Estimating the probable cause of recurrence in Plasmodium vivax malaria: relapse, reinfection or recrudescence?

Abstract

BackgroundRelapse of Plasmodium vivax infection is the main cause of vivax malaria in many parts of Asia. However at the individual patient level, recurrence of a blood stage infection following treatment within the endemic area can be either a relapse (from dormant liver-stage parasites), a recrudescence (blood-stage treatment failure), or a reinfection (following a new mosquito inoculation). Each requires a different prevention strategy, but previously they could not be distinguished reliably. Time-of-event and genetic data provide complimentary information about the cause of P. vivax recurrence, but the optimum approach to genotyping and analysis remains uncertain. MethodsIndividual-level data from two large drug trials in acute vivax malaria patients (Vivax History: VHX; Best Primaquine Dose: BPD) conducted on the Thailand-Myanmar border with follow-up of one year were pooled (n=1299). A total of 710 isolates from both acute and recurrent P. vivax episodes were genotyped using 3-9 highly polymorphic microsatellite markers. These pooled data were analyzed using a novel population statistical model incorporating an assessment of genetic relatedness, treatment drug administered, and the time-to-recurrence. Results99% of genotyped recurrences in individuals who did not receive primaquine (n=365) were estimated to be relapses. In comparison, 14% of genotyped recurrences (n=121) were estimated to be relapses following high-dose supervised primaquine. By comparing episodes across individuals (90194 comparisons), the false-positive rate of relapse December 23, 2018 1/43 identification using genetic data alone was estimated to be 2.2%. We estimated the true failure rate after high-dose primaquine (7mg/kg total dose) to be 2.6% in this epidemiological context, substantially lower the reinfection unadjusted estimate of 12%. Simulation studies show that 9 highly polymorphic microsatellite markers suffice to discriminate between recurrence states. Drug exposures reflected by plasma carboxy-primaquine concentrations were not predictive of treatment failure, but did identify non-adherence. ConclusionUsing this novel statistical model, relapse of P. vivax malaria could be distinguished reliably from reinfection. This showed that in this population supervised high-dose primaquine could avert up to 99% of relapses. In low transmission settings, microsatellite genotyping combined with time-to-event data can accurately discriminate between the different causes of recurrent P. vivax malaria. Author summaryOne hundred years ago, Plasmodium vivax, the most globally diverse cause of human malaria, was present across most of the old-world, throughout tropical and temperate climes. Its main evolutionary advantage over Plasmodium falciparum, responsible for the most deadly type of human malaria, is its ability to stay dormant in the liver, emerging weeks to years later causing recurring illness and continuing transmission. The dormant liver-stage parasites are called hypnozoites. A recurrent infection can either be hypnozoite-derived (...

Estimating Relatedness Between Malaria Parasites

“…In this study, we analyze samples genotyped at nine MSs with no evidence of polyclonality (detection of two or more alleles at one or more MS) from n=204 people, selecting one episode per person uniformly at random from all episodes per person. We use allele frequencies reported in Taylor et al (2018). They have average expected homozygosity trueh¯mmax=0.10 and effective cardinality (defined below, Equation 7) averaged over mmax=9 MSs of K¯′mmax=13.03.…”

Estimating relatedness between malaria parasites

“…In addition to the aforementioned P. falciparum data, we generated results for a single P. vivax data set, freely available online [54]. The P. vivax data were collected between 2010 and 2014 from r m , allele frequencies, (f 1 (g)) g∈G1 , .…”

“…We estimate relatedness between pairs of samples from n = 204 different people, selecting one episode per person uniformly at random from all episodes per person. We use the allele frequencies reported in [54]. They have average expected homozygosityh mmax = 0.10 (equation (3.4)) and average effective cardinality (defined below, equation (3.8))K mmax = 13.03.…”

Efficacy of Primaquine in Preventing Short- and Long-Latency Plasmodium vivax Relapses in Nepal

“…Furthermore, in homologous recurrences, relapse cannot be distinguished with certainty from recrudescences [15, 17]. Using time-to-event information can strengthen the probabilistic discrimination between these states [39]. In this study, recrudescence was considered very unlikely given the low level of CQ resistance.…”