Use of Capillary Blood Samples Leads to Higher Parasitemia ... · malaria control strategies and,...

The Journal of Infectious Diseases

1296 • JID 2018:218 (15 October) • Mischlinger et al

Use of Capillary Blood Samples Leads to Higher Parasitemia Estimates and Higher Diagnostic Sensitivity of Microscopic and Molecular Diagnostics of Malaria Than Venous Blood SamplesJohannes Mischlinger,1,2,3,4,5,6 Paul Pitzinger,1,2 Luzia Veletzky,1,2,5,6 Mirjam Groger,1,2,5,6 Rella Zoleko-Manego,2,3,4,5,6 Ayola A. Adegnika,2,3,4 Selidji T. Agnandji,2,3,4 Bertrand Lell,2,3,4 Peter G. Kremsner,2,3,4 Egbert Tannich,7,8 Ghyslain Mombo-Ngoma,2,3,4,5,6,9 Benjamin Mordmüller,2,3,4 and Michael Ramharter5,6,8

1Department of Medicine I, Division of Infectious Diseases and Tropical Medicine, Medical University of Vienna, Austria; 2Centre de Recherches Médicales de Lambaréné, Gabon; 3Institut für Tropenmedizin, Universität Tübingen, 4German Center for Infection Research, partner site Tübingen, and 5Department of Tropical Medicine, Bernhard Nocht Institute for Tropical Medicine, 6I Department of Medicine University Medical Center Hamburg-Eppendorf, 7Bernhard Nocht Institute for Tropical Medicine, World Health Organization Collaborating Centre for Arbovirus and Hemorrhagic Fever Reference and Research, and 8German Centre for Infection Research, partner site Hamburg-Luebeck-Borstel, Hamburg, Germany; and 9Université des Sciences de la Santé Gabon, Département de Parasitology, Malaria Clinical and Operational Research Unit, Melen Hospital, Libreville, Gabon

Background. Diagnosis of malaria is usually based on samples of peripheral blood. However, it is unclear whether capillary (CAP) or venous (VEN) blood samples provide better diagnostic performance. Quantitative differences of parasitemia between CAP and VEN blood and diagnostic performance characteristics were investigated.

Methods. Patients were recruited between September 2015 and February 2016 in Gabon. Light microscopy and quantitative polymerase chain reaction (qPCR) measured parasitemia of paired CAP and VEN samples. CAP and VEN performance character-istics using microscopy were evaluated against a qPCR gold standard.

Results. Microscopy revealed a median parasitemia of 495/μL in CAP and 429/μL in VEN samples, manifesting in a 16.6% (P = .04) higher CAP parasitemia compared with VEN parasitemia. Concordantly, in qPCR –0.278 (P = .006) cycles were required for signal detection in CAP samples. CAP sensitivity of microscopy relative to the gold standard was 81.5% vs VEN sensitivity of 73.4%, while specificities were 91%. CAP and VEN sensitivities dropped to 63.3% and 45.9%, respectively, for a subpopulation of low-level parasitemias, whereas specificities were 92%.

Conclusions. CAP sampling leads to higher parasitemias compared to VEN sampling and improves diagnostic sensitivity. These findings may have important implications for routine diagnostics, research, and elimination campaigns of malaria.

Keywords. malaria; sensitivity; specificity; capillary; venous.

Malaria constitutes a major global health problem. According to estimates by the World Health Organization (WHO), 216 mil-lion malaria cases had occurred in 2016, with 445 000 resulting in death [1]. Furthermore, 90% of all malaria cases and deaths worldwide occurred in the WHO African Region [1]. These fig-ures demonstrate the high disease burden that resource-limited countries need to withstand and the consequent need for con-trol strategies in these settings that are not only effective, but also affordable.

To date, early diagnosis is among the most important malaria control strategies and, where available, light micros-copy of thick blood smears is considered the gold standard for diagnosis in clinical routine [1, 2]. Currently, no specific rec-ommendation exists with regard to the source of blood sample that should be used for malaria diagnostics. Most commonly sampled blood stems from finger pricks (ie, capillary [CAP] blood) and venous (VEN) blood draws.

Previously, 2 studies have reported potentially superior performance of CAP blood samples for malaria diagnosis compared to VEN blood samples [3, 4]. These studies, how-ever, were underpowered and used methodologies that do not permit to fully support these conclusions. As described above, cost-efficient optimizations of malaria diagnostics are urgently needed in resource-limited settings. Furthermore, the question whether certain blood sources may potentially influence measurements of parasitemia is also important for the clinical management of malaria as well as for research studies [5–7].

M A J O R A R T I C L E

© The Author(s) 2018. Published by Oxford University Press for the Infectious Diseases Society of America. All rights reserved. For permissions, e-mail: [email protected]: 10.1093/infdis/jiy319

Received 13 April 2018; editorial decision 18 May 2018; accepted 23 May 2018; published online May 25, 2018.

Correspondence: M. Ramharter, MD, DTM&H, Department of Tropical Medicine, Bernhard Nocht Institute for Tropical Medicine & I. Department of Medicine University Medical Center Hamburg-Eppendorf, Bernhard-Nocht Straße 74, Hamburg 20359, Germany ([email protected]).

The Journal of Infectious Diseases® 2018;218:1296–305

15

October

218

Dow

nloaded from https://academ

ic.oup.com/jid/article-abstract/218/8/1296/5003449 by guest on 01 N

ovember 2018

mailto:[email protected]?subject=

Diagnostic Sensitivity of Capillary Blood for Detection of Malaria Parasites • JID 2018:218 (15 October) • 1297

This present study used state-of-the-art methods to inves-tigate a potential difference in parasitemia between CAP and VEN samples and to determine the diagnostic performance characteristics of CAP and VEN samples to detect malaria parasites.

METHODS

Study Design and Study Population

The study population of this prospective diagnostic study was recruited at the Centre de Recherches Médicales de Lambaréné and the Centre de Recherches de la Ngunié, 2 clinical research centers in central Gabon [8, 9]. The study region is hyperen-demic for malaria and has a perennial transmission pattern. All patients presenting to the diagnostic services of the 2 health centers for malaria diagnosis were invited to participate in this study. Recruitment took place between September 2015 and February 2016.

Malaria DiagnosisSamplingCAP blood stemmed exclusively from finger pricks and VEN blood was obtained from puncture of the cubital vein. CAP and VEN sampling was performed in parallel within a 5-min-ute interval. Prior to sample preparation, blood from both sources was first collected in separate containers with ethylene-diaminetetraacetic acid (EDTA) coating to avoid blood clotting (250-μL tube for CAP blood and 3-mL tube for VEN blood). After blood sampling the following steps were taken: For each patient, 2 thick smears of CAP and VEN blood were prepared using standardized quantities of 10 μL. One CAP-VEN pair was employed in analysis; the other pair was kept as backup for use in case of insufficient quality of the first pair. In addition, 2 thin smears of CAP and VEN blood using 5 μL blood were prepared and, standardized quantities of 20 μL CAP and VEN blood were applied on filter papers (Whatman 903 Protein Savers). CAP and VEN sample preparation followed the exact same methodology.

Due to logistic reasons EDTA-coated tubes for CAP sam-pling were not available throughout the entire conduct of the study. In case of nonavailability of CAP sampling tubes, CAP blood was directly aspirated from the finger with a pipette and immediately used for sample preparation. These samples were included in the microscopic evaluation but were excluded from molecular analysis.

Light MicroscopyThick and thin smears were stained with 4% Giemsa for 60 minutes. Thick smears were used for quantification of asexual parasitemia (expressed as asexual parasites/μL blood) and thin smears for malaria species determination.

Parasite quantification methods were used as recommended by WHO [10–12]. Each slide was read by 2 expert microsco-pists and the arithmetic mean of these 2 readings was recorded.

Another reading by a third microscopist was performed if the ratio of parasite densities from the higher to the lower count was >1.5 or if there was a discrepancy in positivity. By employ-ing this strict 2- to 3-reader approach, variability of parasite counting was reduced in accordance with standard practice for clinical trials on antimalarial chemotherapy [13, 14]. A thick smear was classified as negative if at least 100 microscopic high-power fields were judged negative. Microscopists were blinded to the blood source during parasitological analysis as well as to quantitative polymerase chain reaction (qPCR) results.

Quantitative Polymerase Chain ReactionDNA purification was performed using QIAamp DNA mini kits (Qiagen) as recommended by the manufacturer [15]. qPCR was conducted employing a validated assay at the German reference center for tropical pathogens (Bernhard Nocht Institute for Tropical Medicine, Hamburg, Germany) with 10 μL of purified DNA using the RealStar Malaria PCR Kit (Altona Diagnostics) according to the manufacturer’s recommendations [16].

Parasitological Definitions, Analyses, and Statistical ConsiderationsComparison of ParasitemiaDue to the skewed distribution of parasitemia (expressed as parasites/μL), a log-transformation of parasitemia estimates was performed. Difference plots of log2(CAPparasitemia) – log2 (VENparasitemia) were computed. In analogy to Bland–Altman analysis, this was performed over the whole range of measure-ments [17, 18]. However, as the objective was to determine dif-ferences rather than agreements, no limits of agreement were computed. As differences of log-transformed data correspond to ratios, the methodology allowed us to demonstrate the mean excess parasitemia in percentage of either one blood source over

the other ie, CAPVEN

parasitemia

parasitemia

[19, 20].

Patients were considered for analysis of the quantitative com-parison of CAPparasitemia and VENparasitemia if they were positive for malaria in both blood sources. In the case of discordance of positivity (CAP positive and VEN negative or vice versa) patients were still eligible for analysis if qPCR proved positivity for malaria in the respective blood sample that was negative in microscopy. It is a common strategy to assign a value below the limit of detection (LOD) to a nondetect if there is reason to con-clude that the unit-to-detect is actually present [21, 22]. Thus, the power of a study is improved, as log-transformation of 0 or a negative number would entail a missing data pair. This concept was applied in this study by assigning half the parasitemia of the paired positive sample for the microscopically negative sample. This conservative approach ensured that no artificial outliers were produced, thus avoiding measurement bias.

Difference plots of CAP cycles – VEN cycles were computed for qPCR analysis. Patients were eligible for analysis if qPCR was positive in both CAP and VEN blood.

Dow



ovember 2018


Comparison of Performance Characteristics

A patient was considered positive for malaria if either CAP or VEN blood was positive for malaria in qPCR. Thus, any of the permutations CAP positive and VEN positive, CAP positive and VEN negative, and CAP negative and VEN positive con-stituted the gold standard definition of malaria for this study, against which performance characteristics of microscopy of CAP and VEN blood were evaluated.

Statistical Considerations

Stata/SE15.1 software (StataCorp) was used for statistical anal-ysis. Stata software packages were downloaded for computation of sensitivity, specificity, positive predictive value (PPV), nega-tive predictive value (NPV) [23], and likelihood ratios [24]. Area under the receiver operating characteristic (ROC) curve was computed with Stata’s “roctab” command. Pearson r was calcu-lated for correlation analysis of CAPparasitemia and VENparasitemia and intraclass correlation coefficients (ICCs) were created to assess the reproducibility of CAPparasitemia in VEN blood, regard-ing patients as random effects and the blood source as fixed effect. Absolute agreement was selected and average ICCs were reported for microscopic data (as multiple readers determined the defini-tive result) and individual ICCs were reported for qPCR (as only one result determined the definitive result). A significance level of 2-sided α < .05 was considered as statistically significant.

Two small and mainly descriptive studies were identified having investigated a potential CAP-VEN parasite difference and using light microscopy as the sole method for parasite detection [3, 4]. Data from these studies were extracted and, alongside microscopic data from this present study, odds ratios (ORs) for paired data were computed as described in the literature [25]. In this context, such ORs quantify the likelihood of parasite detection in one sampling site compared to the other. Odds ratios of these studies were eval-uated in meta-analysis. Meta-analytical summary measures and forest plots were computed with Stata’s “metaan” command [26].

According to sample size calculations, which are demon-strated in the Supplementary Materials, and due to fluctuations of malaria prevalence among local patient populations overall, recruitment was planned to continue until reaching a sample of 375 patients. Biological samples were then shipped to Europe for further parasitological analysis.

Ethical ConsiderationsThe study was conducted according to the Declaration of Helsinki, the applicable guidelines for Good Clinical Practice as defined by the “International Conference On Harmonization Of Technical Requirements For Registration Of Pharmaceuticals For Human Use”, and the applicable laws and regulations of Gabon. The study was approved by the independent ethics committee of the Centre de Recherches Médicales de Lambaréné (reference number: CEI-CERMEL, 009/2015). Upon diagnosis of malaria, an effective anti-malarial treatment was provided to the participants at no cost.

RESULTS

Baseline Characteristics

Three hundred seventy-six patients were recruited into this diagnostic study. Among those, 346 participants were eligible for analysis of performance characteristics, as well as 204 and 108 participants for analysis of quantitative differences between CAPparasitemia and VENparasitemia in microscopy and qPCR, respectively (Figure 1). The median age of participants in var-ious analyses was between 11 and 15 years with a female/male ratio of almost 1 (Table 1).

CAPparasitemia and VENparasitemia: Correlations, Reproducibility,

and Differences

CAPparasitemia and VENparasitemia as determined by microscopy and qPCR show good correlation with Pearson r of >0.9 for the respective analyses (Table 2). ICCs demonstrate excellent reproducibility of CAPparasitemia in VEN blood with values con-sistently >90% in all models (P = .0001). Median (interquar-tile range [IQR]) CAPparasitemia was slightly higher than median VENparasitemia, and inspection of CAP-VEN differences over the whole range of measurements demonstrates a small but signif-icant excess parasitemia in CAP blood (Table 2 and Figure 2). Microscopy quantified the mean CAPparasitemia to be +16.6% (95% confidence interval [CI], +.7% to +35%) higher than VENparasitemia. Concordantly, qPCR analysis demonstrated that fewer cycles (–0.278 [95% CI, –.473 to –.083] were required for parasite signal detection in CAP blood than in VEN blood.

Thirty-four patients were identified as gametocyte carriers. Gametocytes were 4.2 times more likely to be detected in CAP blood than in VEN blood (odds ratio [OR], 4.2; P = .0017) and CAP gametocytemia was +55.9% (95% CI, +22.1% to +99%) higher than VEN gametocytemia (Supplementary Tables 1 and 2).

A stratified analysis per Plasmodium species indicated that this CAP-VEN difference occurred both in Plasmodium falcip-arum monoinfections with +17.9% (95% CI, +6.4% to +37.9%) and in nonfalciparum infections with +14.4% (95% CI, –47% to +146.7%) more parasites in CAP blood relative to VEN blood (Table 2). Mixed infections showed no such difference, however, being highly likely due to sampling variation (P = .95).

Performance Characteristics

The proportion of false-negative microscopic results was higher in VEN samples (26.6%) than in CAP samples (18.5%) (Table 3) and microscopically determined parasitemias of false-negatives were low with a median of 27 (IQR, 16–256) and 43.5 (IQR, 32–86) parasites/μL, respectively. This indicates a difference in sensitivity of approximately 8% (sensitivity, 81.5% for CAP vs 73.4% for VEN), whereas specificities remained roughly equal at 91%. This results in a more favorable positive likelihood ratio for CAP samples, which is approximately 2 units higher than for VEN samples (10.1 vs 8.3), as well as a more favorable negative likelihood ratio (0.2 vs 0.29). Both area under the ROC curves

Dow



ovember 2018

http://academic.oup.com/jid/article-lookup/doi/10.1093/infdis/jiy319#supplementary-data




are >0.8; although 95% CIs did not go below 0.8 for CAP sam-ples (.83–.9), they did so in VEN samples (.78–.86) (Table 4). Investigation of a subpopulation of low-level parasitemias (<250 parasites/μL) decreased CAP sensitivity and VEN sensitivity; however, simultaneously increasing the above-mentioned sensi-tivity difference (sensitivity, 63.3% for CAP vs 45.9% for VEN). Specificities remained high at approximately 92% (Table 4).

A cross-tabulation of quantitative polymerase chain reaction (qPCR) results shows that malaria DNA was twice more likely to be detected in CAP blood than in VEN blood (odds ratio, 2.17; P = .11; Supplementary Table 3).

Meta-analytical summary measures demonstrate that malaria parasites were 2.63 (95% CI, 1.68–4.08) times more likely to be detected by light microscopy in CAP blood than in VEN blood (Figure 3 and Supplementary Table 4). Measures of heteroge-neity did not indicate heterogeneity (P = .96; Supplementary Table 5).

DISCUSSION

This study investigated the diagnostic performance characteris-tics of 2 easy-to-access blood sources, capillary and venous blood. CAP blood was shown to be of diagnostic superiority, manifest-ing in an 8% sensitivity gap between CAP and VEN blood in the overall study population and widening by >2-fold in a subpopu-lation of low-level parasitemias (sensitivity gap of 17%).

This CAP superiority in sensitivity can be explained by the comparatively higher parasite abundance in CAP blood as

diagnosis of malaria in microscopy is based on the demonstra-tion of asexual parasites. Microscopy has a theoretical LOD of approximately 10 parasites/μL [10–12], whereas under real-life settings in endemic areas, a LOD of approximately 90 parasites/μL is typical [27]. Particularly in cases of poorly trained per-sonnel, LODs may rise considerably above these limits. In the study presented here, blood smears were investigated by expert microscopists, who had been assessed by dedicated examina-tions for malaria diagnostics according to WHO standards. However, it is concluded that even a small CAP-VEN difference in parasitemia of approximately 15% may facilitate diagnosis in CAP blood (as just above the LOD), but not in VEN blood (as not yet surpassed the LOD). The diagnostic superiority in favor of CAP blood could potentially rise inversely with the degree of microscopist training. As LODs would accordingly rise to higher parasitemias (eg, 100 parasites/μL vs 10 parasites/μL) for less well-trained microscopists, an approximate 15% difference would result in a higher absolute value of parasite difference (15 parasites/μL vs 1.5 parasites/μL), thereby potentially increasing the sensitivity gap. Given the lack of microscopists with ade-quate training and time to read many fields in resource-limited, malaria-endemic settings, this is a potential topic of concern. However, in high-resource settings where malaria is not en-demic, increased LODs are also likely, as uncommon or rare diseases may not receive sufficient resources and training.

Irrespective of blood source, sensitivity of microscopy can be increased by investigating larger sample volumes. In our

376 patients recruited

348 (93%) patients for whommatched CAP and VEN qPCR

samples were available withsu�cient quality

28 patients excludedReason for exclusion:- No paired DNA sample obtained (either CAP or VEN sample missing) (28)

qPCR

240 patients excludedReasons for exclusion:- Negative for malaria in CAP & VEN (124)- CAP blood directly from finger (97)- Positive in only CAP or VEN sample (19)

- CAP+ & VEN– = (13)- CAP– & VEN+ = (6)

12 patients excludedReason for exclusion:- No matching qPCR sample available (12)

154 patients excludedReason for exclusion:- Negative for malaria in CAP & VEN (143)- Discordant pair and qPCR-negative (11)

Quantification of parasite di�erence inlight microscopy

n = 204 (57%)

Quantification of parasite di�erence inqPCR

n = 108 (31%)

Analysis of performance characteristics

n = 346 (97%)

358 (95%) patients for whom matchedCAP and VEN microscopy samples were

available with su�cient quality

Light microscopy

18 patients excludedReasons for exclusion:- Insu�cient quality of slide pair (8)- No pair obtained (either CAP or VEN sample missing) (10)

Figure 1. Study flowchart. Abbreviations: CAP, capillary blood; qPCR, quantitative polymerase chain reaction; VEN, venous blood.

Dow



ovember 2018

https://academic.oup.com/jid/article-lookup/doi/10.1093/infdis/jiy319#supplementary-data





experience, finger pricks were more accepted than venous punctures, and sampling CAP blood quantities of up to 100 μL was well tolerated. Such quantities allow the preparation of mul-tiple thick smears facilitating the microscopic investigation of >1000 fields, thereby further lowering LODs.

Sensitivity and specificity are fixed assay-specific properties. However, PPVs and NPVs combine these assay properties with characteristics of the respective target population and constitute the degree to which one can trust in the correctness of the respec-tive positive or negative test result [28]. While PPVs are affected by low disease prevalence, NPVs are affected by high prevalence. Examples of changing PPVs and NPVs for different prevalence settings were modeled and are depicted in the Supplementary

Materials. CAP blood may be of particular diagnostic benefit in low-prevalence settings of malaria to rule in malaria as differential diagnosis (by higher PPV) and help in high-prevalence settings to rule out malaria as differential diagnosis (by higher NPV).

Importantly, microscopic findings were further supported by molecular detection of parasite DNA. Results of micros-copy are concordant to findings of qPCR, in that fewer cycles were necessary to detect parasite DNA in CAP blood com-pared to VEN blood. The employed qPCR protocol has a LOD of approximately 1 parasite/μL blood [16, E. Tannich, personal communication, 2017]. Thus, the CAP-VEN difference can be extrapolated to roughly one-tenth of the LOD of microscopy while it remains uncertain how this effect would behave for

Microscopy, quantification of parasite di�erence (n = 204)

6A

B

4

2

0

–2

Log

2 (C

AP

para

site

mia

/VE

N pa

rasi

tem

ia)

–4

–6

1 2 3Log10[(CAP parasitemia + VEN parasitemia)/2]

qPCR, quantification of parasite di�erence (n = 108)

(CAP cycles + VEN cycles)/2

CA

P cy

cles

– V

EN

cyc

les

15

4

2

0

–2

–4

20 25 30

4 5 6

Figure 2. A, Microscopic comparison of capillary blood (CAP) parasitemia and venous blood (VEN) parasitemia. B, Quantitative polymerase chain reaction (qPCR) com-parison of CAP cycles and VEN cycles. Solid lines indicate mean differences; long dashed lines indicate 95% confidence intervals for mean differences; short dashed line indicates line of equality.

Dow



ovember 2018




much lower parasite densities. Interestingly, qPCR detected parasite DNA only in CAP blood in twice as many patients than only in VEN blood (OR, 2.17; P = .11; Supplementary Table 3).

Erythrocytes infected by P. falciparum are known to contain a number of highly variable parasite proteins on their surface, such as P. falciparum erythrocyte membrane protein 1, repeti-tive interspersed family, or sub-telomeric variable open reading frame [29–31]. These molecules show affinity to certain vascular receptors, such as intercellular adhesion molecule 1 (ICAM-1) [32, 33]. ICAM-1 is expressed in CAP blood vessels in response to proinflammatory cytokines whose production is triggered by malarial pathogen-associated molecular patterns during a

malaria infection [29, 34, 35]. This pathophysiological phenom-enon could thus serve as explanation for the higher parasitemia of CAP blood relative to VEN blood. Furthermore, infected erythrocytes that do not bind to vascular endothelium have a higher chance to become eliminated in the spleen, thereby also contributing to a relatively lower parasitemia in VEN blood.

This study demonstrates a significant association between gametocyte carrier status and CAP blood sampling (OR, 4.2; P = .0017), potentially owing to a higher CAP gametocytemia (+55.9%; P = .0008) compared with VEN gametocytemia. This seems plausible as CAP blood is used by Anopheles mosquitos during blood meals and gametocytes are parasite stages aiming

Odds ratios for paired data and 95% CIs

Favors VEN blood

–.5 0 .5 1 1.5 2

Log-odds scale

Favors CAP blood

Ouedraogo, 1991

Njunda, 2013

Studies

Present study, 2018

Overall e�ect (fe)

Figure 3. Forest plot depicting effect sizes of previous studies. Odds ratios (with 95% confidence intervals) for paired data are as follows: Ouedraogo et al, 1991 [4]: 2.8 (.95–7.3). Njunda et al, 2013 [3]: 2.8 (1.18–4.98). Present study, 2018: 2.42 (1.43–5.48). Overall effect: 2.63 (1.68–4.08). Abbreviations: CI, confidence interval; CAP, capillary blood; fe, fixed effects model; VEN, venous blood.

Table 1. Baseline Characteristics

Characteristics Total (N = 376)Microscopy: Quantification of Parasite

Difference (n = 204)qPCR: Quantification of Parasite

Difference (n = 108)Performance Characteristics

(n = 346)

Age, y

Median (IQR) 14 (5–39) 11 (5–24.5) 13 (7–34) 15 (5–40)

<5 73 (19) 44 (22) 16 (15) 66 (19)

5−14 117 (31) 81 (40) 45 (42) 106 (31)

15−24 47 (13) 28 (14) 11 (10) 43 (12)

25−34 38 (10) 18 (9) 10 (9) 33 (10)

35−44 23 (6) 9 (4) 5 (5) 23 (7)

45−54 28 (7) 7 (3) 3 (3) 24 (7)

≥55 50 (13) 17 (8) 18 (17) 50 (14)

Male sex 186 (49) 98 (48) 51 (47) 173 (50)

Data are presented as No. (column %) unless otherwise indicated.

Abbreviations: IQR, interquartile range; qPCR, quantitative polymerase chain reaction.

Dow



ovember 2018



Tabl

e 2.

Co

mpa

riso

n of

Cap

illar

y B

lood

Par

asite

mia

and

Ven

ous

Blo

od P

aras

item

ia U

sing

Mic

rosc

opy

and

Qua

ntita

tive

Poly

mer

ase

Chai

n Re

actio

n

Ana

lysi

sIn

fect

ion

No.

Med

ian

Para

site

mia

(IQ

R)a

Diff

eren

ce C

AP

– V

EN

(95%

CI)

Pear

son

r

Rep

rodu

cibi

lity

of C

AP

para

site

mia in

V

EN

para

site

mia

c

CA

PV

EN

Ligh

t M

icro

scop

y, %

(9

5% C

I)bP

Valu

edqP

CR

e , A

vera

ge

Cyc

les

(95%

CI)

P Va

lued

ICC

, %

(95%

CI)

P Va

luef

Mic

rosc

opic

an

alys

isM

icro

scop

ic a

naly

sis;

to

tals

204

495

(85–

3243

)42

9 (5

2–40

74)

+16

.6 (+

.7 t

o +

35)

.04

––

0.92

95 (9

3.6–

96.3

).0

001

Plas

mod

ium

falc

ipar

um

mon

oinf

ectio

n18

351

2 (7

9–48

00)

433

(46–

4750

)+

17.9

(+6.

4 to

+37

.9)

.04

––

0.92

95 (9

3.4–

96.3

).0

001

Mix

ed in

fect

ion

(P. f

alci

p-ar

um +

non

falc

ipar

um

spp)

1069

9 (2

15–1

1 11

4)79

7 (2

82–1

0 30

6)−

1 (–

29.3

to

+38

.5)

.95

––

0.97

99 (9

6.8–

99.8

).0

001

Non

falc

ipar

um in

fect

ion

(Onl

y Pl

asm

odiu

m

oval

e +

Pla

smod

ium

m

alar

iae)

1119

3 (2

7–76

0)10

1 (2

1–88

8)+

14.4

(–47

to

+14

6.7)

.71

––

0.8

97 (9

0.7–

99.3

).0

001

Mic

rosc

opy

anal

ysis

; w

hen

qPC

R g

old

stan

dard

was

pos

itive

187

519

(85–

4349

)44

8 (5

5–40

91)

+16

.7 (–

.8 t

o +

37.3

).0

6–

–0.

9195

(93.

5–96

.3)

.000

1

qPC

R a

naly

sis

Tota

ls10

825

.735

(23.

27–2

7.35

)25

.95

(23.

585–

27.3

25)

––

–0.2

78 (–

.473

to

–.08

3).0

060.

9493

(90.

1–95

.2)

.000

1

Abb

revi

atio

ns: C

AP,

cap

illar

y bl

ood;

CI,

confi

denc

e in

terv

al; I

CC

, int

racl

ass

corr

elat

ion

coef

ficie

nt; I

QR

, int

erqu

artil

e ra

nge;

qP

CR

, qua

ntita

tive

poly

mer

ase

chai

n re

actio

n; V

EN

, ven

ous

bloo

d.a U

nits

: par

asite

s pe

r m

icro

liter

for

light

mic

rosc

opy;

cyc

les

for

qPC

R.

b Lig

ht m

icro

scop

y; +

, mor

e pa

rasi

tes

in c

apill

ary

bloo

d; –

, mor

e pa

rasi

tes

in v

enou

s bl

ood.

c Unl

ogge

d pa

rasi

tem

ia w

as u

sed

to a

sses

s IC

Cs.

d Pai

red

t tes

t: H

0: lo

g 2(C

AP

para

site

mia) =

log 2(

VE

Npa

rasi

tem

ia).

e qP

CR

: –, m

ore

para

site

DN

A in

CA

P b

lood

; +, m

ore

para

site

DN

A in

VE

N b

lood

.f F

tes

t.

Dow



ovember 2018


at infecting the insect vector. However, these findings need to be treated with caution, as the study was not primarily powered to detect such an effect.

The present study is in discordance with a previously pub-lished study with similar objectives though with different meth-odology. This study concluded that there was no difference in asexual parasitemia between CAP and VEN blood (null hypo-thesis: CAPparasitemia = VENparasitemia; P = .47) [36]. The authors recategorized the continuous outcome parasitemia into an or-dinal response variable of 3 categories—“0 parasites,” “1–5000 parasites,” “>5000 parasites”—and used a cumulative linked mixed model to assess whether either CAP or VEN blood ex-posure had an influence on the ordinal outcome. It is possible that the somewhat arbitrarily chosen bounds may not have been sensitive enough to detect small CAP-VEN differences (~15%), as were demonstrated in the present study. Also, with a sample size of 137, the study was likely underpowered to detect small

effects, as indicated by sample size calculations of the present study. Performing power calculations in analogy to sample size calculations as described in the Supplementary Materials, our study has an observed power >95% for the comparison of par-asitemia of CAP-VEN blood in both microscopy and qPCR analyses, and analyses of performance characteristics have an observed power between 80% and 90% [37]. Last, different observations may also be explained by a biological effect, as the above-cited study recruited asymptomatic carriers, whereas our study included symptomatic malaria patients. Sensitivity analyses, however, should also apply to asymptomatic carriers. In line with this assumption, 2 smaller studies on this topic demonstrated concordant results to the present study [3, 4]. Neither of the studies performed qPCR or sensitivity analyses but demonstrate that malaria parasites were approximately 2.8 times more likely to be detected in CAP samples than in VEN samples (Figure 3).

Strengths of this study include the high methodological qual-ity under which blood sampling and parasitological analysis were performed. Importantly, while sensitivities differed between CAP and VEN blood samples, specificities were approximately equal at 91%. Thus, the resulting similar false-positivity rate is indicative that assessor blinding was not infringed. Yet, it is a limitation of this study that CAP blood sampling tubes were not always available throughout the study period.

This study reports important evidence for research studies of various kinds: (1) clinical trials, particularly on antimalarial chemotherapy, for which reduction of parasitemia is often an important endpoint; (2) clinical trials that are associated with low parasitemias such as malaria vaccine trials and may be af-fected by the source of blood taken for malaria diagnostics; (3) epidemiological studies aiming at determining the true preva-lence of malaria in a given context. To allow generalization of study results in a broad and multinational context, future epi-demiological and clinical studies should mention whether CAP

Table 4. Comparison of Performance Characteristics of Microscopically Analyzed Capillary and Venous Blood Samples Against Quantitative Polymerase Chain Reaction Gold Standard

Performance Characteristic

Totals (All Parasitemias) (n = 346)Low-Level Parasitemias (<250 Parasites/μL)

(n = 231)

CAP VEN CAP VEN

Sensitivity, % 81.5 (77.4–85.6) 73.4 (68.8–78.1) 63.3 (57.1–69.5) 45.9 (39.5–52.3)

Specificity, % 91.9 (89.1–94.8) 91.1 (88.1–94.1) 92.6 (89.3–96.0) 91.8 (88.3–95.3)

PPV, % 94.8 (92.4–97.1) 93.7 (91.1–96.2) 88.5 (84.3–92.6) 83.3 (78.5–88.1)

NPV, % 73.6 (68.9–78.2) 65.7 (60.7–70.7) 73.9 (68.2–79.5) 65.5 (59.4–71.6)

Positive LR 10.11 (5.56–18.38) 8.28 (4.68–14.63) 8.58 (4.5–16.35) 5.6 (2.99–10.49)

Negative LR 0.2 (.15–.27) 0.29 (.23–.37) 0.4 (.31–.51) 0.59 (.49–.71)

Area under ROC curve 0.87 (.83–.9) 0.82 (.78–.86) 0.78 (.73–.83) 0.69 (.64–.74)

Prevalencea, % 64.2 (59.1–69.2) 64.2 (59.1–69.2) 47.2 (40.8–53.6) 47.2 (40.8–53.6)

Data in parentheses indicate the 95% confidence interval.

Abbreviations: CAP, capillary blood; LR, likelihood ratio; NPV, negative predictive value; PPV, positive predictive value; ROC, receiver operating characteristic; VEN, venous blood.aPrevalence of malaria as determined by qPCR.

Table 3. Two-Way Tables of Capillary and Venous Blood as Assessed by Light Microscopy Cross-Tabulated Against a Quantitative Polymerase Chain Reaction Gold Standard

Microscopic Sample (n = 346)

qPCR (CAP or VEN)

Positivea Negativeb

CAPc

Positive 181 (81.5) 10 (8.1)

Negative 41 (18.5) 114 (91.9)

VENc

Positive 163 (73.4) 11 (8.9)

Negative 59 (26.6) 113 (91.1)

Data are presented as No. (column %).

Abbreviations: CAP, capillary blood; qPCR, quantitative polymerase chain reaction; VEN, venous blood.aPositive = CAP positive and VEN positive, CAP positive and VEN negative, or CAP negative and VEN positive.bNegative = CAP negative and VEN negative.cχ2 test: P < .0001.

Dow



ovember 2018



or VEN blood was used for determination of malaria parasit-emia and what sample volume was investigated. If both CAP and VEN blood samples are used to determine parasitemia, a correction factor should be applied especially in the context of repeated measurements. If CAP samples constitute the mi-nority in a series of repeated measurements, then 15% of the respective CAPparasitemia should be subtracted from its value (CAPparasitemia corrected = CAPparasitemia uncorrected – 0.15 × CAPparasitemia uncorrected). Concordantly, if VEN samples con-stitute the minority, then 15% should be added (VENparasitemia corrected = VENparasitemia uncorrected + 0.15 × VENparasitemia uncorrected). Last, this study underlines the importance of CAP sampling in the context of malaria elimination campaigns particularly in settings of perennial endemicity where people with low-level parasitemias constitute an important reservoir of malaria parasites [38]. Approaches aiming at this target group should use CAP blood for sampling purposes.

Currently no distinction is being made with regard to the blood source from the sample that was used for malaria diag-nostics. Considering an 8% increase in sensitivity by using capillary blood, an important proportion of patients may be misdiagnosed if venous blood is used for laboratory analysis. Due to the global burden of malaria, a programmatic recom-mendation of CAP blood for diagnosis of malaria could entail a considerable global public health benefit if these findings are reproduced and confirmed in diverse epidemiological settings.

CONCLUSIONS

This study investigated the diagnostic performance charac-teristics of microscopy and qPCR using CAP and VEN blood samples. A CAP-VEN sensitivity gap of 8% was revealed that increases to 17% for low-level parasitemias. CAP blood sampling is simple and cost-efficient, thus being potentially beneficial to optimize routine diagnostics in both low- and high-resource settings. Thus, using the adequate source of blood samples for malaria diagnostics may be of high importance.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or com-ments should be addressed to the corresponding author.

Notes

Author contributions. Conceptualization: J. M., M. R., G. M.-N., P. G. K., B. L. Data curation: J. M., P. P. Statistical anal-ysis: J. M. Parasitological analysis: J. M., P. P., E. T., L. V., M. G. Funding acquisition: J. M., M. R. Investigation: J. M., G. M.-N., L. V., R. Z.-M. Methodology: J. M., M. R., G. M.-N. Project administration: J. M. Resources: A. A. A., S. T. A., P. G. K., B. L.,

M. R. Supervision: M. R., B. M. Visualization: J. M. Original draft preparation: J. M. Review and editing of the manuscript: all authors.

Acknowledgments. We thank Heide-Maria Winkler, Veronika Reischer, Markus Obermüller, Albert Lalremruata, Stefanie Ruben, and Michael Kundi for providing help regard-ing parasitological analysis.

Financial support. This study was funded by the Federal Ministry of Science, Research and Economy of Austria as part of the EDCTP-2 program, which is supported by the European Union. We also acknowledge project funding from the Österreichische Gesellschaft für Antimikrobielle Chemotherapie.

Potential conflicts of interest. All authors: No potential conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manu-script have been disclosed.

References

1. World Health Organization. World malaria report 2017. Geneva, Switzerland: WHO, 2017.

2. Gubler DJ. Resurgent vector-borne diseases as a global health problem. Emerg Infect Dis 1998; 4:442–50.

3. Njunda AL, Assobc NJC, Nsaghab SD, Kamgaa FHL, Mokenyua MD, Kwentia TE. Comparison of capillary and venous blood using blood film microscopy in the detection of malaria parasites: a hospital based study. Sci J Microbiol 2013; 2:89–94.

4. Ouedraogo JB, Guiguemde TR, Gbary AR. Etude com-parative de la densite parasitaire de Plasmodium falcip-arum dans le sang capillaire et dans le sang veineux chez des porteurs asymptomatiques (région de Bobo-Dioulasso, Burkina Faso) [in French]. Médecine d’Afrique Noire 1991; 38:601–5.

5. Adegnika AA, Breitling LP, Agnandji ST, et al. Effectiveness of quinine monotherapy for the treatment of Plasmodium falciparum infection in pregnant women in Lambaréné, Gabon. Am J Trop Med Hyg 2005; 73:263–6.

6. Borrmann S, Adegnika AA, Moussavou F, et al. Short-course regimens of artesunate-fosmidomycin in treat-ment of uncomplicated Plasmodium falciparum malaria. Antimicrob Agents Chemother 2005; 49:3749–54.

7. Macintyre F, Adoke Y, Tiono AB, et al; OZ-Piperaquine Study Group. A randomised, double-blind clinical phase II trial of the efficacy, safety, tolerability and pharmacokinet-ics of a single dose combination treatment with artefenomel and piperaquine in adults and children with uncomplicated Plasmodium falciparum malaria. BMC Med 2017; 15:181.

8. Ramharter M, Adegnika AA, Agnandji ST, et al. History and perspectives of medical research at the Albert Schweitzer

Dow



ovember 2018


Hospital in Lambaréné, Gabon. Wien Klin Wochenschr 2007; 119:8–12.

9. Manego RZ, Mombo-Ngoma G, Witte M, et al. Demography, maternal health and the epidemiology of malaria and other major infectious diseases in the rural department Tsamba-Magotsi, Ngounie Province, in central African Gabon. BMC Public Health 2017; 17:130.

10. World Health Organization. Basic malaria microscopy part I. Learner’s guide. Geneva, Switzerland: WHO, 2010.

11. World Health Organization. Basic malaria microscopy part ll. Tutor’s guide. Geneva, Switzerland: WHO, 2010.

12. World Health Organization. Microscopy for the detec-tion, identification and quantification of malaria parasites on stained thick and thin blood films in research settings. Geneva, Switzerland: WHO, 2015.

13. Macintyre F, Adoke Y, Tiono AB, et al; OZ-Piperaquine Study Group. A randomised, double-blind clinical phase II trial of the efficacy, safety, tolerability and pharma-cokinetics of a single dose combination treatment with artefenomel and piperaquine in adults and children with uncomplicated Plasmodium falciparum malaria. BMC Med 2017; 15:181.

14. Billo MA, Diakité M, Dolo A, et al. Inter-observer agree-ment according to malaria parasite density. Malar J 2013; 12:335.

15. Qiagen. QIAamp DNA blood mini kit. https://www.qia-gen.com/us/shop/sample-technologies/dna/genomic-dna/qiaamp-dna-blood-mini-kit/#productdetails. Accessed 17 December 2017.

16. Altona Diagnostics. RealStar malaria PCR kit. https://www.altona-diagnostics.com/tl_files/website/downloads/RealStar%C2%AE%20Malaria%20PCR%20Kit%201.0_CE_RevS01_WEB.pdf. Accessed 17 December 2017.

17. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307–10.

18. Quackenbush J. Microarray data normalization and trans-formation. Nat Genet 2002; 32:496–501.

19. Bland JM, Altman DG. The use of transformation when comparing two means. BMJ 1996; 312:1153.

20. Bland JM, Altman DG. Transformations, means, and confi-dence intervals. BMJ 1996; 312:1079.

21. Gibbons R, Coleman D. Statistical methods for detection and quantification of environmental contamination. Vol. 1. New York, USA: John Wiley & Sons, 2001.

22. Krishnamoorthy K, Mallick A, Mathew T. Model-based imputation approach for data analysis in the presence of non-detects. Ann Occup Hyg 2009; 53:249–63.

23. Tobias A. DIAGTEST: Stata module to report summary statistics for diagnostic test. https://ideas.repec.org/c/boc/bocode/s365601.html. Accessed 17 December 2017.

24. Seed P. DIAGT: Stata module to report summary statistics for diagnostic tests compared to true disease status. https://ideas.repec.org/c/boc/bocode/s423401.html. Accessed Accessed 17 December 2017.

25. Kirkwood BR, Sterne JAC. Essential medical statistics. Oxford, UK: Blackwell Science, 2003.

26. Kontopantelis E, Reeves D. METAAN: Stata module to per-form fixed- or random-effects meta-analyses. https://ideas.repec.org/c/boc/bocode/s457071.html. Accessed 3 May 2018.

27. Joanny F, Löhr SJ, Engleitner T, Lell B, Mordmüller B. Limit of blank and limit of detection of Plasmodium falcip-arum thick blood smear microscopy in a routine setting in Central Africa. Malar J 2014; 13:234.

28. Mischlinger J, Schernhammer E. A common trap of diag-nostic tests: disease prevalence and positive predictive value. Wien Klin Wochenschr 2017; 129:583–4.

29. Chakravorty SJ, Craig A. The role of ICAM-1 in Plasmodium falciparum cytoadherence. Eur J Cell Biol 2005; 84:15–27.

30. Goel S, Palmkvist M, Moll K, et al. RIFINs are adhesins implicated in severe Plasmodium falciparum malaria. Nat Med 2015; 21:314–7.

31. Niang M, Bei AK, Madnani KG, et al. STEVOR is a Plasmodium falciparum erythrocyte binding protein that mediates merozoite invasion and rosetting. Cell Host Microbe 2014; 16:81–93.

32. Anchang-Kimbi JK, Achidi EA, Nkegoum B, Sverremark-Ekström E, Troye-Blomberg M. Diagnostic comparison of malaria infection in peripheral blood, placental blood and placental biopsies in Cameroonian parturient women. Malar J 2009; 8:126.

33. Akum AE, Kuoh AJ, Minang JT, Achimbom BM, Ahmadou MJ, Troye-Blomberg M. The effect of maternal, umbilical cord and placental malaria parasitaemia on the birthweight of newborns from south-western Cameroon. Acta Paediatr 2005; 94:917–23.

34. Murphy K. Janeway’s immunobiology. 8th ed. New York, USA: Garland Science, 2011.

35. Newbold C, Craig A, Kyes S, Rowe A, Fernandez-Reyes D, Fagan T. Cytoadherence, pathogenesis and the infected red cell surface in Plasmodium falciparum. Int J Parasitol 1999; 29:927–37.

36. Sandeu MM, Bayibéki AN, Tchioffo MT, et al. Do the venous blood samples replicate malaria parasite densities found in capillary blood? A field study performed in natu-rally-infected asymptomatic children in Cameroon. Malar J 2017; 16:345.

37. Beam CA. Strategies for improving power in diagnostic radiology research. AJR Am J Roentgenol 1992; 159:631–7.

38. World Health Organization. Global technical strategy for malaria 2016–2030. Geneva, Switzerland: WHO, 2015.

Dow



ovember 2018

https://www.qiagen.com/us/shop/sample-technologies/dna/genomic-dna/qiaamp-dna-blood-mini-kit/#productdetails



https://www.altona-diagnostics.com/tl_files/website/downloads/RealStar%C2%AE%20Malaria%20PCR%20Kit%201.0_CE_RevS01_WEB.pdf




https://ideas.repec.org/c/boc/bocode/s365601.html






Use of Capillary Blood Samples Leads to Higher Parasitemia ... · malaria control strategies and,...

Documents

Transcript of Use of Capillary Blood Samples Leads to Higher Parasitemia ... · malaria control strategies and,...