Paper‐Origami‐Based Multiplexed Malaria Diagnostics from Whole Blood

Abstract We demonstrate, for the first time, the multiplexed determination of microbial species from whole blood using the paper‐folding technique of origami to enable the sequential steps of DNA extraction, loop‐mediated isothermal amplification (LAMP), and array‐based fluorescence detection. A low‐cost handheld flashlight reveals the presence of the final DNA amplicon to the naked eye, providing a “sample‐to‐answer” diagnosis from a finger‐prick volume of human blood, within 45 min, with minimal user intervention. To demonstrate the method, we showed the identification of three species of Plasmodium, analyzing 80 patient samples benchmarked against the gold‐standard polymerase chain reaction (PCR) assay in an operator‐blinded study. We also show that the test retains its diagnostic accuracy when using stored or fixed reference samples.

Nucleic acid based tests (NATs) offer the promise of microbial diagnostics,d etermining either the species present or characteristics of the pathogen, such as drug resistance.The gold-standard assay used in many reference laboratories is based upon ap olymerase chain reaction (PCR) amplification-a technology which achieves high sensitivities but which also requires trained staff and external power. In contrast, identification of microbial species in resourcelimited environments requires low cost, simple tests that do not need external or fixed power supplies.O ne example where such asimple low-cost test could transform outcomes is in malaria diagnosis,w here species identification directly informs patient treatment.
Classical malaria diagnosis involves ab lood smear followed by microscopy,w hich, although simple,d oes not provide the required sensitivity and only enables species specific information in the hands of trained experts. [1] New approaches will be required to tackle the disease,w here asymptomatic individuals commonly harbor the disease at levels that are below the sensitivity of microscopy (< 100 parasites/mL). [2] Nucleic acid based tests (NATs) offer the promise of achieving such high sensitivities (1 parasite/mL) with excellent specificity, [3] enabling healthcare professionals to inform treatment. [4] As many people living at risk of malaria infection have no access to diagnosis,p resumptive treatment of all febrile patients as if they were malaria cases is ac ommon practice, which has become as erious problem, especially in sub-Saharan Africa. [5] Currently,t he most widely adopted NAT method in infectious disease diagnosis is still PCR, although the reliance on thermocycling has proven ab arrier to its implementation in low-resource settings. [6] LAMP has emerged as al ow-cost alternative, [7] simplifying hardware requirements whilst enabling visual detection. [8] Although amenable to multiplexing,t he high number of primers required (up to six per target) [9] restricts the number of targets that tests can detect in one reaction. Ac ommercial Plasmodium genus LAMP test is available,b ut the system requires am ulti-step DNAe xtraction based on the PURE methodology,c arried out on ab ench-top instrument. Alternatively,l ateral flow tests,c ommonly referred to as "rapid diagnostic tests" (RDT), also exist, [10] but their sensitivity is poor. [11] Herein, we show an ew capillary-flow platform that combines ease-of-use and low-cost with the sensitivity of LAMP,i nto am ultiplexed three taxon-specific test plus ac ontrol. We overcome the difficulties linked to sample preparation and multiplexing using capillary wicking and paper-folding origami techniques to distribute fluids both vertically and laterally. [12,13] Previously paper microfluidics (see Review [14] )h as enabled single units of aN AT,such as DNAextraction, [15] DNA isothermal amplification, [16] which have been integrated into amanual "machine" [17] with hybridization-based DNAdetection. [18] We now integrate all the required steps into as ingle device to detect Plasmodium falciparum, Plasmodium vivax, [19] and Plasmodium pan directly from af inger-prick volume of whole blood within an operator-blinded study ( Figure 1).
Often in-field testing cannot be performed for logistical reasons,a nd retrospective diagnosis is required. Te sting of archived blood samples is also important for epidemiologists to re-visit reference samples or to analyze historical data sets. [20] We therefore show that we can identify parasites in preserved samples of frozen whole blood, as thick and thin fixed smear samples on glass slides,and as whole blood dried onto paper.
Thefabrication of the device using wax printing, [21] as well as the operating steps,shown in Figure 1and Figure 2A), the paper was folded (structure S1, Figure 2B)toenable the first steps of the assay,i nvolving cell lysis and DNAe xtraction, to yield purified DNA( Figure 2B-D) on the glass-fiber paper.T o transfer the DNAf rom the extraction panel to the amplification panel, the fold S1 is flipped on the opposite side ( Figure 2E), allowing elution (Panels 4-5 of Figure 2A and F). Supporting Information Figure S1(A) illustrates the extraction process.
Multiplexing analysis was enabled by using capillarity to guide the sample to four independent locations on the paper within hot wax printed channels,w here species-specific LAMP reagents were deposited (Figure 2A,p anel 5). The system was sealed by an acetate film to prevent evaporation during incubation ( Figure 1B and D) and amplification was carried out 63 8 8Cf or up to 45 min on as imple hotplate.T he results of species-specific LAMP were initially read-out by the naked eye with ah andheld UV lamp (365 nm;F igure 3A). [8] We also showed that the technique was amenable to quantification ( Figure 3B,C). To test sensitivity,w eu sed the WHO International Standard for P. falciparum DNA, [22] which was serially diluted from 10 to 10 4 times.T he real-time amplification curves ( Figure 3B)w ere normalized to 1f or ease of comparison and show that sensitivity down to 10 5 IU/mL can be achieved within 35 min. As the copy number decreases,s o the exponential phase of signal enhancement starts later ( Figure 3C)a saconsequence of diffusion limited reaction kinetics.F or highly infected samples,t he time to detection can be significantly faster, down to 12 min for 10 8 IU/mL. 80 fully characterized fresh (unfrozen) EDTAblood samples from the PHE MRL were then tested by origami-LAMP in an operator-blind experiment.
Our method showed high specificity and good sensitivity for identifying Plasmodium in blood samples (Table 1), when compared against the benchmark PCR. [3,24] Only for P. falciparum was sensitivity below 80 %, as anumber of samples for this species were of low parasite density.
All samples were also tested with ac ommercially-available LAMP kit for malaria (Eiken Chemical Company Ltd. (Japan)), which covers two of our three targets (Plasmodium pan and P. falciparum;T able 2). Thes pecificity for Plas-  The broken arrows indicate folding direction. Panels numbered as in Figure 1. A) Ah ole from the center of the third Panel has aglass-fiber disc onto which the sample is dispensed;Numbers in the last panel indicate the different reagents placed onto the four different spots for amplification of different species. 1. Internal control (IC);2.Plasmodium pan;3 .P. falciparum;4 .P. vivax.B )The second/third Panel are folded and clamped to form structure 1(S1);C)The fifth Panel is folded onto the back of the fourth frame to form structure 2(S2); D) S1 is folded onto the first Panel before adding lysis/washing buffer for DNA extraction and purification;E -F) S1 is folded onto S2 for elution and elution buffer added. modium pan and Pf alciparum between our method and the LAMP kit, which deploys readout in acommercial turbidim-eter, were above 98 %, while sensitivity was lower at 88 and 69 %r espectively ( Table 2).
Compared to the commercial LAMP kit, our platform has two advantages:o rigami LAMP is able to differentiate between P. falciparum and P. vivax infection, and correctly identify non-vivax/ non-falciparum, while also providing pangenus diagnosis as as ingle test. This has implications in guiding case management since these two species are associated with different treatments. [25,26] P. falciparum is more likely to progress to as evere illness than Pvivax, while P. vivax requires treatment of the dormant form of the parasite.T he two species also present different drug resistance profiles. [26] Thei nclusion of an internal control is an important quality control element for ruling out false negatives due to test failure.
Our method is more sensitive than microscopy and showed ac lose coincidence with both the gold standard benchmark (PCR) and ac ommercial LAMP assay (> 90 %, except for the PCR P. pan assay,88%-detailed calculations in Table S1). Theo rigami LAMP test only failed to detect seven weakly positive samples.T hese included PCR-positive individuals with negative blood films,a nd patients that had already commenced antimalarial treatment. Importantly,i f we exclude these challenging samples from the test panel, the sensitivity increases to 100 %f or P. falciparum and 95 %f or P. pan,w hile coincidence with other methods reach above 93 %f or all assays.
We also studied the analytical sensitivity of the technique using serially diluted cultured samples in whole blood, demonstrating detection down to 5parasites/ml( at hreshold below that of routine microscopy and below the clinical threshold at which symptoms of malaria occur, 500 parasites/ mL [27] -see Methods and Figures S2-3 in the Supporting Information). Clinically,t his allows any user to consider that anegative test in afebrile patient will be indicative of the fact that malaria infection is not the cause of the fever, although the presence of al ow density parasitaemia of Plasmodium spp.c annot be ruled out. Of particular importance is that the high specificity of our method, and thus low false positive ratio,g ives the user confidence that ap ositive result truly indicates the presence of malaria.
To analyze samples retrospectively,w es tored 4b lood samples under four different conditions (frozen, as thick and thin smears fixed in acetone,and dried). Table S2 shows that there is no difference between using frozen and fresh samples ( Figure S4 shows the details of the images obtained), enabling retrospective analysis when testing in the field is not possible 2. 10 7 IU/mL (green right-triangle);3.10 6 IU/mL (cyan down-triangle); 4. 10 5 IU/mL (magenta circle). 5. Negative control (black square:n o target DNA). As the concentration increases, the amplification is initiated earlier,e videnced by the exponential increase in the fluorescence. C) Threshold time (defined as the time corresponding to 50 % of the maximum fluorescence intensity,T t) as af unction of target concentration. This Figure of merit is analogous to the cycle threshold (Ct) of real-time PCR. [23] Data is the average of three repeats and error bars represent the standard deviation. The data was fitted with linear regression( R 2 = 0.98).  or in large cross-sectional studies. [28] As an additional feature of paper-based devices for disease diagnostics,wenoted that samples can be readily disposed of by incineration. [29]