)infective inflammatory diseases and malignancies, also increase during VL. C-reactive protein (CRP), serum amyloid A protein (SAA), and alpha-1-acid glycoprotein (AGP) were increased in Kenyan VL patients upon diagnosis and reached normal levels before or at the end of treatment (SAA and AGP) or at 60 days posttreatment (CRP) (107). Elevation of CRP levels was confirmed for Indian patients with active VL (75). Interestingly, pretreatment CRP levels were lower in patients responding fast to treatment than in slow-responders, with lower splenic parasite counts (107), which was confirmed in a large Indian pediatric VL cohort (108). An increased pretreatment CRP concentration in VL patients was associated with the development of PKDL (109), while CRP levels were not significantly elevated in PKDL patients. However, the specificity of acute-phase proteins in the monitoring of VL treatments is probably low, as they are increased in a myriad of other infectious and noninfectious inflammatory illnesses.

Arginase catalyzes the metabolism of l-arginine into l-ornithine and urea. The resulting diminishing bioavailability of l-arginine is regarded as a potent mechanism of immune suppression and impairment of T-cell responses. In patients with active VL and CL, arginase activity in plasma was found to be significantly increased, and levels returned to control levels for VL patients during SSG treatment (110, 111). VL-HIV coinfection patients appeared to have increased arginase activity compared to VL patients, both in plasma and peripheral blood mononuclear cells (PBMCs) (112). In PKDL patients, arginase activity declined after miltefosine treatment but not after SSG treatment (113).

All of the current first-line diagnostic serological tests for VL are antibody detection tests (114, 115). Two serological tests are currently being used in the field: the direct agglutination test (DAT), based on numbers of killed whole L. donovani promastigotes, and the recombinant K39 (rK39)-based immunochromatographic antibody detection test. Other antigen-based assays have been developed for Leishmania antibody detection, using (recombinant) proteins rK9, rK26, rK28, Leishmania infantum cytosolic tryparedoxin peroxidase (LicTXNPx), rgp63, rLepp12, recombinant open reading frame F (rORFF), BHUP2, rKLO8, rHSP70, guanylate binding protein (GBP), galactosyl-α(1-3)galactose, 9-O-acetylated sialic acids, recombinant peroxidoxin, and amastin (116–130). Unfortunately, antibodies against Leishmania parasites exhibit a long half-life and stay detectable for several months up to several years after an infection [tested by the DAT and for galactosyl-α(1-3)galactose, LicTXNPx, rK26, rK39, and BHUP2] (49, 120, 121, 131–141), which compromises the diagnosis of a relapse case and also the pharmacodynamic application of these markers. However, it was found that for some antibodies (against the recombinant Leishmania antigens rH2A, KMP11, the “Q” protein, and 9-O-acetylated sialic acids), the levels do decrease significantly 30 to 60 days after treatment (129, 142). Furthermore, 1 week posttreatment, only ∼40 to 50% of patients gave a positive signal for rLepp12, compared to 100% for rK39 and for direct agglutination (125). Though not very sensitive (44%), Leishmania-specific immunoglobulin E (L-IgE) has been suggested to be a specific (98.3%) marker of active VL disease (L. chagasi), although it is undetectable at subclinical levels in VL patients, Chagas disease patients, and healthy controls (143–145). Moreover, increased L-IgE concentrations were demonstrated to regress to normal values during the time span of treatment (143, 145, 146). In cases of atypical cutaneous leishmaniasis, IgE levels were not significantly different from those of asymptomatic or healthy controls (147). Anam et al. (144) also hinted at a possible (diagnostic) role for L. donovani-specific IgG3 antibody isotype detection. While the IgG3 antibody level decreases significantly posttreatment (148, 149), the pharmacodynamic value of this marker is probably very low, as the time to normalcy for IgGs is longer than 3 months for both CL and VL patients (150–154).

Besides the drawback of the long half-lives of antibodies, antibody detection tests tend to be positive in a significant proportion of noninfected or otherwise asymptomatic individuals living in areas where VL is endemic (135, 155, 156). Due to these crucial limitations in the use of antibodies to monitor therapies, these markers are excluded from Table 3.

Recently, the quantitative application of molecular parasite detection methods as a pharmacodynamic measure was demonstrated, both in VL and CL. While this method measures the parasite directly and therefore is theoretically the most promising biomarker, there are some issues. First, the sensitivity of this marker for VL is relatively low (∼80%) and seems to vary between geographical regions (27, 28). The parasite loads appear to decrease with clinical cure but are undetectable before clinical cure can be established. The parasites reside within the spleen, bone marrow, and liver, and plasma is therefore only a proxy reservoir of the parasite. Additionally, it remains unknown what the predictive value is of blood parasite loads in relapsing patients and controls in areas of endemicity. Last, it is unsuitable for routine monitoring due to its high costs (considering the ∼€30/sample material costs and the required state-of-the-art laboratory equipment and trained technicians, this tool can be used only in a clinical trial setting).

Another direct biomarker with potential is the Leishmania carbohydrate antigen, which forms the basis for the diagnostic KAtex test. One of the biggest advantages of this biomarker is that it can be detected in urine. Its specificity is consistently high, but its sensitivity appears variable, which may make it suitable only in controlled settings. The Leishmania-specific antigen can be assessed quantitatively by enzyme-linked immunosorbent assay (ELISA), which makes it easier to adopt at primary health care facilities than the molecular detection methods.

Of the indirect biomarkers, the most promising are the macrophage-related markers, as these are directly related to parasitic infection of macrophages. ADA activity is increased in patients with VL and CL, returns to normal during treatment, and shows promising results in patients with HIV-VL coinfection. Unfortunately, this marker has no proven geographical applicability, and there are no data on the relation between ADA activity levels and clinical outcome.

Though most cytokines demonstrated a lack of specificity, a range of them showed promising results with regard to the other evaluation criteria. IL-10 correlated with the parasite load at the time of diagnosis, decreased during treatment, and was even associated with the occurrence of PKDL. However, IL-10 was increased as well in subclinical cases, which complicates its interpretation, certainly in the context of parasite recrudescence. Although studies from different regions contradict each other on its sensitivity, TNF-α shows the highest specificity in comparison to other cytokines, indicating that it might be applicable as a biomarker in certain regions of endemicity. Levels of other indirect markers (e.g., sTNFR, IL-6) appeared predictive for clinical outcome but require further evaluation with regard to the other criteria for us to be able to draw conclusions on their potential. A practical advantage of cytokines is that their ELISA kits are relatively low in cost and may be implemented on a large scale, even though a basic laboratory is still required.