Sensitive extraction-free SARS-CoV-2 RNA virus detection using a novel RNA preparation method

Current conventional detection of SARS-CoV-2 involves collection of a patient sample with a nasopharyngeal swab, storage of the swab during transport in a viral transport medium, extraction of RNA, and quantitative reverse transcription PCR (RT-qPCR). We developed a simplified and novel preparation method using a Chelex resin that obviates RNA extraction during viral testing. Direct detection RT-qPCR and digital-droplet PCR was compared to the current conventional method with RNA extraction for simulated samples and patient specimens. The heat-treatment in the presence of Chelex markedly improved detection sensitivity as compared to heat alone, and lack of RNA extraction shortens the overall diagnostic workflow. Furthermore, the initial sample heating step inactivates SARS-CoV-2 infectivity, thus improving workflow safety. This fast RNA preparation and detection method is versatile for a variety of samples, safe for testing personnel, and suitable for standard clinical collection and testing on high throughput platforms.

6 increased viral detection for calcium/magnesium free media, but not the calcium/magnesium media. Only 1 ~12% of viral RNA was detected in the HBSS containing Ca 2+ /Mg 2+ and 2% heat-inactivated FBS after 2 heating, in contrast, heating in the presence of Chelex allowed the detection of 84% of virions.

3
In summary, LowTE pH 8.0 and TED10 with Chelex produced the highest amounts of viral RNA 4 detected as compared to no heat or heating conditions among the buffers tested.

12
Inactivated SARS-CoV-2 and 293FT cells were mixed and added to these buffers above along with 13 lowTE and heated with or without Chelex ( Figure S2A). We diluted the samples in water before RT-qPCR

16
Of the chemicals tested, RNAlater and RLT appeared to be incompatible with the RNA-extraction-17 free method, as there was either no amplification or the Ct values were much higher than expected for all 18 four targets ( Figure S2A) after 20-fold dilution. Urea, DMSO, TE or MEM α showed minimum RT-qPCR 19 inhibition after dilutions. Chelex appears critical to achieve better detection of viral RNA and cellular RNA 20 in DMSO, 1xTE, lowTE, or MEM α, with the average Ct cycle difference of N1 & N2 between Chelex and 21 heat alone as 2.2, 1.8, 1.4, and 13, respectively. This represents improvement of sensitivity by Chelex of 22 4.5, 3.6, 2.7, and >1000 fold for samples prepared in DMSO, 1xTE, lowTE and MEM α, respectively. In  This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

6
We then determined the maximum concentrations of tested chemicals that were tolerated in a 7 RT-qPCR reaction ( Figure S2B). Samples were heat-treated with Chelex and serial dilutions of 2-fold 8 were used for RT-qPCR. The highest chemical concentrations that did not interfere with RT-qPCR if using 9 5 µl of undiluted sample in a 20 µl reaction were Urea 0.5 M, DMSO 50%, EDTA 0.5 mM ( Figure S2B).

10
The N1 & N2 Ct values for the undiluted sample in lowTE was the lowest, lower than RNA-extraction 11 using the same amount of virions, likely reflecting RNA loss during RNA extraction.

12
Together these results indicate that, by obviating the RNA extraction step, the presence of Chelex 13 in sample buffer increases RNA molecules available for RT-qPCR, as we observed in a variety of buffers 14 simulating nasopharyngeal and saliva collection conditions. Collecting swabs in lowTE appear to provide 15 the highest sensitivity under synthetic conditions.

16
Limit of detection 17 We attempted to further refine the buffer for this RNA-extraction free method by adjusting DMSO 18 concentration and combining TE with DMSO. The RT-qPCR data showed that lowTE with heat and

19
Chelex showed the lowest Ct values for N1 and N2, and combination of TE with DMSO did not improve 20 the Ct ( Figure S3A). Among the saliva samples, the undiluted saliva condition had the lowest Ct for N1 21 and N2 ( Figure S3A), with Ct values ~0.5 cycles above the lowTE sample. Heated saliva samples with or 22 without Chelex lowered Ct values by more than 2 cycles compared to non-heated samples. We further 23 found that increasing Chelex levels in DMSO increased detectable viral and cellular RNA ( Figure S3B).

24
To improve the limit of detection for RT-qPCR, we optimized the reaction conditions by (i) This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

9
We then used RT-ddPCR to determine whether we could detect a lower virion copy number.

10
Because we occasionally observed 1 or 2 positive droplets in the no-template control reactions, we 11 applied 1.8 copies/µl, or 50% higher than the maximum N1/N2 mean of negative controls ( Figure 1B and 12 Figure 2C), for the mean of N1 and N2 as the threshold for being positive for the SARS-CoV-2 N1/N2 RT-13 ddPCR assay. RT-ddPCR confirmed the LoD for conventional RNA extraction method to 2,000 virions per 14 swab and the lowTE-Heat-Chelex method to 200 virions per swab ( Figure 2C and Table 1). The RNA 15 extraction method only detected 30-50% of RNA molecules at the viral loads tested (Table 1).

16
In summary, the lower limit of detection of the Chelex method is 1 copy/µl in NP or saliva under 17 optimized buffer conditions and it avoids virion loss as observed in RNA extraction protocols.

18
Patient samples 19 We then validated our Chelex RNA preparation method using primary patient samples. NP swabs were 20 collected in M4 (N=14, S01 to S14, Figure 3A) or PBS (N=2, S15 & S16). These samples were tested in 21 the NIH Clinical Center diagnostic laboratory using conventional CDC RNA extraction and RT-qPCR 22 method (easyMAG-CL method), then frozen. Three of these samples, S01 to S03, had viral titer above 23 200 genome copies/µl. Twelve samples had viral titer less than 20 genome copies/µl, including eight 24 considered indeterminate because only one of N1 or N2 targets was positive. One sample, S14, was 25 negative.

26
for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted February 1, 2021. and NEB Luna RT-qPCR were comparable when using a common set of purified patient RNA samples 6 with different viral titer (data not shown). The mean difference of N1 Ct between the two methods was 7 2.7, excluding four samples (S10, S11, S12, and S16) that did not show a Ct value in the Chelex-Luna 8 assay.

9
Of the twelve samples with less than 20 genome copies/µl, eight (or 67%) showed as positive in 10 the N1 Chelex-Luna assay, including two samples (S13 and S15) that showed lower Ct values and 11 another (S08) with the similar Ct value. The N2 Ct values were higher than N1 and many were higher 12 than 38, thus were not informative for the low titer samples in the Chelex-Luna assay. Two NP swab 13 samples stored in the CDC-suggested VTM (HBSS with Calcium & Magnesium, S17 and S18) were

17
Next, we compared the RNA extraction and Chelex methods using two primary saliva samples 18 (Saliva01 & 02) and 20 positive saliva samples diluted in negative saliva samples side-by-side (Saliva03 19 to 22, Figure 3B). The mean Ct differences between the Chelex and RNA extraction methods for N1 and 20 N2 were 1.6 and 2.6 respectively. Among the six samples with less than 10 genome copies/µl as 21 determined by the RNA extraction method, five showed as positive and one (Saliva20) was indeterminate 22 in the Chelex assay ( Figure 3B). Thus, the Chelex method demonstrated similar sensitivity as the RNA 23 extraction method for both primary NP swab and saliva samples.

24
We then tested the optimized Chelex RNA isolation method prospectively by collecting This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
samples were heated and used for NEB Luna RT-qPCR, and the samples in VTM were subjected to RNA 1 extraction followed by RT-qPCR performed at the clinical laboratory. Because 1/6 th of buffer in the 2 collection tube was used in the Chelex method and RNA extraction concentrated sample by 4-fold, the Ct 3 values in the Chelex method are expected to be 0.6 lower than the RNA extraction method. Sample P1 4 was found to be negative using both methods, and five of the six NP samples and four of the four saliva 5 samples had lower N1 and N2 Ct values in the Chelex method as compared to the RNA extraction 6 method ( Figure 3C). The NP sample N1 Ct differences for patients P2 to P7 between these two methods 7 were -1.9, -2.1, -4.2, 0.8, -3.2, and -6.4, and their N2 Ct differences were -0.3, -5.5, -6.1, -0.7, -3.2, and -8 9.7. The saliva samples' N1 Ct differences for patients P4 to P7 between these two methods were -1.1, -9 2.1, -2.5, and -1.1, and their N2 Ct differences were -5.6, -3.9, -3.7, and -7.2. Thus, the Chelex method 10 may offer better sensitivities using the procedure here. In addition, the Chelex method allowed sample 11 processing without a Biosafety Cabinet hood as the samples were inactivated before tube opening.

12
Sample stability at room temperature 13 We then determined whether viral and cellular RNA were stable over time. Samples were stored at room 14 temperature then heat-treated in the presence of Chelex and assayed at different timepoints as indicated 15 ( Figure 4). Viral RNAs in lowTE or TED10 were relatively stable at room temperature and ~80% of N1 or 16 N2 RNAs were detected on day 5 ( Figure 4A). Viral RNAs were less stable in MEM α medium as > 80% 17 were degraded after 3 days ( Figure 4A). The cellular RNA was less stable than viral RNA in the similar 18 conditions ( Figure 4A). We then determined the RNA stability at room temperature after heat-treatment 19 with Chelex ( Figure 4B). Heat-treatment stabilized both viral and cellular RNA, and >80% viral RNAs 20 and >60% cellular RNA were detected on day 5 ( Figure 4B).

21
A similar stability experiment was performed for virions prepared in saliva samples. The viral

22
RNAs were stable in saliva before heat-treatment, as a higher amount of viral RNAs were detected after 23 storage at room temperature, possibly because interfering agents in saliva degrade during storage 24 ( Figure 5A). However, the viral RNA stability decreased markedly after heat treatment ( Figure 5B). This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted February 1, 2021.
qPCR and RT-ddPCR detection compared to the conventional method using the simulated samples with 3 the ATCC SARS-CoV-2 virions. When using stored patient samples, RNA extraction provided higher 4 sensitivity ( Figure 3A, B). This is likely due to the enrichment arising from the lower volume of RNA eluate 5 as compared to the input sample volume. The lower sensitivity of the Chelex method seen in Figure 3A 6 could be also due to RNA loss during the freeze-thaw cycle. When the samples were tested side-by-side 7 ( Figure 3B), the sensitivities of the Chelex method and RNA extraction method are comparable. When 8 patient samples were collected in VTM (S17 & S18 in Figure 3A), the N1 and N2 Ct values increased 2 9 and 4, respectively, suggesting that the Chelex method on VTM-collected samples may result in lower 10 sensitivity for some low viral load samples. Further, we found that N1 and N2 responded differently 11 toward components in the VTM as the N1 Ct increased while N2 Ct decreased after 1:2 dilution (S19 & 12 S20 in Figure 3A). Our results in Figure 3C showed that collecting the swab directly in Chelex-TED 13 (TE+DMSO) buffers followed by heating provided the best sensitivity for SARS-CoV-2 detection. The

14
higher LoD associated with conventional RNA extraction method was due to dilution in the VTM and loss 15 of RNA during RNA extraction as shown in Table 1. In addition, viral RNA may be degraded faster when 16 storing in the calcium/magnesium containing VTM, as suggested by data presented in Figure 4A. As 17 such, we recommend primary sample collection for this RNA isolation method in either TED or lowTE 18 buffer, which has similar or better detection performance as RNA extraction methods for primary samples 19 collected in VTM.

20
We also demonstrate that the Chelex method allows for highly-reproducible detection of 1 21 genome copy/µl of the SARS-CoV-2 virions, a 6-15-fold improvement in detection sensitivity. 15, 20 One 22 explanation for this improvement is that the Chelex resin chelates the divalent ions necessary for Rnase 23 activities, and the resin may be able to non-specifically remove inhibitors to reverse-transcription and/or 24 PCR. In addition, un-processed samples may lead to more RNA degradation during sample collection 25 and storage as compared to being stored in the presence of Chelex. We further identified conditions that 26 allow sensitive detection of SARS-CoV-2 in saliva. The potential benefits of saliva testing include lower 27 for use under a CC0 license.
This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted February 1, 2021. ; 12 cost (no swab), reduced variability, and improved patient acceptance over traditional NP swab. 14, 21 Thus, 1 the Chelex method may provide a more sensitive point-of-care method for RNA diagnostics by reducing 2 false negative results.

3
There will be a continued need for SARS-CoV-2 detection during and after the current vaccination 4 period. The sheer volume of preprint and publications in a short period of time illustrates the urgent need 5 and hope to increase testing capacity employing the RNA-extraction free approach. The utility of this RNA 6 isolation method for both NP and saliva samples would increase the number of people tested in the same 7 timeframe as the current method. In addition to improved sensitivity, this method offers a number of 8 additional advantages compared to the current gold standard clinical laboratory testing, including 9 improved cost, reduced sample processing time and complexity, and enhanced workflow safety. The cost 10 of CDC-recommended VTM collection tube is ~$1.70 per tube, and RNA extraction may cost > $6 per 11 sample. We estimate the total cost of Chelex, lowTE, DMSO, and the collection tube is <$1 per sample.

12
Thus, the Chelex method will save cost and reduce supply chain burden by eliminating the need for RNA 13 extraction and VTM. The novel RNA preparation method, amenable for high-throughput processing, is 14 expected to shorten diagnostic testing time by omitting the RNA extraction step and omitting the chemical 15 disinfection of patient samples. This method utilizes a heat-inactivation step that minimizes viral RNA 16 loss, obviates the need for a biological safety cabinet, and eliminates exposure of laboratory personnel to 17 live virus. Therefore, we fully expect that this method will facilitate broader availability and testing capacity 18 for not only, COVID-19, but also for other infectious pathogens. Because of the observed stability of 19 SARS-CoV-2 RNA in collected samples at room temperature, this method should also improve access to 20 COVID-19 testing in resource scarce regions of the world, by improving RNA stability, reducing cost of 21 collection kits and diagnostic reagents, and eliminating the requirement of refrigeration, biosafety cabinet,

22
and storage of RNA extraction kits.

23
Because the Chelex method also allowed cellular RNA detection, we expect that the method This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. One limitation of the current study was the low-level contamination observed in the RT-ddPCR to the background contamination, we used 1.8 copies/µl as the low limit of detection for the SARS-CoV-2 10 N1/N2 RT-ddPCR assay. According to the FDA Emergency Use Authorization (EUA) by the manufacturer

11
Bio-Rad, the low limit of detection in the 1-well RT-ddPCR assay system is 2 positive droplets and no 12 positive droplets observed in no-template controls. Thus the low limit of detection in a RT-ddPCR assay 13 could reach 0.1 copy/µl if performing in a clean room.
14 In summary, we robustly demonstrate improvements in COVID-19 viral testing workflow using  This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted February 1, 2021. advanced posteriorly to the nasopharynx as previously described. 26 Swabs were then inserted into VTM.

11
Subjects were asked to expectorate into a sterile 50 mL conical vial every 30 seconds for 3-5 minutes, or 12 until ~3 mL of saliva was collected. A FLOQswab ® was inserted into the saliva and immediately This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted February 1, 2021. ; were considered positive for SARS-CoV-2 when both N1 and N2 targets were detected with Ct count <40.

5
The positive signal for N1 or N2 alone was defined as an indeterminate result.

17
Chelex was first prepared in H2O, lowTE, or TED99 at 50% (50 grams/100 ml total volume, or 18 500 milligrams Chelex to 550 µl H2O). The 50% Chelex was then added in 1/10 volume to samples to 19 obtain 5% Chelex in a PCR strip with a wide-bore tip. The samples were vortexed briefly then heated in a 20 PCR cycler for 5 min at 98 °C, followed by spinning at 1,000 to 2,000xg for 2 mins in a swing-rotor. The 21 supernatant was then used for RT-qPCR or -ddPCR.

22
For simulated samples involving the ATCC virions, the RNeasy mini kit (Qiagen) was used to 23 extract RNA from virion and cell mixture of less than 10 µl that contained the expected amounts of virions.

24
The RNeasy Protect Saliva Mini Kit (Qiagen) was used for RNA extraction from ATCC virion-simulated  This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted February 1, 2021. ; DNA sequence located in the exon 1 of the RPP30 gene, thus are expected to amplify both cDNA and 6 genomic DNA contents. An additional RPP30 primer specific for RPP30 cDNA was designed to span the 7 exon 1 and exon 2, RPP30cR GCAACAACTGAATAGCCAagGT, where the lower case "ag" denotes the 8 exon junction. RPP30cR was used for RT-qPCR or -ddPCR together with RPP30F and RPP30Hex. The ΔRn. The N1 and N2 Ct of less than 38 were considered positive in the NEB Luna RT-qPCR assay.

20
The 1-Step RT-ddPCR Advanced Kit for Probes (#186-4021, Bio-Rad) was used for RT-ddPCR 21 using the QX200 Droplet Digital PCR System (Bio-Rad). The cycling condition for RT-ddPCR was: 50°C 22 for 60 min, 95°C for 10 min, and 40 cycles of 94°C for 10 sec and 55°C for 60 sec, followed by 98°C for 23 10 min, 4°C for 30 min then hold at 4°C. If DMSO was not present in the sample, DMSO was added to 24 2.5% in the RT-qPCR or -ddPCR reaction, unless specified otherwise. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.    This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted February 1, 2021. CoV-2) diagnosis by RT-PCR to increase capacity for national testing programmes during a 7 pandemic. bioRxiv (2020). This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted February 1, 2021. ; combination of the heat treatment approach and rt-Real-time PCR testing. Emerg Microbes 23 Infect 9, 1393-1396 (2020). 24 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  #: ND, not detectable. The low limit of quantification in the 1-well RT-ddPCR assay is 4 copies/µl of a 4 molecule. The means of N1 and N2 copy numbers for the control samples without virions added were 5 less than 1.2 copies/µl, thus, we applied 1.8 copies/µl as the low limit of detection for the SARS-CoV-2 6 N1/N2 RT-ddPCR assay. 7 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted February 1, 2021. ;   This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Figure Legends
The copyright holder for this preprint this version posted February 1, 2021.  (A) Patient NP swab samples were heated in the presence of 5% Chelex (S01 to S16, and S19, S20) or 9 10% Chelex (S17 & S18). S19 & S20 are 1:2 dilution of S17 & S18 in LowTE, respectively. (B) 50 µl of This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted February 1, 2021. ; volumes of 50% Chelex prepared in H2O or TED99 on the time points indicated and assayed. The RT-1 ddPCR reactions were carried out in one well for N1 and cRPP30 and another well for N2 and RPP30. level detected. Copies/µl refers to concentration in the samples used for RT-ddPCR. The error bars 5 represent Poisson 95% confidence intervals. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted February 1, 2021. ; Figure S4. Optimization of the NEB Luna RT-qPCR assay. Extracted RNA samples were 3 serial diluted and assayed either using 2.5 µl sample in a 10 reaction volume or 5 µl in a 20 µl reaction, This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  Copies/µl c R P P 3 0 R P P 3 0 c R P P 3 0 R P P 3 0 c R P P 3 0 R P P 3 0 c R P P 3 0 R P P 3 0 This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.   This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. S09 S10 S11 S12 S13 S14 S15 S16 S17 S18 S19 S20 This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted February 1, 2021. ; https://doi.org/10.1101/2021.01.29.21250790 doi: medRxiv preprint  This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  Figure S1 for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted February 1, 2021. ; https://doi.org/10.1101/2021.01.29.21250790 doi: medRxiv preprint This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted February 1, 2021. ; https://doi.org/10.1101/2021.01.29.21250790 doi: medRxiv preprint This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.