Spatiotemporal antimalarial drug resistance saturation and emerging artemisinin tolerance in western Kenya during a pilot multiple first-line treatment study.

Emergence of partial resistance to artemisinin-based combination therapies (ACTs) threatens malaria control globally. This pilot multiple first-line treatment (MFT) for uncomplicated malaria study was conducted across three Western Kenya sites namely, Homa Bay County Mainland (intervention sites), Mfangano Island in Homa Bay County (intervention sites) and Migori County (control sites). The MFTs used were artemether plus lumefantrine (AL), dihydroartemisinin plus piperaquine (DHA + PIP), artesunate plus amodiaquine (AS + AQ) and artesunate plus pyronaridine (AS + PYD). The study was a serial cross-sectional analysis (September 2020–September 2022, extended to January 2024 for Mfangano Island) to characterize spatiotemporal resistance patterns following MFT deployment thus: (a) AL at baseline, then DHA + PIP for 8 months, followed by AS + AQ for 8 months, then AL for 8 months in Homa Bay County Mainland; (b) AL at baseline followed by AS + PYD for 36 months in Mfangano Island; (c) AL for 24 months in Migori County. Dried blood spots (n = 310) from malaria patients were analyzed for resistance markers (Pfdhfr, Pfdhps, Pfmdr1, Pfk13) using targeted amplicon deep sequencing. Deep sequencing generated 9,288,278 reads with mean complexity of infection 2.62 and 63% showing polyclonal infections. Core dhfr markers (N51I, S108N) showed saturation (100% non-wildtype), with emerging I164L detected at 12.2%. Dominant IRNI haplotype occurred in 93.7% of samples. For dhps, A437G and K540E reached fixation (100%), while S436H showed 33.4% prevalence. The mdr1 Y184F mutation demonstrated rapid temporal escalation in Homa Bay County Mainland (22.9% to 56.2%, P = 0.005). WHO-validated artemisinin resistance marker k13 A675V was detected (0.5%) and A578S, a k13 polymorphism not yet WHO-validated as a marker of artemisinin partial resistance, occurred in 7.8% of samples. Machine learning models achieved > 90% accuracy with genotype showing strongest predictive power. These findings suggest antimalarial resistance saturation in Western Kenya with emerging artemisinin resistance markers, indicating potential need for enhanced surveillance and alternative treatment strategies.