(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Methaemoglobin as a surrogate marker of primaquine antihypnozoite activity in Plasmodium vivax malaria: A systematic review and individual patient data meta-analysis [1]

['Ihsan Fadilah', 'Oxford University Clinical Research Unit Indonesia', 'Faculty Of Medicine Universitas Indonesia', 'Jakarta', 'Centre For Tropical Medicine', 'Global Health', 'Nuffield Department Of Medicine', 'University Of Oxford', 'Oxford', 'United Kingdom']

Date: 2024-10

For a given primaquine regimen, higher methaemoglobin on day 7 was associated with a reduced risk of P. vivax recurrence. Under our proposed causal model, this justifies the use of methaemoglobin as a population-level surrogate endpoint for primaquine antihypnozoite activity in patients with P. vivax malaria who have normal G6PD activity.

We identified 219 P. vivax efficacy studies, of which 8 provided relevant individual-level data from patients treated with primaquine; all were randomised, parallel arm clinical trials assessed as having low or moderate risk of bias. In the primary analysis data set, there were 1,747 patients with normal glucose-6-phosphate dehydrogenase (G6PD) activity enrolled from 24 study sites across 8 different countries (Indonesia, Brazil, Vietnam, Thailand, Peru, Colombia, Ethiopia, and India). We observed an increasing dose-response relationship between the daily weight-adjusted primaquine dose and day 7 methaemoglobin level. For a given primaquine dose regimen, an observed doubling in day 7 methaemoglobin percentage was associated with an estimated 30% reduction in the risk of P. vivax recurrence (adjusted hazard ratio = 0.70; 95% confidence interval [CI] [0.57, 0.86]; p = 0.0005). These pooled estimates were largely consistent across the study sites. Using day 7 methaemoglobin as a surrogate endpoint for recurrence would reduce required sample sizes by approximately 40%. Study limitations include the inability to distinguish between recrudescence, reinfection, and relapse in P. vivax recurrences.

We conducted a systematic search of Medline, Embase, Web of Science, and the Cochrane Library, from 1 January 2000 to 29 September 2022, inclusive, of prospective clinical efficacy studies of acute, uncomplicated P. vivax malaria mono-infections treated with radical curative doses of primaquine. The day 7 methaemoglobin concentration was the primary surrogate outcome of interest. The primary clinical outcome was the time to first P. vivax recurrence between day 7 and day 120 after enrolment. We used multivariable Cox proportional-hazards regression with site random-effects to characterise the time to first recurrence as a function of the day 7 methaemoglobin percentage (log base 2 transformed), adjusted for the partner schizonticidal drug, the primaquine regimen duration as a proxy for the total primaquine dose (mg base/kg), the daily primaquine dose (mg/kg), and other factors. The systematic review protocol was registered with PROSPERO (CRD42023345956).

The 8-aminoquinolines, primaquine and tafenoquine, are the only available drugs for the radical cure of Plasmodium vivax hypnozoites. Previous evidence suggests that there is dose-dependent 8-aminoquinoline induced methaemoglobinaemia and that higher methaemoglobin concentrations are associated with a lower risk of P. vivax recurrence. We undertook a systematic review and individual patient data meta-analysis to examine the utility of methaemoglobin as a population-level surrogate endpoint for 8-aminoquinoline antihypnozoite activity to prevent P. vivax recurrence.

Competing interests: JAG and GCKWK are former employees of GSK and hold shares in GSK and AstraZeneca. GCKWK reports travel support from AstraZeneca. JKB reports participation on the US National Institutes of Health data safety monitoring board; and membership of the editorial board of Travel Medicine and Infectious Disease and the guidelines development group for malaria control and elimination, Global Malaria Programme, WHO. RJC, JKB, and RNP report contributions to Up-to-Date. All other authors declare no competing interests.

Funding: IF is supported by the Oxford Nuffield Department of Medicine Tropical Network Fund. RNP and RJC are supported by Australian National Health and Medical Research Council (NHMRC) Investigator Grants (2008501 and 1194701, respectively). Shoklo Malaria Research Unit (grant 220211) and FN (grant 089179) are supported by the Wellcome Trust. MVGL is a fellow from the National Council for Scientific and Technological Development (CNPq). KT is a CSL Centenary Fellow. JAW is a Sir Henry Dale Fellow funded by the Wellcome Trust (223253/Z/21/Z). NJW is a Wellcome Trust Principal Fellow (093956/Z/10/C). JKB and KT receive institutional research funding from Medicines for Malaria Venture. JKB reports GSK, Wellcome Trust, and Sanaria. This research was supported by grants from the Wellcome Trust. The sponsors or funders did not play any role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability: Pseudonymised participant data used in this study can be accessed via the WorldWide Antimalarial Resistance Network (wwarn.org). Requests for access will be reviewed by a data access committee to ensure that use of data protects the interests of the participants and researchers according to the terms of ethics approval and principles of equitable data sharing. Requests can be submitted by email to [email protected] via the data access form available at https://www.wwarn.org/working-together/sharing-accessing-data/accessing-data . Code for data analysis and visualisation is available at https://github.com/ihsanfadil/methb7 .

A surrogate endpoint is a patient characteristic, such as a biomarker, intended to substitute for a clinical outcome [ 11 ]—specifically vivax recurrence in this context. To validate surrogacy, high-quality studies need to demonstrate that a putative biomarker is affected by the drug intervention and that drug-induced change in the biomarker level can predict the effect on the outcome of interest [ 12 , 13 ]. We examined the utility of methaemoglobin as a population-level surrogate endpoint for vivax recurrence in a pooled individual patient data meta-analysis using available data from P. vivax patients treated with primaquine.

Both primaquine and tafenoquine cause predictable increases in blood methaemoglobin which results from the action of their oxidative metabolites [ 8 ]. This involves a reversible increase in the conversion rate of intra-erythrocytic reduced (Fe ++ ) haem iron in haemoglobin to its oxidised (Fe +++ ) form [ 8 ]. Following daily administration of primaquine or single dose administration of tafenoquine, blood methaemoglobin concentrations gradually increase, reaching peak concentrations after approximately 1 week [ 9 ]. It has been hypothesised that the same oxidative metabolites responsible for methaemoglobinaemia are also responsible for inducing haemolysis in glucose-6-phosphate dehydrogenase (G6PD) deficiency, killing mature gametocytes of P. falciparum, and killing liver stage hypnozoites [ 8 ]. Early experiments with primaquine and its analogues indicated that there were greater increases in blood methaemoglobin for molecules which had improved radical-cure efficacy [ 8 ]. Two recent pharmacometric studies estimated an adjusted proportional reduction in the risk of vivax recurrence of approximately 10% for primaquine [ 10 ] and 20% for tafenoquine [ 4 ] for each additional percentage-point increase in day 7 methaemoglobin. Based on these results, it is hypothesised that increases in methaemoglobin may serve as a proxy for 8-aminoquinoline antihypnozoite activity and, as such, a potential surrogate endpoint for clinical trials to quantify the antirelapse efficacy of 8-aminoquinoline drugs in vivax malaria.

The only drugs available for radical cure (killing latent hypnozoites) are primaquine and tafenoquine. Both are thought to be prodrugs [ 3 , 4 ], necessitating metabolic activation to produce hypnozonticidal activity. The precise mechanism and active metabolites of primaquine and tafenoquine are poorly characterised [ 5 ]. Patients with vivax malaria, who receive an equivalent 8-aminoquinoline dose per body weight, may variably metabolise the drug leading to varying risks of later relapse. Some of this variation in biotransformation is due to polymorphisms in cytochrome P450 (CYP) 2D6 [ 3 ]. Poorer metabolisers are associated with higher relapse rates [ 6 , 7 ].

The human malaria parasites Plasmodium vivax and Plasmodium ovale are characterised by their ability to form dormant liver stage parasites called hypnozoites, which activate weeks to months later to cause relapsing bloodstream infection [ 1 ]. P. vivax is the most geographically widespread cause of human malaria and is a major challenge in malaria elimination. Relapse contributes substantially to the overall burden of symptomatic vivax malaria, causing over 75% of all symptomatic infections [ 2 ]. Preventing relapse is crucial for eliminating the burden of vivax malaria morbidity and mortality.

Risk of bias related to individual studies was evaluated using the Quality in Prognosis Studies (QUIPS) tool [ 24 ] adapted to the current analysis (signalling questions for risk of bias assessment using the QUIPS tool adapted to the current analysis in S1 Text ). Statistical analysis followed a prespecified plan [ 25 ] and was conducted using R Statistical Software (version 4.3.0). Statistical modelling was implemented using the R packages survival [ 26 ], coxme [ 27 ], metafor [ 28 ], rms [ 29 ], and lme4 [ 30 ].

We provide illustrative sample-size calculations to demonstrate how our findings could contribute to making future studies of drug discovery or regimen optimisation in P. vivax more efficient by using blood methaemoglobin as a surrogate outcome. We estimated that a 0.5-mg/kg increase in daily primaquine dose results in a 0.39 increase in the log 2 day 7 methaemoglobin (i.e., a 30% increase). We estimated the standard deviation of the log 2 day 7 methaemoglobin level, conditional on the daily mg/kg primaquine dose from the pooled data. For example, the standard deviation for the high daily dose group is approximately 1.19. Assuming a normal distribution for the log 2 day 7 methaemoglobin conditional on the daily dose allows for a simple calculation of the required sample size. This is derived from a t test for a difference between 2 normal distributions with mean difference (effect size) of 0.39 and equal standard deviations in both groups. For the power calculation for the clinical endpoints, we assumed recurrence risks of 16% versus 8% (corresponding to primaquine doses of 0.5 versus 1 mg/kg over 7 days) [ 23 ].

We estimated the adjusted predictive effect of the day 7 methaemoglobin percentage (log 2 transformed) on the odds of vivax recurrence using multivariable, random-effects binary logistic regression. This model was limited to patients with at least 120 days of follow-up and adjusted for daily mg/kg primaquine dose, primaquine duration, within-site and across-site linear interactions between daily mg/kg primaquine dose and primaquine duration. A random intercept for study site was specified. The association between the maximum absolute change in haemoglobin concentration from day 0 to days 2 to 3 and the day 7 methaemoglobin percentage was estimated using multivariable, random-effects linear regression. This model included baseline haemoglobin concentration, daily mg/kg primaquine dose, age, sex, schizonticidal drug, baseline parasite density (natural-log transformed) as common-effect covariates, and a random intercept and slope for study site and daily mg/kg primaquine dose, respectively. This model was restricted to patients who started primaquine treatment on day 0. If a haemoglobin measurement was missing, haematocrit was used to impute the haemoglobin concentration using the formula derived from a large series of malaria patients; haemoglobin = (haematocrit − 5.62) ÷ 2.60, where haematocrit was measured in percent and haemoglobin was measured in grams per decilitre [ 22 ]. If haematocrit remained missing, these patients were excluded. We also estimated the association of daily mg/kg primaquine dose and day 7 methaemoglobin percentage using a random-effects linear model, allowing for a random intercept and slope for study site and daily mg/kg primaquine dose, respectively.

To compare our estimates with those from previous studies [ 4 , 10 ], we also specified the day 7 methaemoglobin concentration on its original scale and refitted the main survival analysis model. Additionally, we separately (by study and primaquine regimen) fit a more parsimonious Cox proportional-hazards model that adjusted for daily mg/kg primaquine dose and pooled the estimates obtained from all the study-regimen categories using a two-stage individual-patient data meta-analysis approach [ 20 ]. A similar model specification to the one-stage approach that included a few more adjustment factors was not possible as the data were sparse (i.e., few recurrences). A forest plot was constructed to visualise the results under the common-effect and random-effects models. Assuming that the primaquine regimen was the primary determinant of variation in the outcome, dependence in the estimates derived from the same study [ 9 , 21 ] was expected to be minimal.

In the main analysis, patients were right censored at the time of the first recurrent vivax parasitaemia (outcome), any malaria parasitaemia, loss to follow up, blood smear gap of >60 days, or the last day of study, whichever occurred first. We used multivariable, random-effects Cox proportional-hazards regression to model the time to first recurrence as a function of the day 7 methaemoglobin percentage (log 2 transformed) under a one-stage individual patient data meta-analysis framework. This model adjusted for daily mg/kg primaquine dose, primaquine duration (a proxy for total mg/kg primaquine dose), within-site and across-site linear interactions [ 19 ] between daily mg/kg primaquine dose and primaquine duration, age, sex, schizonticidal drug, and baseline parasite density (natural-log transformed). A random intercept and a random slope for day 7 methaemoglobin concentration were included to account for between-site effect-heterogeneity. Linearity and proportional-hazards assumptions were checked. The adjusted hazard ratio can be interpreted as the estimated predictive effect of each doubling in day 7 methaemoglobin percentage, over and above the adjustment factors.

Red nodes represent the outcomes of interest: relapse and blood methaemoglobin (measured on day 7, for example), between which the association was estimated. Under this causal model, blood methaemoglobin is a proxy measurement for the hypnozontocidal activity of primaquine (but not on the causal pathway mediating the effect of primaquine on relapse). Host factors include but not limited to patient’s genetics (e.g., those related to CYP2D6 and G6PD), behaviours, age, immunity to P. vivax, and geographical location.

We proposed a causal directed acyclic graph for this analysis to guide model specification and aid interpretation of results ( Fig 1 ). Missing day 7 methaemoglobin percentages were linearly imputed using levels measured within ±2 days. If only 1 measurement was available, then the imputation assumed a constant (i.e., the single value observed was used). If no measurements were available within this timeframe (day 5 to day 9), the patient was excluded from the analysis.

The primary predictor of interest (surrogate outcome) was the day 7 methaemoglobin concentration, expressed as a percentage of the total haemoglobin concentration. Day 7 was prespecified and typically, methaemoglobin concentrations peak after approximately a week of commencing the daily primaquine regimens. All studies measured methaemoglobin by transcutaneous pulse CO-oximetry. The day 7 methaemoglobin percentage (log base 2 transformed) was included in the statistical model as a continuous variable. If the day 7 methaemoglobin percentage was recorded as 0, we replaced it with 1, assuming an approximately physiological methaemoglobin level of 1%. Zero recordings are likely to represent mis-readings of the analytical machine.

The secondary outcomes were the binary outcome of any P. vivax recurrence between day 7 and day 120 after starting primaquine and the maximum absolute change in haemoglobin concentration from day 0 (day of randomisation) to days 2 to 3 (expected days of the lowest haemoglobin level [ 18 ]) following the start of any antimalarial treatment. In addition to being the main predictor of interest in the statistical model, we also specified the day 7 methaemoglobin concentration as an outcome and modelled this biomarker as a function of daily mg (base) per kilogram primaquine dose.

All studies included in our meta-analysis provided pseudonymised individual data and had obtained ethical approvals from the corresponding site of origin. Therefore, additional ethical approval was not required for the current analysis, as per the Oxford Tropical Research Ethics Committee. We adhered to the PRISMA-IPD guidelines [ 17 ] for reporting this systematic review and individual patient data meta-analysis (Table A in S1 Text ).

The corresponding authors and/or principal investigators of eligible studies that met the study criteria were invited through direct email to contribute their individual patient data. Relevant data from unpublished studies were requested wherever possible. Shared data were uploaded to the Worldwide Antimalarial Resistance Network (WWARN) repository for curation and standardisation, utilising the IDDO SDTM Implementation Guide [ 16 ]. We excluded individual patients with missing information on age, sex, body weight, baseline parasite density, primaquine regimen, or schizonticidal treatment. Patients with severe malaria, pregnancy, mixed-species infection, or those who received adjunctive antimalarials after the initial schizonticidal treatment were also excluded.

Studies were included if the primaquine regimen was administered within the first 3 days of schizonticidal treatment. Search terms are provided in Supporting information (Systematic search terms for the databases in S1 Text ). This systematic review was conducted by 2 reviewers (IF and RJC), with discrepancies resolved through discussion. The review protocol was registered with PROSPERO (CRD42023345956).

We conducted a systematic search of Medline, Embase, Web of Science, and the Cochrane Library based on an existing living systematic review [ 14 ] of prospective clinical efficacy studies of acute, uncomplicated P. vivax malaria mono-infections treated with primaquine published between 1 January 2000 and 29 September 2022 in any language. Studies were eligible for the individual patient data meta-analysis if they were randomised therapeutic trials and prospective cohort studies with a minimum active follow-up of 42 days that recorded methaemoglobin data (at baseline and at least once in the first week of follow-up between day 5 and day 9) following daily primaquine administration given over multiple days. The systematic review was subsequently updated to 26 July 2024 but 2 potentially eligible clinical trials published between 30 September 2022 and 26 July 2024 were not included in the individual patient data meta-analysis due to time constraints in obtaining approval and standardising these data. Only individuals with normal G6PD levels (i.e., ≥30% G6PD activity or a negative qualitative test) were included in the analysis. In individuals with G6PD deficiency, primaquine administration does not lead to the same methaemoglobin increases [ 15 ].

Assuming a comparison between 2 primaquine doses whereby the higher dose results in half the risk of vivax recurrence with recurrence rates of 16% versus 8% (corresponding to primaquine doses of 0.5 versus 1 mg/kg over 7 days), a two-arm randomised trial aiming to show superiority would require a sample size of 256 individuals per group to achieve 80% power with a 5% false positive rate. In contrast, using day 7 methaemoglobin as a surrogate endpoint for vivax recurrence, the required sample size would be reduced by approximately 42% (n = 148 individuals per group) for high daily dose primaquine and could be even more for lower daily doses (Fig I in S1 Text ).

Of the 1,344 (77%) patients followed for at least 120 days, we observed a comparable relationship between day 7 methaemoglobin and the risk of any observed vivax recurrence (dose-adjusted odds ratio of 0.66 for each doubling in day 7 methaemoglobin; 95% CI [0.52, 0.83]; p = 0.0004). Of the 1,360 (78%) patients with normal G6PD activity with haemoglobin (or haematocrit) measured at baseline, there was little or no evidence of an association between the maximum absolute decrease in haemoglobin concentration from day 0 to days 2 to 3 and the day-7 methaemoglobinaemia (adjusted mean difference = 0.01; 95% CI [–0.21, 0.23], p = 0.90).

On the absolute scale (i.e., percentage of the total haemoglobin), each additional percentage-point increase in day 7 methaemoglobin was associated with an estimated 10% reduction in the risk of vivax recurrence (aHR = 0.90; 95% CI [0.84, 0.96]; p = 0.003). Improved model-fit (p < 0.0001) was observed by specifying day 7 methaemoglobinaemia on the logarithmic scale (as in the main model), indicating linearity between multiplicative changes in day-7 methaemoglobin level and the log-hazard ratio for vivax recurrence.

Dashed line represents no association (i.e., hazard ratio = 1). Estimates to the left of the dashed line indicate a lower risk of recurrence with higher day 7 methaemoglobin levels (blue arrow). Conversely, estimates to the right indicate a higher risk of recurrence with higher day 7 methaemoglobin levels (orange arrow). Both the common-effect and random-effects models yielded comparable estimates. Horizontal axis represents dose-adjusted hazard ratios for vivax recurrence associated with a doubling in day 7 methaemoglobin, and is displayed on a logarithmic scale. CI confidence interval, N number of patients.

After adjusting for the daily and total dose of primaquine and other covariates, a doubling in the observed or imputed day 7 methaemoglobin percentage was associated with an estimated 30% reduction in the risk of vivax recurrence (adjusted hazard ratio [aHR] = 0.70; 95% CI [0.57, 0.86]; p = 0.0005. Table F in S1 Text shows an excerpt of the regression output). These pooled estimates were largely consistent across the study sites (between-site standard deviation in aHR = 1.01; this is a multiplicative factor reflecting the variability in the estimated predictive-effect of the day-7 methaemoglobin concentration on the hazard across different study sites, Fig H in S1 Text shows the study-site specific random-effects estimates). There was no evidence of proportional hazards violations within the 120 days of follow-up. A sensitivity analysis restricted to patients with observed day 7 methaemoglobin values gave similar estimates (aHR = 0.66; 95% CI [0.52, 0.84]; p = 0.0008; n = 1,502). Controlling primaquine daily dose and duration, the relationship between day 7 methaemoglobin and the risk of vivax recurrence was generally consistent between the studies ( Fig 5 ). A sensitivity analysis using a two-stage approach estimated a slightly smaller effect (dose-adjusted HR for vivax recurrence of 0.81 for each doubling in day 7 methaemoglobin percentage; 95% CI [0.67, 0.98]; p = 0.037). This numerically different value estimated by the two-stage approach was primarily a result of less flexibility in model specification (e.g., fewer adjustment factors allowed to avoid estimation issues) and sparsity of data.

There was an increasing trend of day 7 methaemoglobin as daily primaquine dose increased among patients with at least 120 days of follow up. Patients who developed P. vivax recurrences typically had lower day 7 methaemoglobin levels. Top panel highlights comparisons between the dose groups and recurrence status using percentile summaries for day 7 methaemoglobin levels. Bottom panel provides a more detailed view of individual variability in day 7 methaemoglobin levels across different daily primaquine doses received and recurrence statuses. Solid line on the bottom panel denotes a regression line (with a 95% CI) of day-7 methaemoglobin levels on daily primaquine dose. Vertical axis is shown on the logarithmic scale. Box width is proportional to the square root of the number of patients. In a boxplot, the box shows the interquartile range with a line for the median, and the whiskers extend to data points within 1.5 times the interquartile range, highlighting potential outliers (dots).

There was dose-dependent primaquine-induced methaemoglobin production. On average, for every additional 0.1 mg/kg increase in the daily primaquine dose, there was an associated 0.34 percentage-point increase in the day-7 methaemoglobin concentration (95% confidence interval [CI] = [0.16, 0.52]; p = 0.002). This effect remained consistent across all sites (between-site standard deviation in mean difference = 0.20; this quantifies the variability in the estimated effect of the daily primaquine dose on the day-7 methaemoglobin concentration across different sites, Fig G in S1 Text shows the study site-specific random-effects estimates). For a particular mg/kg daily dose, patients who later developed vivax recurrences (largely attributable to relapses) during follow-up had lower day 7 methaemoglobin values ( Fig 4 ). In addition to the heterogeneous background risk of relapse across sites, we observed significant interindividual variation in the methaemoglobin response to primaquine treatment.

Among patients treated with a low-to-intermediate daily primaquine dose, day 7 methaemoglobin was lower in the studies where primaquine was combined with chloroquine or quinine as a partner drug compared with ACT (Fig D in S1 Text ). However, there was no clear evidence suggesting a drug–drug interaction at the patient level as there was insufficient within-site variation in the pooled data (within-site interaction, p = 0.29; across-site interaction, p = 0.028). The highest levels were observed among adolescent patients. Younger and older patients had lower primaquine-induced methaemoglobinaemia (Fig E in S1 Text ). Fig F in S1 Text shows the distribution of day 7 methaemoglobin concentrations by primaquine regimen.

Methaemoglobin levels increased after starting primaquine in both regimens, usually reaching a maximum after about a week. Methaemoglobin increased at a faster rate among the 7-day regimen reflecting the higher daily dose taken and generally methaemoglobin started to decrease during the second week, when primaquine was no longer administered. Meanwhile, after peaking at also day 7 for the 14-day regimen, methaemoglobin appears to be at a more constant level during the second week. Boxplot represents the distribution of methaemoglobin levels on a particular day following primaquine. In a boxplot, the box shows the interquartile range with a line for the median, and the whiskers extend to data points within 1.5 times the interquartile range, highlighting potential outliers. Solid curve summarises the data points by partner drug over time. Box width is proportional to the square root of the number of patients. ACT artemisinin-based combination therapy (artesunate/amodiaquine, artesunate/pyronaridine, dihydroartemisinin/piperaquine).

There were consistent increases in methaemoglobin concentration from baseline to day 7 during primaquine treatment for both the 7-day and 14-day regimens. On average, the peak observed methaemoglobin concentrations occurred on day 7 following the start of primaquine administration, with a median day 7 methaemoglobin level of 6.0% (IQR: 3.3 to 9.0). Subsequently, methaemoglobin levels tended to decrease slowly in the 7-day or plateau in the 14-day regimen groups (Figs 3 and C in S1 Text ).

In the primary analysis (n = 1,747 patients), the median age was 20 years (interquartile range [IQR]: 12 to 32), with 89 patients (5.1%) younger than 5 years. Overall, most patients were male (1,116, 64%), resided in the Asia-Pacific region (1,620, 93%), and were from locations with frequent P.vivax relapse periodicity (i.e., a median interval from the first acute episode to relapse of less than 47 days [ 31 ]; 1,614, 92%) and moderate transmission intensity (i.e., 1 to 9 cases per 1,000 person-years [ 32 ]; 1,325, 76%). The majority of patients (1,138, 65%) were treated with artemisinin-combination therapies (ACT), started primaquine on day of enrolment (day 0; 1,496, 86%), and took daily primaquine over 14 days (1,194, 68%). The overall median daily-dose of primaquine was 0.52 mg/kg (IQR: 0.38 to 0.95, Fig B in S1 Text shows the weight-adjusted dose distribution by primaquine duration). Primaquine administration was fully supervised in most studies (i.e., all doses were directly observed; 1,571, 90%). Further details on baseline patient characteristics are summarised in Table 1 . Eligible studies that were not included [ 33 – 37 ] in the analysis tended to have a shorter duration of follow-up but were otherwise similar in other characteristics (Tables C–E in S1 Text ).

We identified 219 P. vivax efficacy studies published between 1 January 2000 and 29 September 2022 ( Fig 2 ). After review, 206 studies were excluded (almost all because they did not measure methaemoglobin or did not have a primaquine arm which started in the first 3 days of treatment), leaving 13 studies eligible for the pooled analysis. Eight of these studies (8/13, 62%) provided individual-level data from patients treated with primaquine; all were randomised, parallel arm clinical trials and assessed as having low risk of bias (Table B in S1 Text ). Individual-level data were available for 4,122 patients from these 8 trials, of whom 1,747 (42%) satisfied the inclusion and exclusion criteria for our analysis (primary data set). These patients were enrolled from 24 study sites across 8 different countries (Indonesia, Brazil, Vietnam, Thailand, Peru, Colombia, Ethiopia, and India; Fig A in S1 Text ). Most patients had a follow-up of at least 120 days since commencing primaquine radical cure treatment (1,344, 77%–secondary sub-data set) and had haemoglobin concentration (or haematocrit) measured on day 0 of antimalarial treatment (1,360, 78%–haemolysis sub-data set).

Discussion

In this systematic review and individual patient data meta-analysis, we confirmed that higher primaquine-induced methaemoglobin concentrations on day 7 are associated with lower rates of P. vivax recurrence [8]. This finding remained consistent across diverse study sites with varying populations and levels of transmission intensity. Our analysis found no indication of a differential predictive effect of day 7 methaemoglobin between the 7-day and 14-day primaquine regimens. Additionally, we observed a positive dose-response relationship between the daily weight-adjusted primaquine dose and day 7 methaemoglobin level.

Our findings are in line with recent estimates derived from Chu and colleagues [10] (a secondary analysis of primaquine trial [9] data from the endemic northwest Thailand-Myanmar border) and Watson and colleagues [4] (an individual patient data meta-analysis of tafenoquine trials). These estimates suggested that an increase of 1 percentage-point of day 7 methaemoglobinaemia was associated with an estimated reduction of 10% for primaquine and 20% for tafenoquine in the risk of P. vivax recurrence (Fig J in S1 Text provides a graphical comparison). While our analysis of primaquine included data from these studies such that estimates are partially correlated, additional data from different studies contributed most (nearly 70%) of our pooled data set. Watson and colleagues [4] analysed vivax patients who received tafenoquine only (not included in this analysis). According to the White and colleagues review [8] of early experiments with primaquine or its analogues conducted more than 70 years ago [38–40], 8-aminoquinoline analogues which resulted in less than 6% of methaemoglobinaemia during treatment showed reduced efficacy. All these results point towards higher methaemoglobin levels following treatment as indicative of increased antihypnozoite activity at the population level.

In the current analysis of patients with normal G6PD activity, we observed no evidence of an association between the extent of early haemolysis on days 2 to 3 from baseline and day 7 methaemoglobinaemia. This suggests that during the early days of illness, early haemolysis was primarily attributable to the acute parasitaemia and rehydration, rather than iatrogenic haemolysis caused by the primaquine active metabolites. The lower day 7 methaemoglobin observed among younger patients may be partly explained by age-related enzyme immaturity and lower drug exposures [10]. The reason for lower methaemoglobin in older patients is less clear.

Methaemoglobinaemia is a simple and readily measured surrogate endpoint that can be quantified within a week after starting administration of an 8-aminoquinoline and has potential to improve the efficiency of exploratory trials (drug–drug interactions, regimen optimisation, drug screening, dose optimisation). Current antirelapse clinical trials generally require patient follow up for many months to observe recurrent infections in order to have sufficient power to determine comparative efficacy of different treatment arms. Quantifying methaemoglobinaemia also has potential to be a useful approach for monitoring of trials and clinical practices at an individual level, serving as a surrogate marker for patient adherence. Methaemoglobin could also serve as an adherence or exposure surrogate in clinical trials investigating radical curative efficacy (particularly for the investigation of relapses or unexpectedly poor radical curative efficacy). Although methaemoglobin is a surrogate endpoint for adherence to treatment, to our knowledge, these results are not driven by adherence (90% of patients had fully supervised treatment). Early-phase studies, preceding more definitive efficacy trials, may stand to benefit the most from this novel endpoint [8]. Minimum required sample sizes could also be substantially reduced by using day 7 methaemoglobin as an endpoint to improve cost-efficiency in conducting P. vivax trials. An ongoing study (NCT05788094) is currently using day 7 methaemoglobin measurement as a secondary outcome to address the question of whether there is important drug–drug interaction between tafenoquine and chloroquine, dihydroartemisinin (DHA)-piperaquine, or artemether-lumefantrine [41]. The recent INSPECTOR trial in Indonesia used DHA-piperaquine as the partner schizonticidal drug and lower than expected efficacy was observed [42] compared to trials with chloroquine [43–45]. In the INSPECTOR study, tafenoquine plus DHA-piperaquine resulted in a median methaemoglobin of 1.3% (range = 0.7% to 3.7%), which was similar to DHA-piperaquine alone (median = 1.0%, range = 0.5% to 1.8%). In contrast, primaquine plus DHA-piperaquine produced more than twice as much methaemoglobin (median = 2.9%, range = 0.9% to 7.9%) and superior efficacy than the tafenoquine plus DHA-piperaquine arm. While individual patient data from this trial could not be incorporated into our pooled analysis due to substantial time for obtaining approval and standardisation, the results appear to corroborate our findings that higher increases in blood methaemoglobin corresponded to improved radical cure rates.

Our study has several limitations. Some patients missed measurements and some studies by design intentionally did not measure methaemoglobin on day 7. We addressed such missing data through linear interpolation, since methaemoglobin was recorded within 2 days before and/or after day 7. It is important to note that methaemoglobinaemia itself is an inherently noisy measure with substantial variation between patients, especially in patients receiving higher doses of 8-aminoquinoline treatment (Fig I in S1 Text shows higher methaemoglobin variability in higher dose groups). There is also a possibility of less accurate methaemoglobin measurements by using CO-oximetry for patients with darker skin [46]. Improved reliability may be achieved by measuring methaemoglobin on both hands, incorporating multiple or repeated measurements, waiting longer, and blocking fluorescent light (e.g., turning off lights or covering the hand when checking). More invasive methods, such as directly measuring the biomarker in capillary blood, could enhance measurement accuracy but are less acceptable and would be logistically challenging in low-resource settings. In an upcoming analysis, we plan to explore alternative summary metrics for methaemoglobin levels using pharmacometric modelling [25]. This approach aims to capture exposures to primaquine’s active metabolites and improve the overall robustness of our findings.

Only 2 studies [47,48] were conducted in locations with negligible reinfection. Therefore, there is uncertainty regarding the aetiology of P. vivax recurrences for most of the patients, since there is currently no standardised method to differentiate between recrudescence, reinfection, and relapse [1]. It is important to note that the use of highly efficacious schizonticides across the study sites makes recrudescence unlikely. Another potential source of error includes participants who might not have had hypnozoites in the liver to start with, thus having zero risk of relapse, despite primaquine-inducing methaemoglobinaemia. However, this circumstance should bias the effect estimates towards the null (i.e., no association between day 7 methaemoglobinaemia and the risk of vivax recurrence), making our estimates conservative.

The generalisability of our findings is constrained to the large majority of vivax malaria patients with normal G6PD activity (≥30% G6PD activity or a negative qualitative test). For individuals with G6PD deficiency, blood methaemoglobin will not serve as a valid surrogate endpoint for antihypnozoite activity [8,10]. In G6PD deficiency, primaquine does not induce significant methaemoglobinaemia [15,49]. We currently lack sufficient data to explore how CYP2D6 polymorphisms impact primaquine biotransformation to its active metabolites, as most patients were not genotyped [10]. The previous analysis by Chu and colleagues suggested lower day 7 methaemoglobin in null CYP2D6 metabolisers (genotype-based activity score of 0) [8]. Intermediate metabolisers may also have lower active metabolites and higher relapse rates [6,7]. Further research specifically targeted at these vulnerable populations is needed.

Our pooled data set included patients from various countries with different antimalarial treatment policies for their first-line schizonticides, each having different elimination kinetics and thus different durations of suppressive posttreatment prophylaxis. The use of time-to-event models in such contexts may introduce bias in favour of drugs with longer half-lives, such as chloroquine. However, our sensitivity analysis, employing logistic regression that included partner drug as a covariate, yielded similar predictive-effect estimates. The goal of this analysis was not to establish day-7 methaemoglobin levels as an individual level predictor of relapse risk. Predicting relapse at the individual level is very difficult because there is large interindividual variation in liver hypnozoite burdens [1,50]. A patient may have no methaemoglobin response to primaquine but not subsequently relapse because they had no hypnozoites at enrolment. Indeed, in no primaquine treated arms of the included studies, the risk of any recurrent illness by day 120 ranged from around 25% to 80%.

In conclusion, in primaquine-treated individuals, the day 7 methaemoglobin concentration is a pharmacodynamic proxy for exposure to the biologically active metabolites of primaquine, making it a valid population-level surrogate endpoint in patients with P. vivax malaria who have normal G6PD activity. The consistency with results observed in tafenoquine studies suggests a common drug-class phenomenon for 8-aminoquinolines. Direct comparisons of 8-aminoquinoline-induced methaemoglobinaemia between primaquine and tafenoquine in future studies could prove useful. These findings collectively enhance our understanding of the causal mechanisms by which 8-aminoquinoline drugs exert their effects, facilitate drug discovery and regimen optimisation, and influence clinical practices.

[END]
---
[1] Url: https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1004411

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.