PLoS Medicine
Public Library of Science
Abstract

Why was this study done?Complete cure of vivax malaria, the most geographically widespread malaria species, requires the use of 8-aminoquinoline drugs to clear dormant liver stages of the parasite (‘radical cure’); however, these drugs can cause severe haemolysis in individuals with glucose-6-phosphate dehydrogenase (G6PD) deficiency.Ultraviolet (UV) spectrophotometry is used as the reference test to measure G6PD activity, for validating new point-of-care diagnostics, and to determine population-specific definitions of G6PD deficiency.Currently, there is no universal threshold to define G6PD deficiency, and each laboratory must invest time and resources to derive site- and laboratory-specific definitions of G6PD deficiency.What did the researchers do and find?We pooled measurements of G6PD activity from studies conducted across different countries and laboratories worldwide.We assessed the comparability of spectrophotometry results between these laboratories to see whether a universal definition and diagnostic cutoff for G6PD deficiency could be determined.There was substantial variation in the performance and absolute measurements of spectrophotometry conducted in different laboratories, hindering the definition of a universal cutoff for G6PD deficiency.What do these findings mean?These findings highlight the importance of quality-control measures to minimise the influence of laboratory procedures on observed measurements.The data suggest that while a robust universal, assay-specific G6PD activity cutoff value can be established for diagnosis of severe G6PD deficiency (<30% normal enzyme activity), this approach is less robust for diagnosing intermediate G6PD deficiency.Newly developed diagnostic assays that are less sensitive to laboratory conditions and require less sample preparation are required and may help provide more standardised quantitative G6PD activity measurements across different contexts.

Pfeffer, Ley, Howes, Adu, Alam, Bansil, Boum, Brito, Charoenkwan, Clements, Cui, Deng, Egesie, Espino, von Fricken, Hamid, He, Henriques, Khan, Khim, Kim, Lacerda, Lon, Mekuria, Menard, Monteiro, Nosten, Oo, Pal, Palasuwan, Parikh, Pitaloka Pasaribu, Poespoprodjo, Price, Roca-Feltrer, Roh, Saunders, Spring, Sutanto, Ley-Thriemer, Weppelmann, von Seidlein, Satyagraha, Bancone, Domingo, Price, and Garner: Quantification of glucose-6-phosphate dehydrogenase activity by spectrophotometry: A systematic review and meta-analysis

Introduction

Plasmodium vivax and P. ovale both form dormant liver stages (hypnozoites) that can reactivate weeks to months following an initial infection, resulting in relapsing malaria [1]. The complete treatment of either species requires the use of drugs able to clear hypnozoites from the liver (‘radical cure’), alongside standard blood-stage antimalarials. The only licenced antimalarial compounds that can kill hypnozoites and prevent relapses are the 8-aminoquinoline drugs primaquine and tafenoquine. Primaquine is used to treat patients with malaria in two scenarios: either as a single low dose to kill P. falciparum gametocytes and reduce transmission, or at a higher dose administered over 14 days to kill P. vivax hypnozoites. Tafenoquine, is another 8-aminoquinoline drug, which has recently been licensed as a single-dose hypnozoiticidal agent for P. vivax liver stages. While well tolerated in the majority of recipients, standard dosing of either drug can cause severe haemolysis in patients with glucose-6-phosphate dehydrogenase (G6PD) deficiency [2]. Tafenoquine is more slowly eliminated than primaquine; hence, patients are exposed to potentially haemolytic concentrations of the drug for longer. The successful rollout of tafenoquine will therefore require more stringent G6PD screening than is currently available in malaria endemic areas.

G6PD deficiency is a common inherited enzymopathy that is particularly prevalent in malaria-endemic regions [3,4]. Red blood cells (RBCs) of affected individuals are susceptible to haemolysis caused by oxidative stress, induced by a variety of stimuli including drugs (e.g., rasburicase, 8-aminoquinolines, and dapsone), foods (e.g., fava beans), or acute infection [5,6]. The gene encoding G6PD is located on the X chromosome (Xq28). Hence, males inherit a single copy and are either hemizygous G6PD deficient (G6PDd) or G6PD normal (G6PDn), whereas females carry two copies and can be homozygous deficient or normal, or heterozygous for a mutant G6PD allele. Hemizygous and homozygous deficient individuals express >95% deficient RBCs, while heterozygotes harbour two distinct RBC populations, a G6PDd and a G6PDn population. Different patterns of X-inactivation in heterozygotes lead to different proportions of deficient RBCs between individuals and widely varying G6PD phenotypes [7]. Thus, while males are phenotypically either G6PDd or G6PDn, females exhibit a wide range of G6PD levels from very low deficient levels through intermediate to normal. Furthermore, the G6PD gene itself exhibits substantial genetic variability, with over 200 distinct G6PD mutations described [8,9], exhibiting a wide range of enzyme activity levels [10,11].

Ultraviolet (UV) spectrophotometry is the reference standard diagnostic method for quantifying G6PD enzyme activity [12]. Multiple commercial assay kits are available, and although there is strong correlation of measurements between assays, absolute values vary [13,14]. Such absolute values will be important for the universal use of novel point-of-care quantitative diagnostics, such as forthcoming biosensors. The interpretation of quantitative G6PD measurements requires a predetermined definition of ‘normal’ (100%) G6PD activity. However, there is no consensus definition of normal activity. Each laboratory must establish its own reference values and diagnostic thresholds [15]. In 2013, Domingo and colleagues proposed a method of standardising this process [16], deriving normal enzyme activity from the median value of non-deficient males, known as the adjusted male median (AMM). Although this approach is less vulnerable to outliers or the underlying prevalence of G6PDd than the standard mean or median, interpretation of assay results requires derivation of a context-specific local AMM. Once defined, the AMM enables classification of samples based on their relative G6PD activity.

Treatment decisions for primaquine or tafenoquine are currently based on two important thresholds: 30% and 70% normal enzyme activity, respectively. The former is the approximate cutoff activity for most qualitative tests [17]. The latter is designed to exclude heterozygous females with intermediate enzyme activity who are also at risk of haemolysis [18,19]. Setting the threshold too low risks falsely categorising patients as G6PDn and exposing individuals to primaquine-induced haemolysis. Setting the threshold too high potentially excludes G6PDn patients from receiving radical cure, putting them at risk of repeated episodes of vivax malaria and associated morbidity.

The aim of the current study was to pool spectrophotometric data from diverse laboratory and geographic contexts to quantify the degree to which assay-based variability influences inter- and intra-study comparability of G6PD spectrophotometry, and to explore the implications of this variability on diagnosing severe and intermediate G6PD deficiency.

Methods

Data collation

The data for this prospectively planned meta-analysis were pooled from a systematic literature review and unpublished contributions from collaborating investigators (PROSPERO: CRD42019121414) according to PRISMA guidelines (S1 File). Studies involving individual-level spectrophotometric measurements of G6PD activity were identified via a PubMed search (25 April 2018) using the following terms: G6PD OR “glucose-6-phosphate dehydrogenase” OR “glucose 6 phosphate dehydrogenase”) AND (quantitative OR spectrophot*) . Additional studies were identified via reference lists and correspondence with authors. In view of the wide range of assay manufacturers and laboratory methodologies in the published literature, only papers published between January 2005 and April 2018 were screened for inclusion to ensure comparability of diagnostic protocols and quality control procedures. The title, abstract, and full text of studies between these dates were then screened to identify those that used UV spectrophotometry to define G6PD activity. Only studies that measured NADPH formation at a wavelength of 340 nm were included. Authors were contacted via email and invited to contribute individual patient and quality control data. A minimum of two attempts was made to contact authors before excluding studies. Studies targeting specific ethnic groups (e.g., African-American blood donors) were included, but those that only included individuals with known haematological conditions, family studies, or studies for which insufficient details were available were excluded. Studies of patients with malaria were included but analysed separately from individuals without malaria. One paper included studies performed in two different countries and was considered as two separate data sources ([20]; S1 Table).

Demographic data and available haematological parameters were collated and stored in a standardised database along with metadata of the study characteristics. Samples missing data on either G6PD activity (U/g Hb) or sex were excluded from the analysis, as were those for which measured G6PD activity was extreme (>30 U/g Hb). Quality of included studies was assessed using an adapted form of the QUADAS-2 tool ([21]; S2 File).

Data analysis

Intra-laboratory assay repeatability was investigated from replicate measures. For each sample, the coefficient of variation (CV) and absolute difference between replicate measures were calculated, along with a mean CV and mean difference for each study. CV values and proportion of samples with high inter-replicate variability were compared between studies using the Kruskal-Wallis and chi-squared tests.

To assess inter- and intra-laboratory variability in G6PD spectrophotometry, data from the measurement of manufacturer-provided quality control samples were used to quantify the magnitude and variance of control measurements (deficient, intermediate, and normal) and how these differed between assays and studies.

World Health Organization (WHO) prequalification for in vitro diagnostics recently defined G6PD deficiency for males and females at below 30% and intermediate activity between 30% and 80% G6PD activity [22]. The analysis performed here uses the 70% threshold used for the tafenoquine clinical trials [18,19]. For population studies, the AMM [16] was calculated for each study separately and defined as 100% G6PD activity. G6PD deficiency was defined as ‘activity below 30% of the respective study AMM’ and intermediate G6PD deficiency as ‘activity between 30% and 70% of the respective study AMM’. Variation between different study populations was assessed by assay and whether the participant had malaria. Studies with a high risk of bias due to patient selection were excluded from the analysis (S2 Table). To preclude the influence of heterozygosity on the observed variability, only male samples were compared. G6PD activity distributions from control and participant samples were compared using the Kruskal-Wallis test with pairwise Mann-Whitney-Wilcoxon tests using Bonferroni correction (magnitude) and Levene’s test (variance).

A universal AMM was calculated by applying the standard AMM formula to a comparable subset of all included samples. To control for differences due to assay variability or malaria status, the universal AMM was only derived from samples tested using the Trinity assay in patients without malaria. For each study, separately and across the pooled subset of data, the diagnostic performance (sensitivity, specificity, and negative predictive value [NPV]) of this universal AMM was then evaluated at both the 30% and 70% thresholds, considering diagnoses derived from study-specific AMMs as the reference. Exact binomial confidence limits were estimated for both sensitivity and specificity. Performance was compared to a conservative universal threshold, where 100% G6PD activity was determined by the upper limit of the 95% CI of the mean G6PD activity for G6PDn males in the included subset. Due to the limited availability of relevant covariate data from included studies, the universal AMM, conservative threshold, and threshold performances were not adjusted for study-level covariates. Furthermore, such adjustment for study-level covariates would impose an impractical level of complexity in clinical practice.

All data management, visualisation, and statistical analyses were performed using the tidyverse [23], ggplot2 [24], cowplot [25], epiR [26], and base R packages in R version 3.5.2 [27]. Prior ethical approval was obtained for all primary studies included in the analysis.

Results

In total, 312 studies were identified from the literature published between January 2005 and April 2018, as well as 18 unpublished studies. Of these, 243 studies were excluded based on title, abstract, or full text review, and 55 studies were excluded because the corresponding author did not respond or the relevant data were not available (Fig 1). A further two studies were excluded due to methodological criteria (incomparable measurement units or spectrophotometry wavelength). Data from 231 individuals were excluded for reasons stated in Fig 1.

PRISMA flow diagram depicting the literature search and data-screening procedures.
Fig 1
G6PD, glucose-6-phosphate dehydrogenase; RBC, red blood cell.PRISMA flow diagram depicting the literature search and data-screening procedures.

The final dataset comprised spectrophotometric measurements from 15,811 individuals collected from 30 studies (20 in Asia, 5 in Africa, and 5 in the Americas; Table 1). The age of participants ranged from 0 to 86 years, and 44.4% of participants (7,083 individuals) were female. G6PD spectrophotometry results were generated from six different manufacturer’s assays: Trinity Biotech, Ireland (21 studies, 12,222 participants); Randox Laboratories, United Kingdom (4 studies, 1,476 participants); Pointe Scientific, United States (2 studies, 883 participants); Sigma-Aldrich, US (2 studies, 691 participants); BIOLABO, France (1 study, 320 participants), and Spinreact, Spain (1 study, 319 participants).

Table 1
Characteristics of the final pooled database by malaria status.
CharacteristicsMalaria PositiveMalaria NegativeMalaria UnknownTotal
Sex        
Male3,433(21.71)3,360(21.25)1,990(12.59)8,783(55.55)
Female3,446(21.79)1,422(8.99)2,160(13.66)7,028(44.45)
Region
Africa808(5.11)708(4.48)127(0.8)1,643(10.39)
Americas1,588(10.04)324(2.05)346(2.19)2,258(14.28)
Southeast Asia4,483(28.35)3,750(23.72)3,677(23.26)11,910(75.33)
Age
0 to 1115(0.73)8(0.05)566(3.58)689(4.36)
1 to 5601(3.8)194(1.23)123(0.78)918(5.81)
5 to 151,250(7.91)930(5.88)1,104(6.98)3,284(20.77)
15 to 603,408(21.55)2,939(18.59)2,099(13.28)8,446(53.42)
>60179(1.13)29(0.18)50(0.32)258(1.63)
Unknown1,326(8.39)682(4.31)208(1.32)2,216(14.02)
G6PD Assay *
Trinity5,119(32.17)3,964(24.91)3,139(19.73)12,222(77.3)
Randox1,296(8.15)180(1.13)--1,476(9.34)
Pointe Scientific564(3.54)319(2)--883(5.58)
Sigma-Aldrich----691(4.34)691(4.37)
BIOLABO----320(2.01)320(2.02)
Spinreact--319(2)--319(2.02)
Total6,879(43.51)4,782(30.24)4,150(26.25)15,811(100)
Values indicate the number of individuals and percentage of the total database, n (%), in each cell.
*One hundred individuals tested by both Trinity and Randox are counted twice here.

Assay repeatability

Replicate measurements of G6PD activity were available from five studies, including 14.4% (2,204/15,811) of samples (Table 2and Fig 2). One study (n = 609) performed triplicate measures on all samples, while four studies (n = 1,595) performed duplicate measurements on all samples and a third test on a subset of samples with high inter-replicate variability (n = 53). The magnitude of inter-replicate variability differed between studies, with the study-specific maximum difference ranging from 0.65 U/g Hb to 8.64 U/g Hb and the mean CV ranging from 1.6% to 14.9% (Table 2). The percentage of samples that showed a high inter-replicate difference (>2 U/g Hb) differed significantly across all studies (p < 0.001), ranging from 0% to 32% of each study’s samples. Similar inter-study differences were evident when considering relative variability, with the percentage of samples exhibiting a high CV (>10%) ranging from 1.53% to 62%.

Repeatability of G6PD spectrophotometry.
Fig 2
Inter-replicate agreement in five studies enrolling 2,204 participants. (A) Absolute G6PD activity (U/g Hb) of replicate measures. Each point represents a single G6PD activity measurement, with measurements from the same individual connected by vertical lines for clarity. (B) Relative difference (CV; %) between replicate measures of G6PD activity. Five points with CV greater than 50% are shown as ‘50+’ for clarity (CV = 51, 61, 98, 110, 140). For both panels, the assay used is labelled for each study and the study-wise mean (range) inter-replicate difference is shown. Note: One study, Ley et al., 2018, provided measurements from the same samples using both Randox (n = 92) and Trinity (n = 100). CV, coefficient of variation; G6PD, glucose-6-phosphate dehydrogenase.Repeatability of G6PD spectrophotometry.
Table 2
Inter-replicate differences in G6PD spectrophotometry.
StudyAssaySample
Size
Mean ΔaΔ >2 U/g HbMean CVCV >10%
(Range)(% Samples)(Range)(% Samples)
Bancone et al., 2015 [28]Trinity1500.1 (0.0–0.7)0.02.5 (0.0–23.2)3.3
Hailu et al., 2018 bTrinity3670.7 (0.0–2.7)1.64/0 (0.0–15.7)4.9
Ley et al., 2018 b*Randox920.6 (0.0–1.9)0.09.8 (0.0–48.8)34.8
Ley et al., 2018 b*Trinity1001.8 (0.0–8.6)32.014.9 (0.3–60.5)62.0
Oo et al., 2016 [29]Trinity9780.2 (0.0–3.6)0.51.6 (0.0–35.2)1.5
Satyagraha et al., 2016 [30]Trinity6091.0 (0.0–7.8)12.66.7 (0.0–140)16.3
aΔ = absolute inter-replicate difference (U/g Hb).
bUnpublished study.
*Note: These rows refer to a single study which provided measurements of the same samples using both Randox and Trinity.
Abbreviation: CV, coefficient of variation (%)

Inter- and intra-laboratory variability

Quality control data were available from nine studies: seven that used the Trinity assay (using normal, intermediate, and deficient controls) and three studies that used the Randox assay (normal and deficient controls; Fig 3). Results differed significantly across studies for all control categories (p < 0.001), when considering both the Trinity and Randox assays. Smaller inter-study differences were observed between the studies using Randox, all of which were conducted in the same laboratory. Intra-study variability (study-level variance of control measurements) differed significantly for all control categories for studies using both the Trinity (normal, p < 0.001; intermediate, p < 0.001; deficient, p < 0.001), and Randox (normal, p = 0.002; deficient, p = 0.02) assays.

Inter-study variation in control sample measurements.
Fig 3
Quality control data from nine studies are shown (n = 1,509). The three leftmost panels contain measurements of deficient, intermediate, and normal controls from studies using the Trinity assay; panels on the right depict deficient and normal control measurements using the Randox assay. The median and interquartile range for each category are superimposed in black. Points are coloured to indicate studies conducted in the same laboratory. A square-root transformation is applied to the y-axis to depict variation in deficient samples more clearly. Note: One study, Ley et al., 2018, provided measurements from the same samples using both Randox (n = 92) and Trinity (n = 100). G6PD, glucose-6-phosphate dehydrogenase.Inter-study variation in control sample measurements.

To control for differences attributable to assay method, the analysis of the variability in spectrophotometry data between study populations was first addressed in the 18 studies (6,245 individuals) using the Trinity assay. There was considerable inter-study variation in the distribution of G6PD activity and the derived AMM values across studies (Fig 4). The AMM ranged from 5.7 to 12.6 across studies consisting of participants without malaria and 7.8 U/g Hb to 12.4 U/g Hb for those with malaria. The inter-study differences in G6PD activity were statistically significant across all studies (p < 0.05). In the studies that used a spectrophotometry assay other than Trinity, there was significant inter-study variation in G6PDn males for Pointe Scientific (p < 0.001), Randox (p = 0.04), and Sigma-Aldrich (p < 0.001) (S1 Fig).

Inter-study variation in G6PD activity measured using the Trinity assay.
Fig 4
The distribution of G6PD activity measurements is shown for males enrolled in representative studies using the Trinity G6PD assay. Participants with (n = 2,694) and without malaria (n = 3,516) are shown separately. The AMM (black text), 70% threshold (blue), and 30% threshold (red) are labelled for each study (U/g Hb). Point colours indicate the country of origin of each study. Filled points represent G6PDn individuals (>30% study AMM) and hollow points indicate G6PDd individuals (<30% study AMM). Note: Roh et al., 2016, included 63 males <1 year of age, who may have had elevated G6PD activity [31,32]. AMM, adjusted male median; G6PD, glucose-6-phosphate dehydrogenase; G6PDd, G6PD deficient; G6PDn, G6PD normal.Inter-study variation in G6PD activity measured using the Trinity assay.

Universal thresholds

Participants were categorised as being either severely deficient (<30% AMM; ineligible for primaquine) or severely/intermediately deficient (<70% AMM; ineligible for tafenoquine) using the study-specific AMMs. In patients without malaria assessed using the Trinity assay, the pooled AMM (i.e., 100% G6PD activity threshold) was 9.4 U/g Hb, resulting in a 70% threshold of 6.6 U/g Hb and a 30% threshold of 2.8 U/g Hb. The conservative universal threshold (calculated as the upper limit of the 95% CI of the mean G6PD activity, instead of the median used for the AMM) yielded a 100% threshold of 9.7 U/g Hb; 70% threshold of 6.8 U/g Hb, and 30% threshold of 2.9 U/g Hb.

Using the site-specific AMM as the reference, the pooled AMM correctly categorised 89% to 100% of severely deficient patients (both males and females) at the 30% threshold with a pooled sensitivity of 97% (95% CI: 95%–98%). At this threshold, the study-wise specificity was greater than 96% for all studies with a pooled specificity of 100% (95% CI: 100%–100%) (Table 3, S3 File: S7–S12 Figs). Seventeen individuals (12 females and 5 males, out of a total of 7,520) were falsely classified as G6PDn, with a mean (range) G6PD activity of 26.8% (23%–29.6%) of the local AMM (S4 Table, Fig 5, S3 File: S2 and S3 Figs).

Universal diagnostic classifications by G6PD activity—both males and females, using pooled AMM.
Fig 5
Relative G6PD activity is shown (x-axis, normalised to site-specific AMM) for individuals falling into each diagnostic category, as depicted by coloured bars (false normal [FN], red; false deficient [FD], orange; true normal [TN], light grey; true deficient [TD], black). Diagnoses are shown at both the 30% threshold (left; 17 FN, 21 FD, 6,926 TN, 556 TD) and 70% threshold (right; 139 FN, 258 FD, 5,983 TN, 1,140 TD). All individuals tested with Trinity without malaria infection are shown, except 69 individuals with G6PD activity >200% local AMM (n = 7,451). AMM, adjusted male median; G6PD, glucose-6-phosphate dehydrogenase.Universal diagnostic classifications by G6PD activity—both males and females, using pooled AMM.
Table 3
Performance of pooled universal AMM diagnostic thresholds.
Deficient Threshold (30%)
SensitivitySpecificityNPV
Males0.99 (0.97–1.00)1.00 (1.00–1.00)1.00 (1.00–1.00)
Females0.91 (0.85–0.95)1.00 (0.99–1.00)1.00 (0.99–1.00)
Overall0.97 (0.95–0.98)1.00 (1.00–1.00)1.00 (1.00–1.00)
Intermediate Threshold (70%)
SensitivitySpecificityNPV
Males0.95 (0.93–0.96)0.96 (0.95–0.96)0.99 (0.98–0.99)
Females0.84 (0.81–0.87)0.96 (0.95–0.97)0.97 (0.96–0.97)
Overall0.89 (0.87–0.91)0.96 (0.95–0.96)0.98 (0.97–0.98)
Abbreviations: AMM, adjusted male median; NPV, negative predictive value

At the 70% threshold, the study-wise sensitivity of the universal AMM ranged between 64% and 100% with a pooled sensitivity of 89% (95% CI: 87%–91%). At this threshold, the specificity ranged from 35% to 100%, with a pooled specificity of 96% (95% CI: 95%–96%) (Table 3& S3 File: S7–S12 Figs). One hundred and thirty-nine individuals, 107 females and 32 males, were falsely classified as G6PDn at the 70% level, with a mean (range) G6PD activity of 56.5% (52.4%–59.5%) of the local AMM (S4 Table, Fig 5, S3 File: S2 and S3 Figs). At both the 30% and 70% threshold, performance was slightly improved when considering the conservative universal thresholds (S3 and S4 Tables, S3 File: S4–S6, S13–S19 Figs). Diagnostic performance of universal thresholds did not differ substantially in patients with or without malaria. At all thresholds, when there was a difference in diagnostic performance of universal thresholds, the performance in females was worse than in males, across all studies and within individual studies (Table 3 and S3 File: S8–S13 Figs).

Discussion

Safe implementation of radical cure of malaria with primaquine or tafenoquine will be critical for the timely elimination of P. vivax. In view of the risk of drug-induced haemolysis, patients should be tested for G6PD deficiency prior to treatment to avoid exposing vulnerable individuals to oxidative stress. Widespread application of radical cure will thus require robust and easily interpretable diagnostic thresholds for G6PD enzyme activity. Our pooled analysis of G6PD activity across a diverse range of studies and geographical locations highlights significant inter-study and intra-study differences in absolute G6PD measurements derived using spectrophotometry. This variability has strong potential to confound reliable diagnosis of G6PD deficiency.

We observed considerable variability in G6PD activity measurements between sites, even when considering the same spectrophotometric assay. The differences in the quantification of control samples suggest substantial contribution of laboratory- or assay-based factors, although these may be exacerbated by unmeasured genetic or environmental differences between studies. Despite the large sample size of this meta-analysis, the data lacked the granular information necessary to isolate specific laboratory or procedural factors at play. The presence of differences between research laboratories illustrates the likely pervasiveness of interlaboratory differences in G6PD spectrophotometry, with fundamental implications for comparing absolute G6PD activity measurements between studies.

Despite this variability, our findings demonstrate strong performance of a universal threshold for identifying G6PD activity below 30% of the local AMM. At this level, the universal diagnostic threshold (2.8 U/g Hb) demonstrated robust diagnostic performance with sensitivity and specificity exceeding 97%. In these cohorts, 17 out of 7,520 individuals would receive primaquine despite having a G6PD activity less than 30% the local AMM; however, all of these misdiagnoses occurred around the diagnostic cutoff, with a minimum G6PD activity of 23% (S4 Table, Fig 5). The majority of the 17 misdiagnoses were in females, suggesting that a portion of these may come from heterozygous females with G6PD activities spanning the 30% threshold. Hence, this threshold may have utility in certain contexts where a local AMM is unavailable (e.g., validation studies of qualitative assays).

The use of tafenoquine requires a more stringent threshold to reduce the risk of haemolysis in heterozygous females with intermediate deficiency. At the 70% enzyme activity level, the diagnostic performance of pooled universal cutoffs was worse than that for the 30% threshold. This is likely a consequence of both inter-study variation in G6PD activity, as well as natural variation (noise) in G6PD activity levels around this 70% limit. Of the 139 individuals misclassified as G6PDn at this level, only 36 (25.4%) exhibited a G6PD activity less than 60% of the local AMM, and all had a G6PD activity greater than 53%. Again, similar to the 30% threshold, the majority of these were females that are in the 60%–70% activity range. The exact relationship between G6PD activity and the haemolytic risk associated with tafenoquine is unknown, although the risk is thought to be inversely correlated with enzyme activity [33]. Reassuringly, false normal diagnoses at the 70% enzyme activity level using the universal thresholds occurred in a minority of individuals with G6PD activity mostly near the 70% mark.

The universal thresholds defined in our study are based on the Trinity assay kit, which is now discontinued. While Alam and colleagues have shown good correlation between Trinity measurements and results from a Pointe Scientific assay, with little difference in absolute measurements (n = 50, [14]), a study by Pal and colleagues found a more modest correlation in absolute measurements (n = 183; [13]). Consequently, demonstration of suitability of a proposed universal threshold will need to be demonstrated with each specific assay prior to its widespread endorsement and application. Promising point-of-care quantitative biosensors reduce the need for sample storage and complex processing [12,34]. As such, these new assays may exhibit superior interlaboratory comparability and in future may provide the basis for a universal definition of G6PD deficiency. Until there is consensus regarding safe and robust universal thresholds, site- and assay-specific definitions of G6PD deficiency will still be required. Currently, laboratories either use the product insert ranges or determine normal ranges based on normal samples tested in the laboratory.

We identified notable differences in assay repeatability and variance of control measurements between sites. Such assay-based variability may arise from inconsistencies in assay procedures, or sample handling [16,34,35]. This variation demonstrates the importance of performing and monitoring replicate measurements and control sample measurements in order to minimise assay-based variability and maximise comparability of results.

It is worth noting that diagnostic definitions of ‘intermediate’ G6PD activity have varied over time and contexts. The early WHO classifications of G6PD variants considered anything above 60% to be normal G6PD activity [36], current WHO guidelines place the cutoff at 80% normal activity [17], and prescription requirements for tafenoquine in the US and Australia require a G6PD activity of 70% or higher [37]. Regardless of the definition, our study highlights the significant challenges in establishing an estimate of 100% G6PD activity (in clinically relevant units of U/g Hb) from which patients’ relative enzyme activity can be calculated.

Our study has a number of limitations. First, although we excluded studies in which haematological conditions may have influenced observed G6PD activity levels, the final dataset consisted of samples from diverse clinical and community survey contexts. This may have led to unequal influence of undiagnosed health conditions (e.g., other haematological conditions) upon measured G6PD activity. However, there was no clear pattern in G6PD activity by study type. Furthermore, while it has been suggested that neonates have elevated G6PD activity [31,32], introducing possible bias, few of the included studies enrolled newborns ([38,39]). Second, it is common practice to exclude replicate measures that differ by more than a certain amount (e.g., CV > 10%) as well as control measurements falling outside of an expected range. Included datasets may have already excluded these values in some, but not all, cases, meaning that the current study would overestimate true assay performance. Third, the study did not consider lot-to-lot variability in control isolates (which consist of lyophilised blood specimens) and assay reagents; these may have contributed to some of the inter-study variability observed for both the controls and AMM values. Nevertheless, the current findings represent an indication of inter- and intra-study variability in ‘valid’ results of G6PD spectrophotometry.

Conclusions

Interlaboratory variability hinders the definition of universal cutoff values for the classification of G6PD activity using spectrophotometry, particularly at the 70% G6PD activity level. Caution is advised in comparing research findings based on absolute G6PD activity measurements across studies, such as those characterising novel variants or assessing clinical safety in patients exposed to 8-aminoquinolines. In these cases, the derivation of relative G6PD activity using the AMM remains a more appropriate approach. Because assay precision varies considerably between laboratories, the use of replicate measures and control sample measurements is crucial to ensure quality control. Clinical laboratories typically provide the patients’ G6PD value and a normal G6PD range for the laboratory, which is often determined such that it discriminates at or above the 70% threshold required to prescribe tafenoquine. Novel point-of-care assays, such as recently developed quantitative biosensors, are currently being evaluated in field trials. These assays are designed to require less sample preparation and offer more robust performance across diverse temperature ranges and clinical contexts than spectrophotometry. As such, they may provide superior inter-site comparability than the current reference standard of spectrophotometry; however, until this has been shown, routine quantitative diagnosis of G6PDd will require site- and assay-specific local definitions of G6PD activity to ensure that tafenoquine can be administered safely. In any case, no diagnostic assay is perfect, meaning that radical cure treatment policies must be accompanied by patient and health worker training on the warning signs and risks of haemolysis, along with access to transfusion services when needed.

Acknowledgements

We would like to thank all participants of the included studies, as well as all staff who contributed to the primary included articles and assisted in sharing their datasets.

Disclaimer: The views expressed here are solely those of the authors and do not reflect the views, policies or positions of the US Government or Department of Defense. Material has been reviewed by the Walter Reed Army Institute of Research. There is no objection to its presentation and/or publication. The opinions or assertions contained herein are the private views of the author, and are not to be construed as official, or as reflecting true views of the Department of the Army or the Department of Defense. The investigators have adhered to the policies for protection of human subjects as prescribed in AR 70–25.

Abbreviations

AMM

adjusted male median

CV

coefficient of variation

G6PD

glucose-6-phosphate dehydrogenase

G6PDd

G6PD deficient

G6PDn

G6PD normal

NPV

negative predictive value

QUADAS-2

Quality Assessment of Diagnostic Accuracy Studies-2

RBC

red blood cell

UV

ultraviolet

WHO

World Health Organization

References

1 

AA Lover, JK Baird, R Gosling, R Price. . Malaria elimination: Time to target all species. Am J Trop Med Hyg. 2018 Epub 2018/05/16. , doi: 10.4269/ajtmh.17-0869 .

2 

JK Baird, KE Battle, RE Howes. . Primaquine ineligibility in anti-relapse therapy of Plasmodium vivax malaria: the problem of G6PD deficiency and cytochrome P-450 2D6 polymorphisms. Malar J. 2018;17:, pp.42, doi: 10.1186/s12936-018-2190-z PMC5778616.

3 

RE Howes, FB Piel, AP Patil, OA Nyangiri, PW Gething, M Dewi, et al. G6PD deficiency prevalence and estimates of affected populations in malaria endemic countries: a geostatistical model-based map. PLoS Med. 2012;9(11):, pp.e1001339, doi: 10.1371/journal.pmed.1001339

4 

ET Nkhoma, C Poole, V Vannappagari, SA Hall, E Beutler. . The global prevalence of glucose-6-phosphate dehydrogenase deficiency: A systematic review and meta-analysis. Blood Cells, Molecules and Diseases. 2009;42(3):, pp.267–78. , doi: 10.1016/j.bcmd.2008.12.005

5 

L Luzzatto, P Arese. . Favism and glucose-6-phosphate dehydrogenase deficiency. N Engl J Med. 2018;378(1):, pp.60–71. , doi: 10.1056/nejmra1708111

6 

KD Belfield, EM Tichy. . Review and drug therapy implications of glucose-6-phosphate dehydrogenase deficiency. Am J Health Syst Pharm. 2018;75(3):, pp.97–104. , doi: 10.2146/ajhp160961

7 

GJ Domingo, N Advani, AW Satyagraha, CH Sibley, E Rowley, M Kalnoky, et al. Addressing the gender-knowledge gap in glucose-6-phosphate dehydrogenase deficiency: challenges and opportunities. Int Health. 2018;11(1):, pp.7–14. , doi: 10.1093/inthealth/ihy060

8 

S Gómez-Manzo, J Marcial-Quino, A Vanoye-Carlo, H Serrano-Posada, D Ortega-Cuellar, A González-Valdez, et al. Glucose-6-phosphate dehydrogenase: Update and analysis of new mutations around the world. Int J Mol Sci. 2016;17(12):, pp.2069, doi: 10.3390/ijms17122069 PMC5187869.

9 

A Minucci, K Moradkhani, MJ Hwang, C Zuppi, B Giardina, E Capoluongo. . Glucose-6-phosphate dehydrogenase (G6PD) mutations database: Review of the “old” and update of the new mutations. Blood Cells Mol Dis. 2012;48(3):, pp.154–65. , doi: 10.1016/j.bcmd.2012.01.001

10 

L Luzzatto. . Glucose 6-phosphate dehydrogenase deficiency: from genotype to phenotype. Haematologica. 2006;91(10):, pp.1303

11 

AD Cunningham, A Colavin, KC Huang, D Mochly-Rosen. . Coupling between protein stability and catalytic activity determines pathogenicity of G6PD variants. Cell Rep. 2017;18(11):, pp.2592–9. , doi: 10.1016/j.celrep.2017.02.048

12 

B Ley, G Bancone, L von Seidlein, K Thriemer, JS Richards, GJ Domingo, et al. Methods for the field evaluation of quantitative G6PD diagnostics: a review. Malar J. 2017;16(1):, pp.361, doi: 10.1186/s12936-017-2017-3

13 

S Pal, P Bansil, G Bancone, S Hrutkay, M Kahn, G Gornsawun, et al. Evaluation of a novel quantitative test for glucose-6-phosphate dehydrogenase deficiency: bringing quantitative testing for glucose-6-phosphate dehydrogenase deficiency closer to the patient. Am J Trop Med Hyg. 2018:-. , doi: 10.4269/ajtmh.18-0612

14 

MS Alam, MG Kibria, N Jahan, RN Price, B Ley. . Spectrophotometry assays to determine G6PD activity from Trinity Biotech and Pointe Scientific G6PD show good correlation. BMC Res Notes. 2018;11(1):, pp.855, doi: 10.1186/s13104-018-3964-7

15 

E Beutler, KG Blume, JC Kaplan, GW Löhr, B Ramot, WN Valentine. . International Committee for Standardization in Haematology: Recommended Methods for Red-Cell Enzyme Analysis. Br J Haematol. 1977;35(2):, pp.331–40. , doi: 10.1111/j.1365-2141.1977.tb00589.x

16 

GJ Domingo, AW Satyagraha, A Anvikar, K Baird, G Bancone, P Bansil, et al. G6PD testing in support of treatment and elimination of malaria: recommendations for evaluation of G6PD tests. Malar J. 2013;12(1):, pp.391, doi: 10.1186/1475-2875-12-391

17 

World Health Organization. Testing for G6PD deficiency for safe use of primaquine in radical cure of P. vivax and P. ovale malaria. Geneva: WHO, 2016.

18 

MVG Lacerda, A Llanos-Cuentas, S Krudsood, C Lon, DL Saunders, R Mohammed, et al. Single-dose tafenoquine to prevent relapse of Plasmodium vivax malaria. N Engl J Med. 2019;380(3):, pp.215–28. , doi: 10.1056/NEJMoa1710775

19 

A Llanos-Cuentas, MVG Lacerda, TT Hien, ID Vélez, C Namaik-larp, CS Chu, et al. Tafenoquine versus primaquine to prevent relapse of Plasmodium vivax malaria. N Engl J Med. 2019;380(3):, pp.229–41. , doi: 10.1056/NEJMoa1802537

20 

G Henriques, K Phommasone, R Tripura, TJ Peto, S Raut, C Snethlage, et al. Comparison of glucose-6 phosphate dehydrogenase status by fluorescent spot test and rapid diagnostic test in Lao PDR and Cambodia. Malar J. 2018;17(1):, pp.243, doi: 10.1186/s12936-018-2390-6

21 

PF Whiting, AWS Rutjes, ME Westwood, S Mallett, JJ Deeks, JB Reitsma, et al. QUADAS-2: A revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med. 2011;155(8):, pp.529–36. , doi: 10.7326/0003-4819-155-8-201110180-00009

22 

World Health Organization. Technical specifications series for submission to WHO prequalification–diagnostic assessment: in vitro diagnostic medical devices to identify glucose-6-phosphate dehydrogenase (G6PD) activity. Geneva: World Health Organization, 2016.

23 

Wickham H. tidyverse: Easily install and load the ‘tidyverse’ 2017. R package version 1.2.1. Available from: https://CRAN.R-project.org/package=tidyverse

24 

H. Wickhamggplot2: Elegant graphics for data analysis. New York: Springer-Verlag; 2009.

25 

Wilke CO. cowplot: Streamlined plot theme and plot annotations for 'ggplot2' 2019. R package version 0.9.4. Available from: https://CRAN.R-project.org/package=cowplot

26 

Stevenson M. epiR: Tools for the analysis of epidemiological data 2018. R package version 0.9–99. Available from: https://CRAN.R-project.org/package=epiR

27 

R Core Team. R: A Language and Environment for Statistical ComputingVienna, Austria: R Foundation for Statistical Computing; 2018 Available from: https://www.R-project.org.

28 

G Bancone, CS Chu, N Chowwiwat, R Somsakchaicharoen, P Wilaisrisak, P Charunwatthana, et al. Suitability of capillary blood for quantitative assessment of G6PD activity and performances of G6PD point-of-care tests. Am J Trop Med Hyg. 2015;92(4):, pp.818–24. Epub 2015/02/04. , doi: 10.4269/ajtmh.14-0696

29 

NN Oo, G Bancone, LZ Maw, N Chowwiwat, P Bansil, GJ Domingo, et al. Validation of G6PD point-of-care rests among healthy volunteers in Yangon, Myanmar. PLoS ONE. 2016;11(4):, pp.e0152304–e. , doi: 10.1371/journal.pone.0152304 .

30 

AW Satyagraha, A Sadhewa, R Elvira, I Elyazar, D Feriandika, U Antonjaya, et al. Assessment of point-of-care diagnostics for G6PD deficiency in malaria endemic rural Eastern Indonesia. PLoS Negl Trop Dis. 2016;10(2):, pp.e0004457, doi: 10.1371/journal.pntd.0004457

31 

SF Travis, SP Kumar, PC Paez, M Delivoria-Papadopoulos. . Red cell metabolic alterations in postnatal life in term infants. Glycolytic enzymes and glucose-6-phosphate dehydrogenase. Pediatr Res. 1980;14(12):, pp.1349, doi: 10.1203/00006450-198012000-00016

32 

W-C Yang, S Tai, C-L Hsu, C-M Fu, A-K Chou, P-L Shao, et al. Reference levels for glucose-6-phosphate dehydrogenase enzyme activity in infants 7–90 days old in Taiwan. J Formos Med Assoc. 2019, doi: 10.1016/j.jfma.2019.03.010.

33 

R Rueangweerayut, G Bancone, EJ Harrell, AP Beelen, S Kongpatanakul, JJ Mohrle, et al. Hemolytic potential of tafenoquine in female volunteers heterozygous for glucose-6-phosphate dehydrogenase (G6PD) deficiency (G6PD Mahidol variant) versus G6PD-normal volunteers. Am J Trop Med Hyg. 2017;97(3):, pp.702–11. Epub 2017/07/28. , doi: 10.4269/ajtmh.16-0779 .

34 

MS Alam, MG Kibria, N Jahan, K Thriemer, MS Hossain, NM Douglas, et al. Field evaluation of quantitative point of care diagnostics to measure glucose-6-phosphate dehydrogenase activity. PLoS ONE. 2018;13(11):, pp.e0206331, doi: 10.1371/journal.pone.0206331

35 

CA Major, L Pretlow, E Fincher, V Koziatek. . Validating the assessment of glucose-6-phosphate dehydrogenase (G6PD). Clin Lab Sci. 2006;19(3):, pp.134–9.

36 

WHO Working Group. . Glucose-6-phosphate dehydrogenase deficiency. Bull World Health Organ. 1989;67(6):, pp.601–11. PMC2491315.

37 

RJ Commons, JS McCarthy, RN Price. . Tafenoquine for the radical cure and prevention of malaria: the importance of testing for G6PD deficiency. Med J Aust. 2020;212(4):, pp.152–3.e1.

38 

P Charoenkwan, W Tantiprabha, S Sirichotiyakul, A Phusua, T Sanguansermsri. . Prevalence and molecular characterization of glucose-6-phosphate dehydrogenase deficiency in northern Thailand. Southeast Asian J of Trop Med Public Health. 2014;45(1):, pp.187–93. Epub 2014/06/27. .

39 

ME Roh, C Oyet, P Orikiriza, M Wade, J Mwanga-Amumpaire, Y Boum 2nd, et al. Screening for glucose-6-phosphate dehydrogenase deficiency using three detection methods: A cross-sectional survey in Southwestern Uganda. Am J Trop Med Hyg. 2016;95(5):, pp.1094–9. , doi: 10.4269/ajtmh.16-0552 .


8 Oct 2019

Dear Dr. Pfeffer,

Thank you very much for submitting your manuscript "Quantification of glucose-6-phosphate dehydrogenase activity by spectrophotometry: A systematic review and meta-analysis" (PMEDICINE-D-19-02590) for consideration at PLOS Medicine.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to three independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by Oct 29 2019 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see http://journals.plos.org/plosmedicine/s/submission-guidelines#loc-methods.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Thomas McBride, PhD

Senior Editor

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

1- Thank you for providing your PRISMA statement. Please replace the page numbers with paragraph numbers per section (e.g. "Methods, paragraph 1"), since the page numbers of the final published paper may be different from the page numbers in the current manuscript.

2- Thank you for agreeing to include the aggregate data in the supplementary files. However, I did not see a data file, please include this with your revision. Additionally, it would be helpful to include a supplementary text file listing the data contacts for each of the studies included in your analysis.

3- In the Abstract Methods and findings, please provide the dates of search, data sources, types of study designs included, eligibility criteria, and synthesis/appraisal methods.

4- In the last sentence of the Abstract Methods and Findings section, please describe the main limitation(s) of the study's methodology.

5- Thank you for including an Author Summary. Please reformat the Author Summary into 2-3 single sentence bullet points for each of the following questions. Bullet points should be objective, brief, succinct, specific, accurate, and avoid technical language.

>Why Was This Study Done? Authors should reflect on what was known about the topic before the research was published and why the research was needed.

>What Did the Researchers Do and Find? Authors should briefly describe the study design that was used and the study’s major findings. Do include the headline numbers from the study, such as the sample size and key findings.

>What Do These Findings Mean? Authors should reflect on the new knowledge generated by the research and the implications for practice, research, policy, or public health. Authors should also consider how the interpretation of the study’s findings may be affected by the study limitations.

Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary

Comments from the reviewers:

Reviewer #2: Pfeffer and colleagues have presented the results of a systematic review and meta-analyses of individual participant data evaluating lab-based variation in diagnostic results for spectrophotometry derived diagnosis for G6PD deficiency malaria and non-malaria samples. They further pool data to derive a universal adjusted male median (AMM) from a sub-set of samples derived from Trinity assays and apply this threshold to define G6PD deficiency compared to lab-specific thresholds as the reference standard. This resulted in identifying deficiency with 89-100% sensitivity and 96-100% specificity. I commend the authors for a large undertaking in collating the international study data globally and the results could potentially have some policy implications on determining a universal reference standard. There are a few comments I have on the methodology, in particular relating to how the samples were pooled that should be clarified and addressed prior to publication.

The main issue the authors should clarify is that it isn't clear how the authors pooled the AMM in the IPD and how they accounted for between study heterogeneity (starting 257). The authors are rather vague in their presentation of the statistical methods here: "A pooled universal AMM was calculated across a comparable subset of all included samples." Given the challenges in pooling medians across studies (non-normality assumed), I would presume the authors would have had to calculate a AMM across the entirety of the pooled sample directly from the pooled data in one-stage, and hence one of the limitations may in fact be not being able to account for between study-heterogeneity (as the dominant IPD methodology if pooling means would be to account for this in an error-term in the regression model). This would be an important consideration/limitation as that 75% of the samples come from Asia (Table 1) and study labs have large variations: the authors report range of difference from 0.65- 8.64 with coefficient of variation between 1.6% to 14.9%. Inter-replicate difference was significantly difference between studies. Despite showing that 5/6 assays were from Trinity assay - this still had high levels of variability.

This means that in Line 263 where a conservative threshold was derived as the upper bound of the 95% CI of the mean G6PD activity, the pooled mean G6DP activity should in fact take into consideration between study differences. As this can be derived from a pooled regression analyses, covariate adjustments, study heterogeneity can be incorporated into a hierarchical model either in one-stage of two-stages. It's unclear how the authors pooled the mean here as well and it would beneficial to clarify the methods.

Finally, it's nice to see the authors give an overall pooled sensitivity and specificity (Lines 365-367), but it's also not described in the methods how this was done and if study-level covariates were incorporated in these pooled sensitivity and specificity. For completeness of presentation, it would be also useful to present PPV, NPV at varying prevalence and derive an overall accuracy measure summary ROC plot. See Riley et al. 2008 (https://onlinelibrary.wiley.com/doi/epdf/10.1002/sim.3441) and Cochrane handbook suggests having this metric: https://methods.cochrane.org/sites/methods.cochrane.org.sdt/files/public/uploads/Chapter%2010%20-%20Version%201.0.pdf

Other items to address:

Line 206: It's still not clear why the lower bound for was 2005 for laboratory changes and methodology. Was there some particular change around this time or was this just a pragmatic choice to limit the number of studies?

Line starting 243: The analysis uses AMM to define 100% G6PD activity, then uses WHO definitions to define categories for < 30% of the AMM is deficiency and activity between 30-70% for intermediate from tafenoquine clinical trials and not the recent WHO definition between 30-80% activity. It would be advantageous to benchmark also against the recent WHO definition as this should enhance the results policy implications (30-80% definition for intermediate activity)

Despite these suggestions above - this is a comprehensive review and should warrant consideration for publication.

Reviewer #3: This is an important and timely paper as countries with Vivax malaria especially those aiming to eliminate malaria within relatively short time frames have to address the recurrent nature of vivax infections in areas where G6DPdeficiencies are common but vary widely in nature. Vivax also occurs now mainly in remote and mobile and migrant populations which require management by lower level health workers so the aim to develop a universal gold standard against which to measure any new tests is important. With the introduction of the only new drug to be developed to address the presence of hypnozoites this is even more urgent.

The number of studies and the range of countries where studies were done give a range of countries with different prevalences and varieties of G6DPd but did not include data from countries with very high rates and severe varieties such as Pakistan.

In light of the fact that universal testing of newborns for G6DP is being suggested and indeed implemented in some countries the important variation in sensitivity and specificity of the tests in newborns is an important area to explore further.

Although patients with and without malaria were looked at it might have been interesting to know if other blood dyscrasias common in these populations such as thalassemia and sickle cell disease had an influence on the outcome of the tests if any

The use of the term gold standard seems a little ambitious as the standard cannot be a universal standard if the standard has to be set for the different assays and in different populations as seems to have to be the case. The intra-laboratory variation as well as the variation across different assay types requires that quality control is especially important and this is a very useful fact to emphasise especially in countries where quality control systems are weak.

When introducing the drugs in line 162 It might have been a good idea to explain the different ways these drugs are deployed ie Primaquine is used as a single low dose to deal with gametocytes of Pf which WHO says is safe for all levels of G6DP activity (although in countries with very weak pharmacovigilance systems this contention may not be completely true) Primaquine for 14 or 7 days (or over 8 weeks) for management of hypnozoites and tafenoquine which is a single dose and therefore carries more hazards and requires a higher degree of safety measures before administration.

When relying on qualitative tests ( those are the only ones universally available at the moment to National programmes the sensitivity and specificity of these tests against a locally derived gold standard is important Even when Primaquine is in the national protocol many health workers are reluctant to prescribe this drug due to perhaps exaggerated fears of Haemolysis. However good a test is it can never be perfect and so cases of haemolysis will occur and when training health workers and drug recipients it will be necessary to advise patient of how to detect haemolysis and provide access to transfusion services when needed. This will be even more important when tafenoquine is used as management of drug induced haemolysis will be even more challenging

Any attachments provided with reviews can be seen via the following link:

[LINK]

Submitted filename: PMEDICINE-PaulEder.docx

29 Oct 2019

Submitted filename: Pfeffer_et_al_G6PDSpectro_Response.docx

10 Feb 2020

Dear Dr. Pfeffer,

Thank you very much for re-submitting your manuscript "Quantification of glucose-6-phosphate dehydrogenase activity by spectrophotometry: A systematic review and meta-analysis" (PMEDICINE-D-19-02590R1) for consideration at PLOS Medicine.

I have discussed the paper with editorial colleagues and our academic editor and it was also seen again by one reviewer. I am pleased to tell you that, provided the remaining editorial and production issues are dealt with, we expect to be able to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

Our publications team (plosmedicine@plos.org) will be in touch shortly about the production requirements for your paper, and the link and deadline for resubmission. DO NOT RESUBMIT BEFORE YOU'VE RECEIVED THE PRODUCTION REQUIREMENTS.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We hope to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

Please let me know if you have any questions. Otherwise, we look forward to receiving your revised manuscript shortly.

Sincerely,

Richard Turner PhD, for Thomas McBride, PhD

Senior Editor, PLOS Medicine

rturner@plos.org

------------------------------------------------------------

Requests from Editors:

To your abstract, please add summary demographic details for study participants.

In your abstract and elsewhere, please add p values alongside 95% CI where available.

At line 122, would "... lack of availability of covariate data ..." be more appropriate?

At line 124, please begin the sentence "Our findings indicate that ..." or similar.

If appropriate, please adapt the wording at the start of the Methods section of your main text (around line 239 where you refer to the PROSPERO record) to note that the study was prospectively planned

Please refer to the attached PRISMA checklist in the Methods section.

Towards the end of the Methods section, please add a sentence to note that specific ethics approval was not required for the present analysis, and if available the identity of the body responsible for that assessment.

At line 305, please adapt the text "a total of ... sample data" to indicate whether this refers to number of individuals.

We suspect that the reference call-outs in table 2 should not be in superscript.

Please adapt reference call-outs to this style: "... endemic regions [3,4].".

Please add full access details to reference 7.

Comments from Academic Editor:

This is an interesting, pertinent, and well conducted study in an area with high levels of interest in malaria and concomitant funding. This analysis provides one part of the picture in moving forward. It’s clearly written and relatively short. I have only minor comments about language in a couple of places.

I agree with previous comments of referees that the authors should not refer to spectrophotometry as the “Gold standard”; it is not, this is a notoriously difficult condition to measure, there are so many variants, and the functional significance, in relation to the extent of the haemolysis with with tafenoquine, is really not clear. Thus “reference standard” should be used throughout.

The authors conclude that there is “substantial heterogeneity in G6PD measurements between sites…” I would encourage they describe this as “substantial variation in….”. The word heterogeneity implies that G6PD actually does vary sites but in fact it appears that it is the result largely of laboratory measures varying.

Under, what did the researchers find, third bullet, please try and avoid the use of the word “significant”. This is an ambiguous word, it can be qualitatively important, or it can mean some statistical evaluation using P 0.05. “Important” or “considerable” is more helpful.

On line 189, the authors don’t quite communicate to the reader that the concern with tafenoquine is the long half-life, and thus the persistence of the harmful haemolytic effect with this drug-it may be worse than PQ.

Otherwise the message is clear

Comments from Reviewers:

*** Reviewer #2:

The revised version of the manuscript is much improved. The authors have responded appropriately the reviewers comments and have substantiated their responses well. The methods are clearer now related to the pooling process and the authors have described the limitations of obtaining study-level covariates to explore heterogeneity. Thank you for also producing summary ROC curves in supplemental, stratified by gender and threshold. On this basis, my recommendation would for acceptance for publication, given two minor points:

1) Thank you for the detailed response to the IPD methodology. The authors mention that "had we adjusted our pooled threshold with the covariates available, this would have meant that practitioners would also have to adjust their assay cut-offs based upon local factors - a level of complexity which is unlikely to be practical or offer utility beyond current practice." This is an interesting point and I thought warrants inclusion in the methodology to strengthen their rationale.

2) I thank the authors for educating me on the WHO guidelines being out of date - I think this also is a nice point that should have a place in the discussion - The policy implications are that both thresholds being inadequate.

***

Any attachments provided with reviews can be seen via the following link:

[LINK]


13 Apr 2020

Submitted filename: Pfeffer_et_al_G6PDSpectro_Response.docx

13 Apr 2020

Dear Dr. Pfeffer,

On behalf of my colleagues and the academic editor, Dr. Paul Garner, I am delighted to inform you that your manuscript entitled "Quantification of glucose-6-phosphate dehydrogenase activity by spectrophotometry: A systematic review and meta-analysis" (PMEDICINE-D-19-02590R2) has been accepted for publication in PLOS Medicine.

PRODUCTION PROCESS

Before publication you will see the copyedited word document (in around 1-2 weeks from now) and a PDF galley proof shortly after that. The copyeditor will be in touch shortly before sending you the copyedited Word document. We will make some revisions at the copyediting stage to conform to our general style, and for clarification. When you receive this version you should check and revise it very carefully, including figures, tables, references, and supporting information, because corrections at the next stage (proofs) will be strictly limited to (1) errors in author names or affiliations, (2) errors of scientific fact that would cause misunderstandings to readers, and (3) printer's (introduced) errors.

If you are likely to be away when either this document or the proof is sent, please ensure we have contact information of a second person, as we will need you to respond quickly at each point.

PRESS

A selection of our articles each week are press released by the journal. You will be contacted nearer the time if we are press releasing your article in order to approve the content and check the contact information for journalists is correct. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact.

PROFILE INFORMATION

Now that your manuscript has been accepted, please log into EM and update your profile. Go to https://www.editorialmanager.com/pmedicine, log in, and click on the "Update My Information" link at the top of the page. Please update your user information to ensure an efficient production and billing process.

Thank you again for submitting the manuscript to PLOS Medicine. We look forward to publishing it.

Best wishes,

Thomas McBride, PhD

Senior Editor

PLOS Medicine

plosmedicine.org

https://www.researchpad.co/tools/openurl?pubtype=article&doi=10.1371/journal.pmed.1003084&title=Quantification of glucose-6-phosphate dehydrogenase activity by spectrophotometry: A systematic review and meta-analysis&author=Daniel A. Pfeffer,Benedikt Ley,Rosalind E. Howes,Patrick Adu,Mohammad Shafiul Alam,Pooja Bansil,Yap Boum,Marcelo Brito,Pimlak Charoenkwan,Archie Clements,Liwang Cui,Zeshuai Deng,Ochaka Julie Egesie,Fe Esperanza Espino,Michael E. von Fricken,Muzamil Mahdi Abdel Hamid,Yongshu He,Gisela Henriques,Wasif Ali Khan,Nimol Khim,Saorin Kim,Marcus Lacerda,Chanthap Lon,Asrat Hailu Mekuria,Didier Menard,Wuelton Monteiro,François Nosten,Nwe Nwe Oo,Sampa Pal,Duangdao Palasuwan,Sunil Parikh,Ayodhia Pitaloka Pasaribu,Jeanne Rini Poespoprodjo,David J. Price,Arantxa Roca-Feltrer,Michelle E. Roh,David L. Saunders,Michele D. Spring,Inge Sutanto,Kamala Ley-Thriemer,Thomas A. Weppelmann,Lorenz von Seidlein,Ari Winasti Satyagraha,Germana Bancone,Gonzalo J. Domingo,Ric N. Price,Paul Garner,Thomas J. McBride,Thomas J. McBride,Thomas J. McBride,&keyword=&subject=Research Article,Medicine and Health Sciences,Hematology,Anemia,Hemolytic Anemia,Glucose-6-Phosphate Dehydrogenase Deficiency,Medicine and Health Sciences,Parasitic Diseases,Malaria,Medicine and Health Sciences,Tropical Diseases,Malaria,Research and Analysis Methods,Spectrum Analysis Techniques,Spectrophotometry,Medicine and Health Sciences,Pharmacology,Drugs,Primaquine,Biology and Life Sciences,Parasitology,Parasite Groups,Apicomplexa,Plasmodium,Engineering and Technology,Industrial Engineering,Quality Control,Medicine and Health Sciences,Hematology,Medicine and Health Sciences,Pharmacology,Drugs,Antimalarials,