An International Survey of Glucose-6-Phosphate Dehydrogenase Laboratory Reporting Practices: Implications for Tafenoquine Eligibility Assessment

In October 2019, supplemental questions regarding percent of normal G6PD reporting practices were distributed to 327 laboratories participating in the CAP G6PDS-B 2019 Survey. In October 2020, additional supplemental questions regarding G6PD RI reporting practices were distributed to 324 laboratories participating in the G6PDS-B 2020 Survey. Supplemental questions are included in the Supplemental Digital Content (at https://meridian.allenpress.com/aplm in the October 2022 table of contents) of the present manuscript. “Do not know” responses to supplemental questions in G6PDS-B 2019 are shown in Table 1 but were excluded before statistical analysis. This exclusion was performed as “do not know” responses do not capture practice characteristics of the laboratory, and therefore cannot be used to draw comparisons in practice across laboratories.

Participant responses to assay manufacturer, unit of measure, temperature, result, and interpretation for proficiency testing (PT) challenges designed to represent “normal” (2020 B mailing; specimen G6PD-03), “intermediate” (2020 A mailing; specimen G6PD-01), and “deficient” (2020 B mailing; specimen G6PD-04) G6PD status were also used for analysis of quantitative and qualitative results. Descriptive statistics for G6PD assessed using quantitative methods were summarized for assays with a peer group size of at least 10. Outlier screening was performed using the criterion of 1.5 times the interquartile range. X²  and Fischer exact tests were used to examine the association between participant interpretation and type of assay for each specimen. The analytic specimen was restricted to participants that provided either a quantitative or qualitative interpretation to ensure that the groups were mutually exclusive.

Statistical significance was set at the 5% level. For analysis of the strength and directionality of the relationship between the components of the RI (low end, high end, mean of the low and high end) and the “normal” G6PD result represented by the G6PD-03 specimen in the 2020 B mailing, and owing to the skewed distribution of the analyte, Spearman rank correlation was performed with a Bonferroni correction to adjust for multiplicity in testing. All statistical analyses and graphics were generated using SAS 9.4 (Cary, North Carolina). Data are presented as mean ± SD unless otherwise indicated.

Two hundred fifty-seven laboratories (US, 45.5%, n = 117; Canada, 12.5%, n = 32; international, 42.0%, n = 108) responded to questions regarding percent of normal reporting practices in the 2019 B mailing (Table 1). A smaller percentage of clinical laboratories in the US and Canada (2.0% [3 of 149]) report having received questions from clinicians about average normal G6PD activity than was reported by international laboratories (8.4% [9 of 107]) (Question 1; P = .04). In addition, a smaller percentage of clinical laboratories in the US and Canada (6.7% [10 of 149]) have determined what their average normal G6PD activity is for their assay and patient population than was reported by international laboratories (19.4% [21 of 108]) (Question 2; P = .008). Last, a smaller percentage of clinical laboratories in the US and Canada (2.7% [4 of 149]) currently report G6PD results as a percent of normal activity than was reported by international laboratories (8.3% [9 of 108]) (Question 3; P = .05). As noted above, only a small minority of laboratories overall are involved in responding to clinician questions about average normal G6PD activity, evaluating laboratory-specific average normal G6PD results, or reporting G6PD results in percent of normal format.

One hundred nineteen laboratories (US, 33.6%, n = 40; Canada, 11.8%, n = 14; international, 54.6%, n = 65) responded to questions regarding percent of normal and/or RI reporting practices in the 2020 B mailing. The G6PDS Survey questions were designed to assess which RIs were used by participant laboratories for their assays, whether they currently perform percent of normal reporting, and if so, which RIs or interpretive information is provided with percent of normal results. Figure 1, A through G, shows participant G6PD RIs by assay and configuration temperature, displaying results for grouping with more than 2 assay/temperature pairings, including Pointe Scientific (37°C, n = 34; Figure 1, A), United Diagnostics Industry (UDI) (25°C, n = 4; Figure 1, B), UDI (37°C, n = 7; Figure 1, C), and Randox (37°C, n = 3; Figure 1, E). Participants that provided “other” (37°C; n = 14) as a method response with unit per gram of hemoglobin (Hb) reporting units are shown in Figure 1, D. Only a limited number of RIs were provided from laboratories reporting G6PD (U/10¹² red blood cell [RBC]) units of measurement, with assays configured at 37°C (n = 6; Figure 1, F) and 25°C (n = 2; Figure 1, G). RIs for laboratories reporting RBC units (U/10¹² RBC) are therefore not displayed by specific assay.

Of 107 participants that responded to the follow-up G6PDS-B 2020 Survey question about percent of normal reporting (Question 3), only 1.9% (n = 2) reported that they provide percent of normal G6PD reporting, whereas 94.4% (n = 101) responded “no'” and 3.7% (n = 4) responded “do not know.” Of the 2 participants that responded “yes,” 1 reported a percent of normal RI of 80% or more, whereas the other responded that they provide interpretive comments only.

To further assess G6PD results across laboratories using the same reagents and temperature conditions, PT specimens from the 2020 G6PDS Survey A and B mailings designed to represent “normal,” “intermediate,” and “deficient” G6PD status were analyzed by method. Results for assays that had greater than or equal to 10 participants are shown in Figure 2, A through C, and included assays from Pointe Scientific and UDI, both configured at 37°C. Results for the 3 PT specimen challenges are presented in Figure 2, A through C, and Table 2 and were as follows: normal specimen (Figure 2, A, G6PD-03)—Pointe Scientific, n = 48, 12.74 ± 2.15 U/g Hb (% coefficient of variation [CV] 16.9); UDI, n = 11, 10.17 ± 0.88 U/g Hb (%CV 8.7); intermediate specimen (Figure 2, B, G6PD-01)—Pointe Scientific, n = 48, 5.26 ± 1.17 U/g Hb (%CV 22.2); UDI, n = 11, 4.18 ± 0.34 U/g Hb (%CV 8.1); and deficient specimen (Figure 2, C, G6PD-04)—Pointe Scientific, n = 46, 1.59 ± 0.34 U/g Hb (%CV 21.7); UDI, n = 11, 1.45 ± 0.12 U/g Hb (%CV 8.3). These results demonstrate that while the absolute variation in G6PD results across laboratories (ie, SD) decreases with lower G6PD concentrations for both assays, these differences, when normalized to the mean (ie, %CV), are relatively consistent within a specific-vendor reagent grouping.

To assess whether there was a correlation between a laboratory's result for the “normal” specimen PT challenge (2020 B mailing, G6PD-03) and the range of results specified in the laboratory-specific RI provided in responses to the supplemental questions, participant “normal” specimen G6PD-03 results for the largest assay group—Pointe Scientific (37°C)—were plotted against the mean of the respective laboratory-specific low and high RI limits and were also analyzed separately across all methods. This calculated mean was chosen as an estimate of the middle peak of a presumed parametric RI. Figure 3 presents the results of this analysis for Pointe Scientific (37°C). Spearman rank correlation coefficients (r) for the correlation between the RI and the normal specimen G6PD-03 results (expressed as r for low RI, high RI, mean of the low and high RI) were as follows: Pointe Scientific at 37°C, n = 33 (r = .31, .52°, .51*); all methods, n = 66 (r = .31, .52*, .51*); all methods, n = 66 (r = .53*, .36*, .46*); *P value of < .02.

Finally, the ability of quantitative and qualitative assays to correctly categorize PT challenge specimens as “normal” (2020 B mailing, G6PD-03), “intermediate” (2020 A mailing G6PD-01), and “deficient” (2020 B mailing, G6PD-04) was then assessed. For this analysis, participants were excluded if a laboratory provided both quantitative and qualitative results (2020 A mailing, specimen G6PD-01, n = 27 exclusions; 2020 B mailing, specimens G6PD-03 and G6PD-04, n = 35 exclusions). As shown in Table 3, both quantitative and qualitative assays were able to correctly categorize the “normal” and “deficient” PT specimens for more than 90% of participants. The specimen designed as “intermediate” (G6PD-01) had a low-to-intermediate quantitative result (5.26 ± 1.17 U/g Hb, Point Scientific, 37°C; 4.18 ± 0.34 U/g Hb, UDI, 37°C). A total of 70.5% (62 of 88) of participants using quantitative assays categorized it as “deficient,” whereas 17.0% (15 of 88) categorized it as “intermediate” and 12.5% (11 of 88) categorized it as “normal.” Participants using qualitative assays had more difficulty categorizing it as “intermediate” or “deficient.” Of participants using qualitative assays, 41 of 178 (23.0%) categorized the G6PD-01 specimen as “deficient,” whereas 52 of 178 (29.2%) categorized it as “intermediate” and 85 of 178 (47.8%) categorized it as “normal.” An overall P value of < .001 was obtained from the X²  test of association for this comparison (ie, participant interpretations for G6PD-01 derived from quantitative versus qualitative methods).

The present investigation provides an overview of current G6PD reporting practices among participants in the CAP G6PDS Survey. Despite longstanding phenotypic G6PD WHO classifications that rely upon G6PD activity expressed as a percent of normal, and the description of and reference to percent of normal thresholds in WHO guidelines for antimalarial use,⁹  few laboratories presently report G6PD results in a percent of normal format. Variation in quantitative PT results for “normal,” “intermediate,” and “deficient” specimens across participants, as well as clinical use of a variety of RIs, both support the notion that variability in patient G6PD result distributions exists across clinical laboratories. Addition of percent of normal G6PD reporting could help clinicians better interpret the G6PD status of their patients, regardless of where testing had been performed.

Determination of a laboratory's average normal G6PD activity—specific to its assay, configuration, temperature, and patient population—is required for accurate percent of normal reporting. A 2012 G6PD workshop held in Bangkok, Thailand outlined critical issues associated with G6PD testing, and its subsequent 2013 published report includes a detailed description of how one can determine male medians (as well as adjusted male medians, to exclude patients with G6PD activity <10% of the male median) to define 100% normal G6PD activity for a laboratory.⁸  Numerous other studies have similarly defined median or mean normal G6PD activity to assess, derive, and/or discuss percent of normal activity cutoffs.^16–24  Calculation of adjusted male medians can be conducted using retrospective analysis of clinical data.¹⁰ 

Quantitative G6PD results are typically normalized either to Hb or RBC count, to account for the variation in number of erythrocytes present in a patient's whole blood specimen submitted for testing. As demonstrated in the G6PDS Survey Participant Summary, the majority of laboratories report quantitative G6PD results in unit per gram Hb units of measure and at 37°C reaction temperature.²⁵  A smaller minority of participants report RBC units of measure (U/10¹² RBC), and/or perform testing at 25°C or other reaction conditions.²⁵  Standardization toward unit per gram Hb (37°C), where technically feasible, in addition to incorporation of percent of normal reporting, would provide an opportunity for more harmonization within clinical and laboratory communities involved in G6PD testing and interpretation. Of note, variability in G6PD results reported by participants in the G6PDS Surveys may also be due to the method(s) used for Hb or RBC count determination. The G6PDS Survey does not collect participant technical information on these methods, thus further investigation into this possibility cannot be conducted with the current dataset.

Clinical laboratories are required to validate and/or verify RIs appropriate for their assay and patient populations.^26,27  Review of package insert RIs, as well as those found within textbooks or published RI studies, is often a first step in this process. Examples of methods for the establishment and verification of RIs have previously been reported.^28–31 Figure 1, A through G, illustrates the diversity of RIs currently used across participant clinical laboratories for G6PD reporting. As an example, the Pointe Scientific G6PD package insert describes “expected values” of 12.1 ± 2.09 U/g Hb (ie, 10.01–14.19 U/g Hb) when performed at 37°C (Glucose-6-Phosphate Dehydrogenase Reagent Set; Pointe Scientific, Canton, Missouri). The insert lists the third edition of the Tietz Textbook of Clinical Chemistry as reference for these values.³²  More recent editions of the textbook include a wider RI of 7.9 to 16.3 U/g Hb.³³  As shown in Figure 1, A, 4 participants reported using the package insert range of 10.0 to 14.2 U/g Hb, whereas 1 participant reported using the range of 7.9 to 16.3 U/g Hb. This suggests that other assay configuration–specific, literature-derived, and/or population-specific RIs are alternatively being used by most clinical laboratories.

The present report also provides evidence demonstrating a moderate correlation between a participant laboratory's quantitative result for a “normal” PT specimen and the average of its laboratory-specific upper and lower RI limits. Given the result distribution observed across participant laboratories (Figure 3), there is likely an opportunity for optimization and/or re-evaluation of RIs used in some clinical laboratories. This is particularly important for laboratories that use common RIs (eg, Figures 1, A, and 3). Common RIs may be observed when laboratories verify vendor-supplied RIs, when a laboratory evaluates publicly available RIs (eg, from the published literature), or when laboratories within a corporate or hospital network standardize assays and/or reporting conventions across multiple sites. Re-verification and/or validation of an RI could be conducted at the same time as the median normal male result is derived, thus enabling percent of normal reporting.

It should also be noted that the data presentation in Figure 1 can also help to visualize RIs that may be clinically inappropriate or erroneously reported. For example, the RI for 1 participant using the UDI assay (37°C) was reported as 0 to 26 U/g Hb (Figure 1, C). Such a range would not be expected to provide flags useful to facilitate clinician identification of G6PD deficiency states or could alternatively be representative of an analytical measuring range erroneously reported in the questionnaire as an RI.

Interpretation of G6PD results can also sometimes be challenging, as results from individual patients with severe-deficiency and moderate-deficiency WHO G6PD deficiency classifications can overlap with the low end of traditional RIs delineated by the central 95th percentile of a “normal” population.^4,5,10  This may be due to a variety of factors, including the large number of genetic variants associated with G6PD deficiency,³  the complexity of genotype-phenotype correlations,⁵  and X-chromosome lyonization in females.⁶  It is also important to emphasize that significant within-subject and between-subject biological variation in erythrocyte G6PD measurement has been described.^34,35 

This report also highlights the challenges in evaluating intermediate and/or borderline G6PD results, particularly when using qualitative assays (Table 3). Intermediate quantitative results can be observed in heterozygous females with G6PD variants given X-chromosome lyonization as noted above.^36,37  This presents a particular challenge when using qualitative assays that may be designed primarily to differentiate between G6PD “normal” and “deficient” (typically <30% normal activity) categorizations; some qualitative assays are not capable of differentiating intermediate results. Future developments in, and availability of, cost-effective G6PD quantitative point-of-care devices, and with temperature stable reagents, will be necessary to address these concerns, particularly in resource-limited settings that currently rely on qualitative options.

Limitations of the present report include the distribution of supplemental questions only to enrollees in the CAP G6PDS Survey. The extent of G6PD testing outside of these participant laboratories, as well as the reporting practices in these settings, could therefore not be evaluated. In addition, completion of supplemental questions in the G6PDS Surveys is considered voluntary. Furthermore, questionnaires (supplemental questions) may have been completed by individuals in a variety of laboratory roles (eg, technologists, supervisors, laboratory managers, directors, and/or pathologists). Differences in familiarity with reporting practices may therefore have contributed to the “do not know” responses provided by some participants. Finally, the response rate to the 2020 B supplemental questions was lower than in the 2019 B Survey. This may have been due to the directions included in the 2020 B supplemental questions emphasizing a focus on quantitative reporting. It could also be due to PT Survey questionnaire fatigue, or alternatively, a de-emphasis on voluntary activities as clinical laboratorians focused on maintaining operations during the COVID-19 pandemic.

In the absence of more widespread use of percent of normal reporting, it is unclear how clinicians will be able to make effective tafenoquine eligibility determinations to limit the risk of intravascular hemolysis. Determination of average normal G6PD activity is a critical step toward enabling percent of normal G6PD activity reporting. Addition of percent of normal reporting could also enable more accurate assessments of hemolytic risk in other settings where G6PD status is frequently considered, including hematology and oncology.

The authors would like to thank Patrick A. Erdman, DO, for input during survey development. Preliminary analysis of qualitative and quantitative assays with intermediate G6PD specimens was provided in an education discussion in the Glucose-6-Phosphate Dehydrogenase G6PDS-B 2020 Participant Summary Report. College of American Pathologists. Northfield, Illinois; 2020.