Miltefosine has been used successfully to treat visceral leishmaniasis (VL) in India, but it was unsuccessful for VL in a clinical trial in Brazil. To identify molecular markers that predict VL treatment failure whole genome sequencing of 26 L. infantum isolates, from cured and relapsed patients allowed a GWAS analysis of SNPs, gene and chromosome copy number variations. A strong association was identified (p = 0·0005) between the presence of a genetically stable L. infantumMiltefosine Sensitivity Locus (MSL), and a positive response to miltefosine treatment. The risk of treatment failure increased 9·4-fold (95% CI 2·11-53·54) when an isolate did not have the MSL. The complete absence of the MSL predicted miltefosine failure with 0·92 (95% CI 0·65-0·996) sensitivity and 0·78 (95% CI 0·52-0·92) specificity. A genotyping survey of L. infantum (n = 157) showed that the frequency of MSL varies in a cline from 95% in North East Brazil to <5% in the South East. The MSL was found in the genomes of all L. infantum and L. donovani sequenced isolates from the Old World (n = 671), where miltefosine can have a cure rate higher than 93%. Knowledge on the presence or absence of the MSL in L. infantum will allow stratification of patients prior to treatment, helping to establish better therapeutic strategies for VL treatment. FUND: CNPq, FAPES, GCRF MRC and Wellcome Trust.

In India, visceral leishmaniasis (VL) caused by Leishmania donovani has been successfully treated with miltefosine with a cure rate of > 90%. To assess the efficacy and safety of oral miltefosine against Brazilian VL, which is caused by Leishmania infantum, a phase II, open-label, dose-escalation study of oral miltefosine was conducted in children (aged 2–12 years) and adolescent-adults (aged 13–60 years). Definitive cure was assessed at a 6-month follow-up visit. The cure rate was only 42% (6 of 14 patients) with a recommended treatment of 28 days and 68% (19 of 28 patients) with an extended treatment of 42 days. The in vitro miltefosine susceptibility profile of intracellular amastigote stages of the pretreatment isolates, from cured and relapsed patients, showed a positive correlation with the clinical outcome. The IC₅₀ mean (SEM) of eventual cures was 5.1 (0.4) µM, whereas that of eventual failures was 12.8 (1.9) µM (P = 0.0002). An IC₅₀ above 8.0 µM predicts failure with 82% sensitivity and 100% specificity. The finding of L. infantum amastigotes resistant to miltefosine in isolates from patients who eventually failed treatment strongly suggests natural resistance to this drug, as miltefosine had never been used in Brazil before this trial was carried out.

Miltefosine treatment failure in visceral leishmaniasis in Brazil has been associated with deletion of the miltefosine susceptibility locus (MSL) in Leishmania infantum. The MSL comprises four genes, 3′-nucleotidase/nucleases (NUC1 and NUC2); helicase-like protein (HLP); and 3,2-trans-enoyl-CoA isomerase (TEI). In this study CRISPR-Cas9 was used to either epitope tag or delete NUC1, NUC2, HLP and TEI, to investigate their role in miltefosine resistance mechanisms. Additionally, miltefosine transporter genes and miltefosine-mediated reactive oxygen species homeostasis were assessed in 26 L. infantum clinical isolates. A comparative lipidomic analysis was also performed to investigate the molecular basis of miltefosine resistance. Deletion of both NUC1, NUC2 from the MSL was associated with a significant decrease in miltefosine susceptibility, which was restored after re-expression. Metabolomic analysis of parasites lacking the MSL or NUC1 and NUC2 identified an increase in the parasite lipid content, including ergosterol; these lipids may contribute to miltefosine resistance by binding the drug in the membrane. Parasites lacking the MSL are more resistant to lipid metabolism perturbation caused by miltefosine and NUC1 and NUC2 are involved in this pathway. Additionally, L. infantum parasites lacking the MSL isolated from patients who relapsed after miltefosine treatment were found to modulate nitric oxide accumulation in host macrophages. Altogether, these data indicate that multifactorial mechanisms are involved in natural resistance to miltefosine in L. infantum and that the absence of the 3'nucleotidase/nuclease genes NUC1 and NUC2 contributes to the phenotype. 10.13039/501100000265 MRC GCRF and 10.13039/501100006182 FAPES .