• View in gallery
    Figure 1.

    The locations of the three schools are presented in relation to the national highway system, the national capital, and the study communes of Gressier and Leogane, located in the Ouest Department of Haiti. This figure appears in color at www.ajtmh.org.

  • View in gallery
    Figure 2.

    The distributions of the glucose-6-phosphate dehydrogenase (G6PD) activity (U/g Hgb) from the sample population measured using the spectrophotometric assay and the biosensor are presented with the classifications for G6PD deficiency (dotted lines from left to right): severe (< 10%), moderate (10–30%), mild (30–60%), normal (60–150%), and increased (> 150%) activity. This figure appears in color at www.ajtmh.org.

  • View in gallery
    Figure 3.

    A Bland–Altman plot is used to compare the difference between the two glucose-6-phosphate dehydrogenase activity (U/g Hgb) values obtained from the spectrophotometric assay and the biosensor with the average of the two values and their distribution with respect to the mean difference and two standard deviations from the mean difference (dotted lines). This figure appears in color at www.ajtmh.org.

  • View in gallery
    Figure 4.

    The distributions of the glucose-6-phosphate dehydrogenase (G6PD) activity (U/g Hgb) from the sample population measured using the biosensor and the spectrophotometric assay are presented separately for males and females with the threshold values (dotted lines) at which drug-induced hemolysis can occur (30% residual activity, 2.624 U/g Hgb) and the threshold for G6PD deficiency (60% residual activity, 5.249 U/g Hgb). This figure appears in color at www.ajtmh.org.

  • View in gallery
    Figure 5.

    A scatterplot of the glucose-6-phosphate dehydrogenase activity (U/g Hgb) measured using the spectrophotometric assay and the biosensor is presented with respect to the number of copies of the non-A-allele: hemizygous males or homozygous females with only A-alleles (red), heterozygous females with one non-A-allele (yellow), and homozygous females or hemizygous males with only non-A-alleles (gray). This figure appears in color at www.ajtmh.org.

  • View in gallery
    Figure 6.

    The population distributions of glucose-6-phosphate dehydrogenase (G6PD) activity measured by the spectrophotometric method and the biosensor are presented with respect to G6PD A-hemizygous males and homozygous females (red bars) and heterozygous females/normal population members (gray bars). The receiver operator characteristics and area under the curve (AUC) are presented for the spectrophotometric assay (purple, left) and the biosensor (green, right) to illustrate the sensitivity and specificity of the assays to correctly classify participants with only A-alleles (homozygous A-females and hemizygous A-males) who are at the highest risk for drug induced hemolysis. The gray dotted line shows the characteristics of a test that is not more informative than random chance (AUC = 0.5). This figure appears in color at www.ajtmh.org.

  • View in gallery
    Figure 7.

    Scatterplots of the glucose-6-phosphate dehydrogenase activity readings using the biosensor (U/dL) in the laboratory compared with the field are presented by school and overall with the R2 values from a linear regression of the laboratory and field values. This figure appears in color at www.ajtmh.org.

  • 1.

    World Health Organization, 2015. World Malaria Report. Geneva, Switzerland: WHO. Available at: http://www.who.int/malaria/publications/world-malaria-report-2015/report/en/. Accessed December 8, 2016.

  • 2.

    Mikolajczak SA et al.., 2015. Plasmodium vivax liver stage development and hypnozoite persistence in human liver-chimeric mice. Cell Host Microbe 17: 526553.

    • Search Google Scholar
    • Export Citation
  • 3.

    Galappaththy GNL, Tharyan P, Kirubakaran R, 2013. Primaquine for preventing relapse in people with Plasmodium vivax malaria treated with chloroquine. Cochrane Database Syst Rev 10: CD004389.

    • Search Google Scholar
    • Export Citation
  • 4.

    Beutler E, 1996. G6PD: population genetics and clinical manifestations. Blood Rev 10: 4552.

  • 5.

    Mason PJ, Bautista JM, Gilsanz F, 2007. G6PD deficiency: the genotype-phenotype association. Blood Rev 21: 267283.

  • 6.

    Luzzatto L, Metha A, Vulliamy T, 2001. Glucose-6-phosphate dehydrogenase deficiency. Scriver CR, Beaudet AL, Sly WS, eds. The Metabolic and Molecular Bases of Inherited Disease, 9th edition. Columbus, OH: McGraw-Hill, 4517–4553.

  • 7.

    Beutler E, Duparc S, 2007. Glucose-6-phosphate dehydrogenase deficiency and antimalarial drug development. Am J Trop Med Hyg 77: 779789.

  • 8.

    World Health Organization, 2015. Point-of-care G6PD testing to support safe use of primaquine for the treatment of vivax malaria. Geneva, Switzerland: WHO. Available at: http://www.who.int/malaria/mpac/mpac-march2015-erg-g6pd.pdf. Accessed November 10, 2016.

  • 9.

    Osorio L et al.., 2015. Performance of binaxNOW G6PD deficiency point-of-care diagnostic in P. vivax-infected subjects. Am J Trop Med Hyg 92: 2227.

    • Search Google Scholar
    • Export Citation
  • 10.

    Kim S et al.., 2011. Performance of the CareStart G6PD deficiency screening test, a point-of-care diagnostic for primaquine therapy screening. PLoS One 6: e28357.

    • Search Google Scholar
    • Export Citation
  • 11.

    Access Bio, Inc. Manufacturer Information and Instruction for the Operation of the CareStartTM G6PD Biosensor Analyzer. Available at: http://www.accessbio.net/eng/products/products03.asp. Accessed January 1, 2017.

  • 12.

    World Health Organization, 1989. Glucose-6-phosphate dehydrogenase deficiency. WHO Working Group. Bull World Health Organ 67: 601611.

    • Search Google Scholar
    • Export Citation
  • 13.

    Howes RE et al.., 2013. Spatial distribution of G6PD deficiency variants across malaria-endemic regions. Malar J 12: 418426.

  • 14.

    Carter TE, Maloy H, von Fricken ME, St Victor Y, Romain JR, Okech BA, 2014. Glucose-6-phosphate dehydrogenase deficiency A-variant in febrile patients in Haiti. Am J Trop Med Hyg 91: 412414.

    • Search Google Scholar
    • Export Citation
  • 15.

    Tishkoff SA et al.., 2001. Haplotype diversity and linkage disequilibrium at human G6PD: recent origin of alleles that confer malarial resistance. Science 293: 455462.

    • Search Google Scholar
    • Export Citation
  • 16.

    Landis JR, Koch GG, 1977. The measurement of observer agreement for categorical data. Biometrics 33: 159174.

  • 17.

    Weppelmann TA, Carter TE, Chen Z, von Fricken ME, Victor YS, Existe A, Okech BA, 2013. High frequency of the erythroid silent Duffy antigen genotype and lack of Plasmodium vivax infections in Haiti. Malar J 12: 30.

    • Search Google Scholar
    • Export Citation
  • 18.

    Weppelmann TA, von Fricken ME, Lam B, Telisma T, Existe A, Lemoine JF, Larkin J III, Okech BA, 2016. Sparse serological evidence of Plasmodium vivax transmission in the Ouest and Sud-Est departments of Haiti. Acta Trop 162: 2734.

    • Search Google Scholar
    • Export Citation
  • 19.

    von Fricken ME, Weppelmann TA, Lam B, Eaton WT, Schick L, Masse R, Beau De Rochars MV, Existe A, Larkin J III, Okech BA, 2014. Age-specific malaria seroprevalence rates: a cross-sectional analysis of malaria transmission in the Ouest and Sud-Est departments of Haiti. Malar J 13: 361.

    • Search Google Scholar
    • Export Citation
  • 20.

    von Fricken ME, Weppelmann TA, Eaton WT, Alam MT, Carter TE, Schick L, Masse R, Romain JR, Okech BA, 2014. Prevalence of glucose-6-phosphate dehydrogenase (G6PD) deficiency in the Ouest and Sud-Est departments of Haiti. Acta Trop 135: 6266.

    • Search Google Scholar
    • Export Citation
  • 21.

    von Fricken ME, Weppelmann TA, Eaton WT, Masse R, Beau de Rochars MV, Okech BA, 2014. Performance of the CareStart glucose-6-phosphate dehydrogenase (G6PD) rapid diagnostic test in Gressier, Haiti. Am J Trop Med Hyg 91: 7780.

    • Search Google Scholar
    • Export Citation
  • 22.

    Domingo GJ et al.., 2013. G6PD testing in support of treatment and elimination of malaria: recommendations for evaluation of G6PD tests. Malar J 12: 391.

    • Search Google Scholar
    • Export Citation
  • 23.

    Luzzatto L, Seneca E, 2013. G6PD deficiency: a classic example of pharmacogenetics with on-going clinical implications. Brit J Haem 164: 469480.

    • Search Google Scholar
    • Export Citation
Past two years Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 263 121 10
PDF Downloads 124 48 4
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 

 

 

Field Trial of the CareStart Biosensor Analyzer for the Determination of Glucose-6-Phosphate Dehydrogenase Activity in Haiti

Thomas A. WeppelmannHerbert Wertheim College of Medicine, Florida International University, Miami, Florida;

Search for other papers by Thomas A. Weppelmann in
Current site
Google Scholar
PubMed
Close
,
Michael E. von FrickenDepartment of Global and Community Health, George Mason University, Fairfax, Virginia;

Search for other papers by Michael E. von Fricken in
Current site
Google Scholar
PubMed
Close
,
Tara D. WilfongDepartment of Environmental and Global Health, University of Florida, Gainesville, Florida;
Emerging Pathogens Institute, University of Florida, Gainesville, Florida;

Search for other papers by Tara D. Wilfong in
Current site
Google Scholar
PubMed
Close
,
Elisa AguenzaDepartment of Global and Community Health, George Mason University, Fairfax, Virginia;

Search for other papers by Elisa Aguenza in
Current site
Google Scholar
PubMed
Close
,
Taina T. PhilippeChristianville Foundation Clinic, Gressier, Haiti

Search for other papers by Taina T. Philippe in
Current site
Google Scholar
PubMed
Close
, and
Bernard A. OkechDepartment of Environmental and Global Health, University of Florida, Gainesville, Florida;
Emerging Pathogens Institute, University of Florida, Gainesville, Florida;

Search for other papers by Bernard A. Okech in
Current site
Google Scholar
PubMed
Close

Throughout many developing and tropical countries around the world, malaria remains a significant threat to human health. One barrier to malaria elimination is the ability to safely administer primaquine chemotherapy for the radical cure of malaria infections in populations with a high prevalence of glucose-6-phosphate dehydrogenase (G6PD) deficiency. In the current study, a field trial of the world’s first quantitative, point-of-care assay for measuring G6PD activity was conducted in Haiti. The performance of the CareStart Biosensor Analyzer was compared with the gold standard spectrophotometric assay and genotyping of the G6PD allele in schoolchildren (N = 343) from the Ouest Department of Haiti. In this population, 19.5% of participants (67/343) had some form of G6PD deficiency (< 60% residual activity) and 9.9% (34/343) had moderate-to-severe G6PD deficiency (< 30% residual activity). Overall, 18.95% of participants had the presence of the A-allele (65/343) with 7.87% (27/343) considered at high risk for drug-induced hemolysis (hemizygous males and homozygous females). Compared with the spectrophotometric assay, the sensitivity and specificity to determine participants with < 60% residual activity were 53.7% and 94.6%, respectively; for participants with 30% residual activity, the sensitivity and specificity were 5.9% and 99.7%, respectively. The biosensor overestimated the activity in deficient individuals and underestimated it in participants with normal G6PD activity, indicating the potential for a systematic measurement error. Thus, we suggest that the current version of the biosensor lacks adequate sensitivity and should be improved prior to its use as a point-of-care diagnostic for G6PD deficiency.

INTRODUCTION

In many parts of the world, malaria remains a significant threat to public health, with an estimated incidence of more than 200 million cases in 2015.1 Along with drug-resistant Plasmodium falciparum infections, Plasmodium vivax malaria represents another barrier to malaria elimination due to the formation of dormant hypnozoite stages in hepatocytes, which can lead to recurrent infections and facilitate ongoing transmission.2 Currently, a 14-day course of primaquine is the only treatment able to eradicate P. vivax liver hypnozoites and one of few drugs with gametocidal activity against P. falciparum.3 However, primaquine therapy is contraindicated in patients with glucose-6-phosphate dehydrogenase (G6PD) deficiency, which is common in malaria-endemic regions.4 G6PD deficiency is an X-linked disorder with multiple clinical phenotypes resulting from loss-of-function mutations in the G6PD gene (band Xq28).5 G6PD catalyzes the first step in the pentose–phosphate pathway that produces nicotinamide adenine dinucleotide phosphate (NADPH), allowing for the regeneration of reduced glutathione and the ability to resist intracellular oxidative stress.6 Because erythrocytes lack mitochondria, treatment with chemotherapeutic agents that induce oxidative stress, such as primaquine, rapidly deplete NADPH and glutathione, leading to acute hemolytic anemia (AHA) that can be fatal.7 For these reasons, the research and development of rapid, point-of-care diagnostics for G6PD deficiency represent a critical step in the process toward malaria elimination in countries where G6PD is common.8 Although multiple screening tests currently exist for the determination of G6PD deficiency, only two are designed to be used at the point of care (BinaxNow G6PD Test, Alere; CareStart G6PD RDT, Access Bio, Somerset, NJ); neither of which gives a quantitative measure of G6PD activity.9,10 In this study, the performance of the CareStart G6PD Biosensor Analyzer, the world’s first quantitative, point-of-care screening test was compared with gold standard methods for the determination of G6PD activity in both field and laboratory settings.

MATERIALS AND METHODS

Study location and sample collection.

G6PD activity was measured in 343 children (169 males and 174 females) enrolled from three primary schools, Tiboukan (school A), Tikouzen (school B), and Jean-Jean (school C), located in the communes of Gressier and Leogane, in the Ouest Department of Haiti (Figure 1). After informed consent was obtained by the parent and/or legal guardian, 3 mL of venous blood was drawn into collection tubes (Lavender Vacutainers; BD Diagnostics, Franklin Lakes, NJ) and stored at 4°C for analyses. After measurement of the hemoglobin concentration using a digital hemoglobin meter each sample was screened with the both the CareStart™ G6PD Biosensor Analyzer and the Trinity Biotech quantitative G6PD assay. Relevant demographic characteristics from this population are summarized in Table 1, which included a population composed of 49.3% male and 50.7% female students with an average age of 8.7 ± 3 years and an average hemoglobin concentration of 11.4 ± 1.6 g/dL.

Figure 1.
Figure 1.

The locations of the three schools are presented in relation to the national highway system, the national capital, and the study communes of Gressier and Leogane, located in the Ouest Department of Haiti. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene 97, 4; 10.4269/ajtmh.16-0714

Table 1

Demographic characteristics of study populations

Enrollment locationObserved (n)GenderAge (years)Hemoglobin (g/dL)
(% Male)(% Female)MeanSDMeanSD
School A10350.549.59.92.611.62.3
School B13050.050.08.33111
School C11346.953.18311.51.4
Total34349.350.78.7311.41.6

SD = standard deviation. The number of schoolchildren observed (n), their demographic statistics of age and gender, and hemoglobin concentration are presented by school and for the entire sample.

Ethical approval for research on human subjects.

This study was approved by the Haiti Ethical Review Committee and the University of Florida’s Institutional Review Board. Informed consent (in Haitian Creole) was obtained by physicians and health-care workers from the parents or legal guardians of all primary school children prior to participation in this study.

Determination of G6PD activity and hemoglobin concentration.

The CareStart G6PD Biosensor Analyzer (Cat. No. BGB-E00182; Access Bio) was used to screen each study participant for G6PD activity following the instructions provided by the manufacturer.11 The biosensor uses an electrochemical method to measure enzyme activity, where G6PD present in the sample reacts with a substrate causing an electron transfer between the donor and acceptor molecules. The resulting electric current produced is directly proportional in magnitude to the level of G6PD activity in the blood sample. Only three steps are required to use the device: 1) an unused test strip (Cat. No. BGS-E02582) is inserted into the biosensor, 2) an aliquot from each blood sample (∼5 µL) is placed on the exposed end of the test strip until the device indicates (by beeping and visual cue) that the automated sample intake is complete, and 3) approximately 4 minutes after sample loading, the G6PD activity in U/dL is displayed on the biosensor screen. The Trinity Biotech G6PD Assay (Kit No. 345B; Trinity Biotech, St. Louis, MO) was also used to analyze G6PD activity according to the manufacturer guidelines. The assay uses a spectrophotometric method to quantify G6PD enzyme activity by measuring the amount of NADPH produced in the presence of a known concentration of substrate (glucose-6-phosphate) and the G6PD present in the blood sample. The change in absorbance (λ = 340 nm) after 5 minutes of incubation at a constant temperature (25°C) was recorded and used to calculate the enzyme activity. All samples were analyzed within 24 hours of collection with quality assurance maintained by running control blood samples with normal, intermediate, and deficient levels of G6PD provided by the manufacturer (Cat. No. G6888, G5029, and G5888). The G6PD activity was standardized to the hemoglobin concentration present in each sample after measurement by a digital hemoglobin meter (HemoCue Hb 201 Plus; Hemocue Incorporated, Cypress, CA) for both assays. Additionally, 20 µL of whole blood was spotted onto filter paper cards (Whatman 903 Protein Saver, GE Healthcare, Pittsburgh, PA), allowed to dry, and stored for genotypic analyses. Along with the use of the biosensor in the field at the time of sample collection, the biosensor and spectrophotometric assay were run in duplicate on each sample in the laboratory, with the average of the two values for each assay used for subsequent statistical analyses.

Classification of G6PD activity.

The mean G6PD activity for this population was calculated by taking the average of nonanemic participants (> 12 g/dL hemoglobin) after exclusion of population members with G6PD activity below the lower limit of the manufacturer range for normal activity (4.6 U/g Hgb).12 Using this method, the mean G6PD activity level in nondeficient population members was 8.75 U/g Hgb, which was used to determine the following G6PD activity level classifications relative to the mean (residual activity): severe deficiency with enzyme activity less than 10% residual activity (less than 0.875 U/g Hgb), moderate deficiency with enzyme activity between 10% and 30% residual activity (0.875–2.624 U/g Hgb), mild deficiency with enzyme activity between 30% and 60% residual activity (2.624–5.249 U/g Hgb), normal with enzyme activity between 60% and 150% residual activity (5.249–13.121 U/g Hgb), and increased activity with enzyme activity > 150% residual activity (above 13.121 U/g Hgb).

Determination of G6PD genotype.

Though clinical G6PD deficiency has a considerable amount of allelic heterogeneity, Haitians descended almost exclusively from west African populations where the G6PD A-variant is the most common genotype associated with G6PD deficiency, which is supported by a previous study in the Ouest Department that identified a high population frequency of the A-allele.13,14 As such, the prevalence of G6PD-deficient A-alleles was examined by restriction fragment length polymorphism using a procedure described previously.15 Genomic DNA was extracted and purified from dried blood spots on cards using a bead-based extraction kit (ZR DNA-Card Extraction Kit; Zymo Research Corp., Irvine, CA). Oligonucleotide primers were used to amplify a 320-base pair (bp) fragment from exons 3 and 4 and a 342-bp fragment from exon 5 of the G6PD gene. After enzymatic digestion with NlaIII for exons 3 and 4 and FokI for exon 5, the presence of the A-allele was determined by two single nucleotide polymorphisms; one at position 202 from G to A that resulted in two fragments (113 and 207 bp) and one at position 376 from A to G that resulted in two fragments (173 and 169 bp). Products were separated by gel electrophoresis for 1 hour at 140 volts using a 2.0% agarose gel stained with SYBR safe (Life Technologies, Carlsbad, CA) and imaged using a ultraviolet (UV) transilluminator (Gel Doc XR+ Imager; BioRad Laboratories, Hercules, CA).

Statistical analyses of G6PD assay performance.

The activity measurement using the spectrophotometric assay at 25°C was standardized to activity at 30°C using a temperature correction factor supplied by the manufacturer; the biosensor does this automatically. The average G6PD measurements from the biosensor (U/dL) were divided by the hemoglobin concentration (g/dL) to obtain G6PD activity in U/g Hgb. The resulting G6PD activity was visualized using histograms with thresholds for the classification of severe, moderate, mild, normal, and increased G6PD enzyme activity levels and reported as median values with interquartile ranges (IQRs) in tabular form. Bland–Altman plots were constructed by plotting the average G6PD activity measured using both assays by the difference between the two assays on the x and y axes, respectively. This allowed for the accommodation of both measurements within subjects to show the differences between the two G6PD measurements with 95% limits of agreement (±2 standard deviation [SD]).

Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were calculated with the G6PD activity classification obtained using the spectrophotometric assay as the reference. The ability of the biosensor to correctly classify G6PD activity was evaluated for severe (less than 10% residual activity), moderate-to-severe (less than 30% residual activity), and mild-to-severe G6PD deficiency (less than 60% residual activity). The degree of homogeneity between the G6PD classes determined using the biosensor compared with the reference values from the spectrophotometric assay was evaluated using an intraclass correlation coefficient (ICC). The ICCs were interpreted as follows: < 0.01, poor; 0.01–0.20, slight agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, substantial agreement; and 0.81–1.00, almost perfect agreement.16 Using the genotyping results, traditional performance measures and receiver operator characteristic (ROC) curves were created to compare the ability of both assays to correctly identify hemizygous A-males and homozygous A-females, who are at the greatest risk for AHA following treatment with primaquine. All statistical analyses were conducted in STATA (StataCorp, College Station, TX).

RESULTS

Comparison of quantitative G6PD activity measurements.

The mean G6PD activities of the sample population measured using both assays were similar, with an average value of 8.75 U/g Hgb for the spectrophotometric assay and 8.13 U/g Hgb for the biosensor. The population distributions are presented in Figure 2 with more separation of the moderate to severely deficient population members using the spectrophotometric assay and a smaller range in G6PD activity measured by the biosensor. The number of participants and range for each G6PD activity level as measured by the spectrophotometric assay are presented in Table 2 with the median, IQRs, and difference in activity measured by both assays. The mean difference between the two measurements of G6PD activity from the biosensor and the spectrophotometric assays was 0.78 (95% confidence interval: 0.36, 1.21). Though the distribution of the difference in G6PD activity measured using both assays was approximately normal, the biosensor gave higher readings than the spectrophotometric assay when the activity was below 5 U/g Hgb and lower readings with activity above 10 U/g Hgb. This resulted in a weak, positive trend that can be observed in the Bland–Altman plot (Figure 3) and could be indicative of a systematic measurement error between the two assays.

Figure 2.
Figure 2.

The distributions of the glucose-6-phosphate dehydrogenase (G6PD) activity (U/g Hgb) from the sample population measured using the spectrophotometric assay and the biosensor are presented with the classifications for G6PD deficiency (dotted lines from left to right): severe (< 10%), moderate (10–30%), mild (30–60%), normal (60–150%), and increased (> 150%) activity. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene 97, 4; 10.4269/ajtmh.16-0714

Table 2

Biosensor classification of G6PD activity compared with reference categories

G6PD activityObservedRangeSpectrophotometryBiosensorDifference
classification(n)(U/g Hgb)Median (IQR)Median (IQR)Average (SE)
Severe4< 0.870.72 (0.62–0.79)3.96 (3.45–4.52)−3.29 (0.36)
Moderate300.87–2.621.63 (1.31–2.17)4.15 (3.58–5.66)−3.14 (0.31)
Mild332.62–5.233.80 (3.37–4.63)6.75 (5.00–8.62)−3.29 (0.45)
Normal2225.23–13.128.81 (7.23–10.69)8.37 (7.26–9.72)0.50 (0.18)
Increased54> 13.1216.08 (14.31–19.24)9.57 (8.29–10.85)6.89 (0.36)
Total3430–19.58.63 (5.71–11.65)8.22 (6.59–9.70)0.78 (3.98)

G6PD = glucose-6-phosphate dehydrogenase; IQR = interquartile range; SE = standard error. Sample size (n). The number of schoolchildren observed (n) with severe (< 10%), moderate (10–30%), mild (30–60%), normal (60–150%), or increased (> 150%) G6PD activity levels are presented with the range (in U/g Hgb) in activity, the median activity using the gold standard spectrophotometric assay and IQR, the median activity using the biosensor and the IQR, along with the difference between the assays in U/g Hgb and the SE for the difference.

Figure 3.
Figure 3.

A Bland–Altman plot is used to compare the difference between the two glucose-6-phosphate dehydrogenase activity (U/g Hgb) values obtained from the spectrophotometric assay and the biosensor with the average of the two values and their distribution with respect to the mean difference and two standard deviations from the mean difference (dotted lines). This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene 97, 4; 10.4269/ajtmh.16-0714

Performance measures of the biosensor compared with spectrophotometry.

The prevalence of G6PD deficiency in the sample population using both assays and the performance measures of the biosensor compared with the spectrophotometric assay are presented in Table 3. Of the four population members with severe G6PD deficiency, none were classified as severe by the biosensor (0% sensitivity), only two of 34 were correctly classified as moderate or severe (5.9% sensitivity), and only 36 of 67 were correctly classified as having some form of deficiency (53.7% sensitivity). The specificity of the biosensor was 100% for severe deficiency, 99.7% for moderate or severe deficiency, and 94.5% determining any form of deficiency.

Table 3

Performance measures of the G6PD biosensor

Prevalence of glucose-6-phosphate dehydrogenase deficiency
G6PD activitySpectrophotometricBiosensor
classification[n (%)][n (%)]
Severe4 (1.17)0 (0)
Moderate30 (8.75)3 (0.87)
Mild33 (9.62)48 (13.99)
Normal222 (64.72)285 (83.09)
Increased54 (15.74)7 (2.04)
Comparison of biosensor compared with spectrophotometric assay
Performance measures (%)< 10%< 30%< 60%
Sensitivity0.05.953.7
Specificity100.099.794.6
Accuracy98.890.486.6
Positive predictive value66.770.6
Negative predictive value98.890.689.4

G6PD = glucose-6-phosphate dehydrogenase. The number of schoolchildren observed (n) with different levels of G6PD activity and their prevalence in the study population (%) are presented for the spectrophotometric assay and the biosensor along with the performance of the biosensor to classify participants with G6PD activity levels with < 10% residual activity, 10–30% residual activity, and 30–60% residual activity compared the average activity in nondeficient study population members.

Comparison of G6PD activity level classifications between assays.

The G6PD activity level determined using the spectrophotometric assay as a reference and corresponding classifications determined using the biosensor are presented in Table 4. Though 208 of the 222 normal population members are correctly classified by the biosensor, all four severely deficient participants are classified as having mild deficiency, 19 of 30 moderately deficient participants are classified as having mild deficiency, 20 of 33 mildly deficient participants are classified as normal, and 48 of 54 participants with increased activity were classified as normal. For males, 14.8% of participants (25/169) had moderate-to-severe deficiency; however, only three were correctly identified as moderate to severe, with 17 identified as mild, and six as normal by the biosensor. For females, 5.2% of participants (9/174) had moderate-to-severe deficiency; however, none were correctly identified as moderate to severe, with six identified as mild and three as normal by the biosensor. The overall agreement between the two assays to classify a participant into the correct G6PD activity levels was poor for severe (ICC = 0.01), slight for moderate (ICC = 0.11), slight for mild (ICC = 0.18), fair for normal (ICC = 0.37), and slight for increased G6PD activity (ICC = 0.11). The systematic overestimation of the G6PD activity in deficient population members using the biosensor was also apparent from the histograms of the G6PD activity stratified by gender (Figure 4), where almost all moderately deficient males were misclassified as having higher enzyme activities compared with the spectrophotometric method.

Table 4

Classification of G6PD activity by participant gender

Reference values*nPercentage of normal G6PD activity (%)
< 1010–3030–6060–150> 150
Total34303482857
Severe400400
Moderate30021990
Mild330011202
Normal22201122081
Increased54002484
Male16902261383
Severe200200
Moderate23021560
Mild14003110
Normal105005990
Increased26001223
Female17401221474
Severe200200
Moderate700430
Mild1900892
Normal1180171061
Increased28001261

G6PD = glucose-6-phosphate dehydrogenase. The number of schoolchildren observed (n) with different levels of G6PD activity is presented by gender to show the results obtained from using the biosensor to measure G6PD activity when compared with the gold standard spectrophotometric method to classify participants by G6PD activity level.

Reference values defined as percentage of normal G6PD activity by spectrophotometric assay: severe (< 10); moderate (10–30); mild (30–60); normal (60–150); increased (> 150).

Figure 4.
Figure 4.

The distributions of the glucose-6-phosphate dehydrogenase (G6PD) activity (U/g Hgb) from the sample population measured using the biosensor and the spectrophotometric assay are presented separately for males and females with the threshold values (dotted lines) at which drug-induced hemolysis can occur (30% residual activity, 2.624 U/g Hgb) and the threshold for G6PD deficiency (60% residual activity, 5.249 U/g Hgb). This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene 97, 4; 10.4269/ajtmh.16-0714

Comparison of quantitative G6PD measures by genotype.

In this study population, 18.95% had the presence of the A-allele (65/343), 1.17% were homozygous females (4/343), 6.71% were hemizygous males (23/343), 11.08% were heterozygous females (38/343), and 81.05% did not carry the A-allele (278/343). The average G6PD activity by genotype (±SD and range in U/g Hgb) was as follows: 1.56 ± 0.67 (0.53–3.14) for hemizygous A-males and homozygous A-females, 4.59 ± 1.51 (1.56–6.9) for heterozygous females, and 10.22 ± 4.02 (1.57–20) for population members without the A-allele. A comparison of the G6PD activity values obtained from both tests and the genotypic results are presented (Figure 5). Overall, 7.87% (27/343) of the population members had only A-alleles (hemizygous A-males and homozygous A-females), and were considered at the highest risk for drug-induced hemolysis. The distributions of the G6PD activity by genotype and the ability to correctly identify high-risk population members using either assay were explored using ROCs (Figure 6). Using the genotypic results as a reference, both tests had great performance at discriminating between high-risk population members, and those that would likely tolerate primaquine therapy. The area under the curve (AUC) for the spectrophotometric assay and biosensor were 0.992 and 0.976, respectively; with the spectrophotometric assay having slightly, but statistically significantly greater AUC (P = 0.032) when compared with the biosensor. The activity thresholds at which all high-risk individuals were correctly classified as A-variants (no false positives) were 3.14 and 5.67 U/g Hgb for the spectrophotometric assay and biosensor, respectively.

Figure 5.
Figure 5.

A scatterplot of the glucose-6-phosphate dehydrogenase activity (U/g Hgb) measured using the spectrophotometric assay and the biosensor is presented with respect to the number of copies of the non-A-allele: hemizygous males or homozygous females with only A-alleles (red), heterozygous females with one non-A-allele (yellow), and homozygous females or hemizygous males with only non-A-alleles (gray). This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene 97, 4; 10.4269/ajtmh.16-0714

Figure 6.
Figure 6.

The population distributions of glucose-6-phosphate dehydrogenase (G6PD) activity measured by the spectrophotometric method and the biosensor are presented with respect to G6PD A-hemizygous males and homozygous females (red bars) and heterozygous females/normal population members (gray bars). The receiver operator characteristics and area under the curve (AUC) are presented for the spectrophotometric assay (purple, left) and the biosensor (green, right) to illustrate the sensitivity and specificity of the assays to correctly classify participants with only A-alleles (homozygous A-females and hemizygous A-males) who are at the highest risk for drug induced hemolysis. The gray dotted line shows the characteristics of a test that is not more informative than random chance (AUC = 0.5). This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene 97, 4; 10.4269/ajtmh.16-0714

DISCUSSION

Ideally, the biosensor would have given the same values as the spectrophotometric assay; however, it consistently overestimated the activity of population members with moderate or severe G6PD deficiency. This led to a sensitivity of only 5.9% to diagnose moderate-to-severe G6PD deficiency and only poor, slight, or fair agreement between the assays to correctly determine the ordinal G6PD activity classification. Although the biosensor yielded higher measurements of G6PD activity in ranges clinically relevant on the reference assay (< 30% residual activity), certain observations indicated that this new device had the ability to separate normal and deficient population members. The bimodal distribution in males obtained from the biosensor (Figure 4) was similar to that of the spectrophotometer, albeit with a right shift in activity units. Likewise, correct identification of all homozygous A-females and hemizygous A-males was achieved using a higher threshold (5.67 versus 3.24 U/g Hgb). Using this threshold to guide treatment decisions, none of the 34 at-risk individuals with moderate or severe G6PD deficiency would have been treated with primaquine and treatment would have been withheld from only 5.8% (18/309) of population members without G6PD deficiency.

Though Haiti has been reported to have little evidence of P. vivax infections and low-level transmission of P. falciparum, the relatively high prevalence of glucose-6-dehydrogenase deficiency, rugged conditions, and tropical climates made Haiti an excellent location for a field trial of this new point-of-care diagnostic.1720 In a previous evaluation of a qualitative G6PD rapid diagnostic test (BinaxNOW G6PD Test) differences in the humidity and temperature in the field compared with laboratory conditions led to low reliability of the assay.9 Although the average field conditions during sample collection (60–80% and 85–95°F) were certainly hotter and more humid than the controlled laboratory setting (30% and 75–77°F), humidity and temperature were not measured with each use of the biosensor, thus we are unable to speculate on any an systematic effects on measurement of G6PD activity using the biosensor. Overall, the field readings were not significantly different (P = 0.3, student t test) than those from the laboratory (mean biosensor activity 92.5 versus 90.8 U/dL) with moderate agreement (R2 values between 76% and 80%) when stratified by sample location (Figure 7).

Figure 7.
Figure 7.

Scatterplots of the glucose-6-phosphate dehydrogenase activity readings using the biosensor (U/dL) in the laboratory compared with the field are presented by school and overall with the R2 values from a linear regression of the laboratory and field values. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene 97, 4; 10.4269/ajtmh.16-0714

Unlike the spectrophotometric assay, which requires a trained technician, UV spectrophotometer, calibrated pipettes, and other wet laboratory equipment, the biosensor was intuitive and easily operated by community health workers with minimal training. Compared with the qualitative G6PD assay from the same manufacturer (CareStart G6PD RDT), the biosensor provides a quantitative measure of G6PD activity, but demonstrated a markedly lower sensitivity to determine those most at risk for AHA.21 If adequate improvements could be made to the performance of the biosensor, the quantitative determination would allow for the standardization of treatment guidelines within countries and for researchers to study the relationships between patient G6PD activity, primaquine dosage, and adverse clinical events to help optimize ongoing malaria elimination programs.22 Additionally, this biosensor could be used to screen for potential susceptibility to drug-induced hemolysis from other pharmacological agents that are contraindicated in G6PD-deficient patients including certain anthelmintics, antimalarials, antibiotics, and antimycobacterials; many of which are routinely used for treating tropical infectious diseases in developing countries where G6PD deficiency is common.23

LIMITATIONS

A limitation of the current study was that a convenience sample of primary school students from a single department in Haiti was used to evaluate the performance of the biosensor. As such, the results might not be representative of other departments of Haiti, or the country as a whole. To support the preliminary results provided in the current study, future work should include more comprehensive field trials in Haiti and other countries with moderate-to-high prevalence of G6PD deficiency and future evaluations of any improvements made to the current device. Although both quantitative assays were compared with the presence of the African A-allele, the current study design would have only allowed for the conclusion that the performance of the biosensor is at least not inferior to, but not better than the current gold standard for the phenotypic determination of G6PD activity.

CONCLUSIONS

By comparing the portable biosensor under field conditions to the current gold standard spectrophotometric assay in a laboratory setting, this study was able to provide the first evaluation of this new device for the point-of-care diagnosis of G6PD deficiency. Though this novel biosensor showed promise, we conclude that its performance is currently unacceptable relative to the gold standard and must be improved prior to its use for the diagnosis of G6PD deficiency. If this were to be accomplished, the CareStart G6PD Biosensor Analyzer could prove to be an invaluable tool in the elimination of malaria and treatment of tropical diseases in which G6PD deficiency is a contraindication to pharmacotherapy.

Acknowledgments:

We extend a special thanks and sincere gratitude to the children and legal guardians who participated in this study, the Christianville Foundation Schools, the local medical staff, and Kaylee Sills; without whom this study would not have been possible. We would also like to acknowledge Access Bio, Inc., for donation of the newly developed CareStart G6PD Biosensor Analyzer for independent evaluation by our research team in this field trial.

REFERENCES

  • 1.

    World Health Organization, 2015. World Malaria Report. Geneva, Switzerland: WHO. Available at: http://www.who.int/malaria/publications/world-malaria-report-2015/report/en/. Accessed December 8, 2016.

  • 2.

    Mikolajczak SA et al.., 2015. Plasmodium vivax liver stage development and hypnozoite persistence in human liver-chimeric mice. Cell Host Microbe 17: 526553.

    • Search Google Scholar
    • Export Citation
  • 3.

    Galappaththy GNL, Tharyan P, Kirubakaran R, 2013. Primaquine for preventing relapse in people with Plasmodium vivax malaria treated with chloroquine. Cochrane Database Syst Rev 10: CD004389.

    • Search Google Scholar
    • Export Citation
  • 4.

    Beutler E, 1996. G6PD: population genetics and clinical manifestations. Blood Rev 10: 4552.

  • 5.

    Mason PJ, Bautista JM, Gilsanz F, 2007. G6PD deficiency: the genotype-phenotype association. Blood Rev 21: 267283.

  • 6.

    Luzzatto L, Metha A, Vulliamy T, 2001. Glucose-6-phosphate dehydrogenase deficiency. Scriver CR, Beaudet AL, Sly WS, eds. The Metabolic and Molecular Bases of Inherited Disease, 9th edition. Columbus, OH: McGraw-Hill, 4517–4553.

  • 7.

    Beutler E, Duparc S, 2007. Glucose-6-phosphate dehydrogenase deficiency and antimalarial drug development. Am J Trop Med Hyg 77: 779789.

  • 8.

    World Health Organization, 2015. Point-of-care G6PD testing to support safe use of primaquine for the treatment of vivax malaria. Geneva, Switzerland: WHO. Available at: http://www.who.int/malaria/mpac/mpac-march2015-erg-g6pd.pdf. Accessed November 10, 2016.

  • 9.

    Osorio L et al.., 2015. Performance of binaxNOW G6PD deficiency point-of-care diagnostic in P. vivax-infected subjects. Am J Trop Med Hyg 92: 2227.

    • Search Google Scholar
    • Export Citation
  • 10.

    Kim S et al.., 2011. Performance of the CareStart G6PD deficiency screening test, a point-of-care diagnostic for primaquine therapy screening. PLoS One 6: e28357.

    • Search Google Scholar
    • Export Citation
  • 11.

    Access Bio, Inc. Manufacturer Information and Instruction for the Operation of the CareStartTM G6PD Biosensor Analyzer. Available at: http://www.accessbio.net/eng/products/products03.asp. Accessed January 1, 2017.

  • 12.

    World Health Organization, 1989. Glucose-6-phosphate dehydrogenase deficiency. WHO Working Group. Bull World Health Organ 67: 601611.

    • Search Google Scholar
    • Export Citation
  • 13.

    Howes RE et al.., 2013. Spatial distribution of G6PD deficiency variants across malaria-endemic regions. Malar J 12: 418426.

  • 14.

    Carter TE, Maloy H, von Fricken ME, St Victor Y, Romain JR, Okech BA, 2014. Glucose-6-phosphate dehydrogenase deficiency A-variant in febrile patients in Haiti. Am J Trop Med Hyg 91: 412414.

    • Search Google Scholar
    • Export Citation
  • 15.

    Tishkoff SA et al.., 2001. Haplotype diversity and linkage disequilibrium at human G6PD: recent origin of alleles that confer malarial resistance. Science 293: 455462.

    • Search Google Scholar
    • Export Citation
  • 16.

    Landis JR, Koch GG, 1977. The measurement of observer agreement for categorical data. Biometrics 33: 159174.

  • 17.

    Weppelmann TA, Carter TE, Chen Z, von Fricken ME, Victor YS, Existe A, Okech BA, 2013. High frequency of the erythroid silent Duffy antigen genotype and lack of Plasmodium vivax infections in Haiti. Malar J 12: 30.

    • Search Google Scholar
    • Export Citation
  • 18.

    Weppelmann TA, von Fricken ME, Lam B, Telisma T, Existe A, Lemoine JF, Larkin J III, Okech BA, 2016. Sparse serological evidence of Plasmodium vivax transmission in the Ouest and Sud-Est departments of Haiti. Acta Trop 162: 2734.

    • Search Google Scholar
    • Export Citation
  • 19.

    von Fricken ME, Weppelmann TA, Lam B, Eaton WT, Schick L, Masse R, Beau De Rochars MV, Existe A, Larkin J III, Okech BA, 2014. Age-specific malaria seroprevalence rates: a cross-sectional analysis of malaria transmission in the Ouest and Sud-Est departments of Haiti. Malar J 13: 361.

    • Search Google Scholar
    • Export Citation
  • 20.

    von Fricken ME, Weppelmann TA, Eaton WT, Alam MT, Carter TE, Schick L, Masse R, Romain JR, Okech BA, 2014. Prevalence of glucose-6-phosphate dehydrogenase (G6PD) deficiency in the Ouest and Sud-Est departments of Haiti. Acta Trop 135: 6266.

    • Search Google Scholar
    • Export Citation
  • 21.

    von Fricken ME, Weppelmann TA, Eaton WT, Masse R, Beau de Rochars MV, Okech BA, 2014. Performance of the CareStart glucose-6-phosphate dehydrogenase (G6PD) rapid diagnostic test in Gressier, Haiti. Am J Trop Med Hyg 91: 7780.

    • Search Google Scholar
    • Export Citation
  • 22.

    Domingo GJ et al.., 2013. G6PD testing in support of treatment and elimination of malaria: recommendations for evaluation of G6PD tests. Malar J 12: 391.

    • Search Google Scholar
    • Export Citation
  • 23.

    Luzzatto L, Seneca E, 2013. G6PD deficiency: a classic example of pharmacogenetics with on-going clinical implications. Brit J Haem 164: 469480.

    • Search Google Scholar
    • Export Citation

Author Notes

Address correspondence to Bernard A. Okech, Emerging Pathogens Institute, University of Florida, 2055 Mowry Road, Gainesville, FL 32611. E-mail: bokech@ufl.edu

These authors contributed equally to this work.

Authors’ addresses: Thomas A. Weppelmann, Herbert Wertheim College of Medicine, Florida International University, Miami, FL, E-mail: twepp002@fiu.edu. Michael E. von Fricken and Elisa Aguenza, George Mason University, Global and Community Health, Fairfax, VA, E-mails: mvonfric@gmu.edu and echon@masonlive.gmu.edu. Tara D. Wilfong and Bernard A. Okech, Department of Environmental and Global Health, University of Florida, Gainesville, FL, and Emerging Pathogens Institute, University of Florida, Gainesville, FL, E-mails: twodoc@ufl.edu and bokech@ufl.edu. Taina T. Philippe, Christianville Clinic, Christianville Foundation Inc., Gressier, Haiti, E-mail: tainatelisma@yahoo.com.

Save