A modular paper-and-plastic device for tuberculosis nucleic acid amplification testing in limited-resource settings

We present a prototype for conducting rapid, inexpensive and point-of-care-compatible nucleic acid amplification tests (NAATs) for tuberculosis (TB). The fluorescent isothermal paper-and-plastic NAAT (FLIPP-NAAT) uses paper-based loop mediated isothermal amplification (LAMP) for DNA detection. The cost of materials required to build a 12-test-zone device is $0.88 and the cost of reagents per reaction is $0.43. An inexpensive imaging platform enables filter-free fluorescence detection of amplified DNA using a cell-phone camera. FLIPP-NAAT can be operated by an untrained user and only requires a regular laboratory incubator as ancillary equipment. All reagents can be dry-stored in the device, facilitating storage and transportation without cold chains. The device design is modular and the assay demonstrated high specificity to Mycobacterium tuberculosis (Mtb), analytical sensitivity of the order of 10 copies of Mtb gDNA, and tolerance to complex samples. The clinical sensitivity and specificity of sputum-based FLIPP NAAT tests were 100% (zero false negatives) and 68.75% (five false positives), respectively (N = 30), using Xpert MTB/RIF assay as the reference standard. FLIPP-NAAT has the potential to provide affordable and accessible molecular diagnostics for TB in low- and middle-income countries, when used in conjunction with an appropriate sample preparation technique. Although demonstrated for the detection of TB, FLIPP-NAAT is a platform technology for amplification of any nucleic acid sequence.


Supplementary Information
Supplementary  Fluorescence imaging was also performed using optical filters to check if filters helped in improving the imaging sensitivity. The mean intensities for each of the copy numbers was less than half of the intensities observed for imaging without the filters. While imaging without filters, it was seen that the mean intensities were co-related with the starting copy numbers with statistically significant differences. But for imaging through optical filters, the mean intensities for 100 and 10 copies were not statistically different. Though mean intensities for reactions with 1000 starting copies continued to be statistically higher than 100 copies and same was true for 10 starting copies in comparison with the negative control. Apart from non-specific interaction with single-stranded DNA, the dye also seemed to have a component of green color even in its unbound state, which led to a prominent green signal in the negative controls when imaged through the filter.
Supplementary Figure S4. The protocol for testing the twenty clinical samples in FLIPP-NAAT is described in the methods section of the main manuscript. Images were captured using a cell phone camera and intensity analysis was done in the green channel using ImageJ. Supplementary Figure S7 shows AatII. All the negative controls did not show any amplification. Consistent with the FLIPP-NAAT results, sample 17_I in device H did not show any amplified products on the gel, while sample 17_I from device F showed very faint bands. This was also corroborated by the smear microscopy and GeneXpert results, both of which had classified sample 17 as weakly infected.
Since one of the duplicates for sample 17 showed consistent amplification, sample 17 was considered as positive for FLIPP-NAAT test results.
Even though sample 9_I and 9_II from the device in panel H appear slightly green to the eye, the intensity from the test zones was not found to be higher than the threshold during the intensity analysis (Supplementary table S4). The gel results for sample 9_I and 9_II from device in panel H helped in resolving the ambiguity and confirming that sample 9 is a true positive. The resulting suspension was vortexed and kept at 95˚C for 30 minutes to lyse the cells. The turbid solution obtained was centrifuged at 8000 rpm for 5 minutes and 1 µl of the supernatant was directly added to the LAMP mix.

Supplementary Notes
Supplementary Note S1. Why LAMP?
While several isothermal nucleic acid amplification techniques exist, we chose LAMP for FLIPP-NAAT because of its multiple distinctive characteristics, e.g. i) enhanced specificity because of the need of at least 4 primers, ii) an operating temperature range of 60-65˚C that is far from ambient, reducing baseline enzyme activity during storage, and iii) it's non-proprietary nature that enables acquiring individual reaction components from different vendors, providing flexibility for cost reduction. There are, however, certain challenges associated with LAMP. Primer dimer formation in LAMP leads to ladder-like patterns on gels reminiscent of target-specific amplification. Target-specific and non-specific amplification can be distinguished only by conducting enzyme digestion of products. The larger challenge with LAMP, however, is that because the number of amplicons produced is high, the method is prone to carryover amplicon contamination via aerosols. Post-amplification analysis was always conducted in a separate laboratory from the one in which LAMP reactions were set-up to avoid carryover contamination.

Supplementary Note S2. Comparison of FLIPP-NAAAT with existing similar designs
The device design by Seok et al. 1 and Ahn et al. 2 consisted of an assembly of polysulphone membrane, wax patterned polyethersulphone membrane, and glass fiber; physically stacked one above another and covered with ELISA sealing tape. The assembly was placed in a petri dish containing moist toilet tissues to maintain humidity and the petri dish was sealed using an aluminum tape. Nucleic acid amplification was demonstrated using LAMP and recombinase polymerase amplification (RPA). Fluorescence measurements were recorded at 10-min intervals by taking the set-up out of the oven and it was placed back at 63 °C in the oven immediately after measurement. The device from Trinh and Lee 3 consisted of a CNC milled pattern on a middle polycarbonate layer sandwiched between two sealing films. The chip was centrifuged after sample addition to distribute the sample to the respective reaction zones. Both the studies used fluorescence-based detection with imaging done in a high-end Bio-Rad Molecular Imager Gel ChemiDoc. Both the designs had multiple user steps, required ancillary equipment, and fluorescence imaging was done in an expensive set-up. FLIPP-NAAT, on the other hand, can contain all the components required for NAATs within the device and the user experience is very simple and minimal. One of the most desirable features of FLIPP NAAT is that the material cost of making a 12-reaction-zone device is only $0.88 and the cost of reagents required per reaction zone is $0.43 (see Supplementary Table S3 and S4 online). Modular design is a powerful feature of FLIPP-NAAT and using the current definition of a module, devices may be designed to contain 4N reaction zones (where N = 1, 2, 3…), without any major modification in fabrication methods.
The device design is a simple layer-by-layer assembly of paper pads, plastics, and adhesives, and therefore it is compatible with mass manufacturing using injection molded plastic components.
The very low-cost imaging box designed for cell phone-based filter-free fluorescence imaging is also an important development for enabling point-of-care NAATs.

Supplementary Note S3. Optimizing FLIPP-NAAT design
Several design challenges were overcome to develop the final FLIPP-NAAT prototype. Because the paper reaction zones in FLIPP-NAAT are in close proximity to each other, cross contamination was a challenge. The device dimensions were modified to provide more area of contact for the PSA to ensure it did not detach during heating at 63˚C. A vice was used to pressurize the device after fabrication to ensure that all layers were tightly secured. In the current design, there exists a 1-mm gap between modules, where there is discontinuity in both the top black acrylic covers and the bottom layer of PSA. The sudden change in surface properties from PSA to acrylic in these gaps helps in avoiding spread of leaking fluid, if any. Different types of materials were iterated to determine the most suitable material for each layer to optimize fluorescence detection. Transparent acrylic bottom allowed imaging from the bottom. Black acrylic covers, PSA, and the transparency cover gave minimum background fluorescence. The entirely plastic assembly of the device (i.e. no