key: cord-0739628-eir065mk
authors: Kidd, Stephen P.; Burns, Daniel; Armson, Bryony; Beggs, Andrew D.; Howson, Emma L.A.; Williams, Anthony; Snell, Gemma; Wise, Emma L.; Goring, Alice; Vincent-Mistiaen, Zoe; Grippon, Seden; Sawyer, Jason; Cassar, Claire; Cross, David; Lewis, Thomas; Reid, Scott M.; Rivers, Samantha; James, Joe; Skinner, Paul; Banyard, Ashley; Davies, Kerrie; Ptasinska, Anetta; Whalley, Celina; Ferguson, Jack; Bryer, Claire; Poxon, Charlie; Bosworth, Andrew; Kidd, Michael; Richter, Alex; Burton, Jane; Love, Hannah; Fouch, Sarah; Tillyer, Claire; Sowood, Amy; Patrick, Helen; Moore, Nathan; Andreou, Michael; Morant, Nick; Houghton, Rebecca; Parker, Joe; Slater-Jefferies, Joanne; Brown, Ian; Gretton, Cosima; Deans, Zandra; Porter, Deborah; Cortes, Nicholas J.; Douglas, Angela; Hill, Sue L.; Godfrey, Keith M.; Fowler, Veronica L.
title: RT-LAMP has high accuracy for detecting SARS-CoV-2 in saliva and naso/oropharyngeal swabs from asymptomatic and symptomatic individuals
date: 2022-02-02
journal: J Mol Diagn
DOI: 10.1016/j.jmoldx.2021.12.007
sha: 9de942934eed2ef7a8f5690f1ab9f6209822b694
doc_id: 739628
cord_uid: eir065mk

Previous studies have described RT-LAMP methodology for the rapid detection of SARS-CoV-2 in nasopharyngeal (NP) and oropharyngeal (OP) swab and saliva samples. This study describes the validation of an improved sample preparation method for extraction free RT-LAMP and defines the clinical performance of four different RT-LAMP assay formats for detection of SARS-CoV-2 within a multisite clinical evaluation. Direct RT-LAMP was performed on 559 swabs and 86,760 saliva samples and RNA RT-LAMP on extracted RNA from 12,619 swabs and 12,521 saliva from asymptomatic and symptomatic individuals across healthcare and community settings. For Direct RT-LAMP, overall diagnostic sensitivity (DSe) of 70.35% (95% CI 63.48-76.60%) on swabs and 84.62% (79.50-88.88%) on saliva was observed, with diagnostic specificity (DSp) of 100% (98.98-100.00%) on swabs and 100% (99.72-100.00%) on saliva when compared to RT-qPCR; analysing samples with RT-qPCR ORF1ab C(T) values of <25 and <33, DSe of 100% (96.34-100%) and 77.78% (70.99-83.62%) for swabs were observed, and 99.01% (94.61-99.97%) and 87.61% (82.69-91.54%) for saliva, respectively. For RNA RT-LAMP, overall DSe and DSp were 96.06% (92.88-98.12%) and 99.99% (99.95-100%) for swabs, and 80.65% (73.54-86.54%) and 99.99% (99.95-100%) for saliva, respectively. These findings demonstrate that RT-LAMP is applicable to a variety of use-cases, including frequent, interval-based testing of saliva with Direct RT-LAMP from asymptomatic individuals that may otherwise be missed using symptomatic testing alone.

Rapid diagnostic testing to identify and isolate symptomatic and asymptomatic individuals potentially 81 transmitting infectious viral pathogens is an essential requirement of any pandemic response. The novel 82 betacoronavirus, SARS-CoV-2, initially identified after an outbreak of viral pneumonia in Wuhan, China in 83

December 2019 1 , has rapidly spread throughout the world, causing over 223 million confirmed cases and 84 over 4.6 million deaths (https://coronavirus.jhu.edu/ -September 10, 2021). 85

Conventional diagnostics for SARS-CoV-2 consist of RNA enrichment followed by reverse-transcriptase 86 quantitative real-time PCR (RT-qPCR) against one or more viral gene targets 2 . However, this methodology 87 requires sample inactivation, RNA extraction and RT-qPCR thermal cycling, meaning that the time from 88 sample-to-result can often be several hours, and requires centralised equipment and personnel trained 89 in Good Laboratory Practice to perform testing. 90

The utility of reverse-transcriptase loop mediated isothermal amplification (RT-LAMP) for the detection 91 of SARS-CoV-2 both from extracted RNA (RNA RT-LAMP) and directly from NP/OP swabs (Direct RT-92 LAMP) 3 has previously been shown. RT-LAMP utilises a rapid and stable DNA polymerase that amplifies 93 target nucleic acids at a constant temperature. This removes the requirement for conventional thermal 94 cycling allowing RT-LAMP reactions to be performed in shorter reaction times using less sophisticated 95 platforms. 96

In a study of 196 clinical samples 3 , testing of RNA extracted from NP/OP swabs collected into viral 97 transport media (VTM) using RNA RT-LAMP demonstrated a diagnostic sensitivity (DSe) of 97% and a 98 diagnostic specificity (DSp) of 99% in comparison to RT-qPCR of the ORF1ab region of SARS-CoV-2. For 99 Direct RT-LAMP on crude swab samples, the DSe and DSp were 67% and 97%, respectively. When a cycle 100 threshold (CT) cut-off for RT-qPCR of < 25 was considered, reflecting the increased likelihood of detecting 101 viral RNA from active viral replication, the DSe of Direct RT-LAMP increased to 100% 3 . 102

However, the collection of a swab sample is invasive and during the time of the pandemic there have 103 been considerable shortages in swab supplies. Exploring the use of alternative sample types that are both 104 easy to collect and more comfortable from a sampling perspective 4,5,6 is desirable particularly when 105 repeat sampling is performed 7, 8, 9 . Saliva presents an ideal bio-fluid that fulfils both these objectives and 106 previous studies have shown that SARS-CoV-2 is readily detectable in such a sample type 10, 11, [12] [13] [14] [15] [16] [17] [18] . To 107 improve the diagnostic sensitivity of previously described saliva Direct RT-LAMP 3 , optimisation of saliva 108 preparation for the detection of SARS-CoV-2 was undertaken utilising a cohort of 3100 saliva samples 109 from an asymptomatic population 19 of healthcare workers; where saliva was diluted 1:1 in Mucolyse, 110 followed by a 1 in 10 dilution in 10% (w/v) Chelex 100 Resin ending with a 98°C heat step prior to Despite the benefits of this optimisation, the protocol added additional steps and reagents which 113 increased chance for user error and made the automation of the process more challenging. This study 114 therefore aimed to investigate a simpler process using a novel reagent, RapiLyze (OptiGene Ltd, 115 Camberley, UK), which is a sample dilution buffer, followed by a two-minute heat-step. This novel sample 116 preparation method was evaluated in combination with Direct RT-LAMP using samples collected from 117 symptomatic National Health Service (NHS) primers using the AgPath-ID PCR kit (Thermo Fisher) according to manufacturer's instructions for use (IFU) 188 on an Aria qPCR Cycler (Agilent, Cheadle, UK) and results analysed using the Agilent AriaMX 1.5 software, 189 using 5 µl of extracted RNA per reaction. RNA extracted using the MagMAXCORE Nucleic acid purification 190 kit were analysed using this assay. with molecular-grade nuclease free water (Ambion) to a final volume of 15 μl. 5 μl of AVE buffer extract 197 was used at a template and added to the 15 μl mastermix. Cycling conditions were 55°C for 10 min, 198 followed by 94°C for 3 min and 45 cycles of 95°C for 15 s and 58°C for 30 s. 199

Single step RT-qPCR against the N1 and N2 gene targets of SARS-CoV-2 was carried out using integrated 201 design technologies kit (IDT; Catalogue number: 10006606) according to manufacturer's instructions for 202 use (IFU) on either a LC480 II or ABI 7500 FAST instrument. RNA extracted on Qiagen QIAsymphony and 203 the Roche FLOW system were analysed using this RT-qPCR assay. 204

RT-LAMP assays were performed using OptiGene Ltd. COVID-19_RT-LAMP kits, as described previously 3 , 206 with the following modifications. The COVID-19_RNA RT-LAMP KIT-500 kit was used for RNA and the COVID-19_Direct PLUS RT-LAMP KIT-500 was used for Direct RT-LAMP directly on 208 oropharyngeal/nasopharyngeal swabs or saliva samples. The COVID-19_Direct PLUS RT-LAMP KIT-500 kit 209 also includes a sample preparation buffer, RapiLyze. For RNA RT-LAMP 5 μl of extracted RNA was added 210 to the reaction. For the Direct PLUS RT-LAMP, 50 µl sample (swab VTM or neat saliva) was added to 50 µl 211

RapiLyze, vortexed and placed in a dry heat block pre-heated to 98 o C for 2 mins. 5 μl of the treated sample 212 was added to each reaction. 213

The anneal temperature (Ta) that confirmed a positive result for Direct RT-LAMP was modified to 81. where , represent the expected logit sensitivity and specificity, 2 , 2 represent the between-253 study variance in the logit sensitivity and specificity, and , represents the covariance between the 254 logit sensitivity and specificity. For Direct RT-LAMP, we fit a univariate normal distribution for the logit-255 transformed sensitivity only, due to the absence of false positives across all sites. 256

In addition, the sensitivity as a function of viral load was assessed for RNA RT-LAMP and Direct RT-LAMP 257 on both swab and saliva samples. This was performed through the conversion of each sample CT value to 258 viral load in gene copies/ml for all sample sets. As the relationship between CT value and viral load varied 259 according to the RT-qPCR method used; a dilution series was utilised for each method to standardise 260 these values for two of the four aforementioned RT-qPCR methods (CerTest VIASURE SARS-CoV-2 real 261 time PCR kit, and Corman et al RT-qPCR assay E gene), which was used for testing 100% of the swab 262 samples, 90% of the saliva samples used for Direct RT-LAMP, and 83% of the saliva samples used for RNA 263 RT-LAMP. The logarithm of the viral load was then fitted to the CT values for both methods using linear 264 regression followed by converting the CT values to viral load based on which method had been used to 265 evaluate the samples. For the remaining samples (n= 56) that utilised one of the other two RT-qPCR 266 methods for which viral load was not standardised against a CT value, the conversion derived from the CerTest VIASURE SARS-CoV-2 real time PCR kit dilution series was applied, the assumption that the N gene 268 CT values are the most similar 21-23 . 269 For the CerTest VIASURE SARS-CoV-2 real time PCR kit, the following relationship between log viral load 270 and CT value was applied: 271 where represents the viral load in copies/ml. 275 Viral load was grouped according to the following categories (in copies/ml): <10 3 , 10 3 -10 4 , 10 4 -10 5 , 10 5 -276 10 6 , 10 6 -10 7 and >10 7 then the diagnostic sensitivity was calculated according to viral load group with 277 associated Clopper-Pearson 95% confidence intervals. 278

The site meta-analysis was produced using R 3.5.3. Confusion matrices, sensitivity, specificity, sensitivity 279 as a function of viral load calculations, and the production of scatter graphs showing the relationship 280 between RT-LAMP results and CT were performed using Python 3.8.6. 281

Results 283

Heat inactivation experiments demonstrated that SARS-CoV-2 was completely inactivated by heating at 285 60°C (20 min plus) or ≥70°C (after 2, 5 or 10 min) (Supplemental Table S1 ). Importantly optimised RapiLyze 286 Sample Lysis Buffer did not inactivate live virus on its own without a heat step. Further, inactivation at 287 56°C was not 100% effective at shorter incubation times, and additionally showed a loss in sensitivity 288 following a 4 x 2-fold dilution (Supplemental Table S2 , P07102) at 10 and 30 minutes. Following 289 optimisation of heat inactivation of live virus, pre-treatment of samples was assessed to determine any 290 impact of pre-treatment on assay sensitivity. Interestingly, a pre-treatment 70°C for 5 mins carried out 291 on spiked samples prior to the proposed direct RT-LAMP assay had no effect on subsequent LAMP or PCR 292 results. It recommended that even if a pre-treatment is effective in inactivating the virus that downstream 293 processes are carried out in UV hoods or with effective air-flow management to prevent cross Table 2 ). 195 samples were tested in duplicate and 364 tested as single replicates. Seven of 195 samples 332 tested in duplicate were positive by Direct RT-LAMP in only one replicate (CT 27.51, 27.95, 28.15, 28.15, 333 28.87, 28.92, and 28.95) analysis (CT 20.27, 21.28, 22.01, 24.42, 25.85, 27.35, 28.52, and 30.37 The relationship between CT value and Tp was explored with the results shown in Figure 1 . Whilst there 364 is a weak linear relationship between CT value and Tp across all methods and sample types, a stronger 365 linear relationship was observed in swab samples with 2 = 0.431 for RNA RT-LAMP and 2 = 0.462 for 366

Direct RT-LAMP. There was a notably weaker linear relationship in saliva samples 2 = 0.201 for RNA RT-367 LAMP and 2 = 0.204 for Direct RT-LAMP. For RNA RT-LAMP, there was a notable increase in Tp variance, 368 2 , after CT = 20 across both sample types. On saliva samples, 2 = 0.81 for CT <20, and 2 = 20.41 369 for CT >20; on swabs samples, 2 = 1.96 and CT <20, and 2 = 15.72 for CT >20. Given the relationship 370 between CT value and viral load, this indicates that Tp is not a reliable indicator for viral load beyond the 371 CT = 20 threshold. 372

Although not a large sample size; a negative result via Direct RT-LAMP indicates that the presence of 374 culturable virus is less probable and for samples with a CT >25 (RDRP/ORF1ab target) recoverable virus is 375 less likely (Table 5 ). The sensitivity of 1 PFU/ml of the viral culture assay is presented in Supplemental 376 Table S3 . No CPE was observed in the flasks inoculated with 0.1 or 0.01 PFU after the two passes. AVL 377 samples were taken from the flasks at the beginning and end of each passage and the CT values of the 378 extracted nucleic acids shown in Supplemental Table S3 . 379

In the time course experiment SARS-CoV-2 RNA was detected from day 5 (at the onset of symptoms) up 381 to day 12 post suspected initial exposure using Direct RT-LAMP and up to day 13 by RNA RT-LAMP, 382 encompassing the full six days where symptoms were recorded, Supplemental Table S4 . 383

The sensitivity of the RNA and Direct RT-LAMP assays across viral load groups is shown in Figure 2 . For 385 RNA RT-LAMP, samples which were positive by RT-qPCR containing >10 5 copies/ml were consistently 386 identified as positive with no samples returning a negative result. Below this copy number, sensitivity is 387 reduced for both saliva and NP/OP swab samples, reaching ~60% in NP/OP swab samples exclusively with 388 viral loads <10 3 copies/ml, and an approximately linear drop in sensitivity from 100% to 0% between viral 389 loads of 10 5 and 10 3 copies/ml respectively in saliva samples. For Direct RT-LAMP, all but one saliva sample 390 were detected above 10 6 copies/ml. On swab samples, sensitivity is reduced on samples containing below 391 <10 5 copies/ml, dropping from 85% at viral loads of 10 5 -10 6 copies/ml, to 30% in the 10 4 -10 5 range. On 392 saliva samples, sensitivity is reduced in the 10 4 -10 5 range to a sensitivity of 80% but then reduces further 393 within the 10 3 -10 4 range, to 30%. 394

Site-level confusion matrices, sensitivity, and specificity per method and sample type are shown in Figures  396 3 and 4. For specificity, heterogeneity between sites was minimal for all combinations of method and 397 sample type, with the random effects model matching the overall aggregated sample calculation. For 398 sensitivity, heterogeneity was minimal between sites for RNA RT-LAMP. However, for Direct RT-LAMP, 399 sensitivity showed significant overall heterogeneity (bivariate model variance: 2 = 1.817 on saliva 400 samples; 2 = 0.228 on swab samples). Between-site variations in the viral load of the samples tested 401 contributed a minority of the heterogeneity, but sensitivity was consistently high in samples with higher 402 viral loads (i.e., >10 6 copies/ml, as shown in Figure 2 CoV-2 viral RNA could be detected from saliva for a similar duration post onset of clinical signs when 418 compared to combined NP/OP swabs 29-31 , highlighting saliva as a valuable tool for SARS-CoV-2 detection. 419

Direct detection negates the requirement for RNA extraction 32,33 , for which there has previously been 420 competition for reagents and often requires expensive extraction equipment including liquid handling 421 automation. This extraction-free method decreases turnaround time from sample collection to result. 422

The Direct RT-LAMP method is straight forward and rapid, allowing the test to be performed in a wide 423 range of settings, including near-patient hospital laboratories and pop-up or mobile laboratories. 424

However, previously evaluated extraction-free sample preparation methods using RT-LAMP from saliva 425 samples have demonstrated reduced sensitivity 3,19 , likely due to the inhibitory factors found within saliva 426 that may affect molecular tests such as 35 . The simple sample preparation method 427 evaluated in the study aimed to improve upon these methods by utilising the addition of a novel step of 98 o C for two minutes prior to addition to the RT-LAMP master mix renders SARS-CoV-2 inactive as 430 confirmed by infectivity analysis using live virus inactivation studies (Supplemental Tables S1 and S2) . 431

Downstream steps are then able to proceed outside of traditional containment level laboratory settings 432 broadening its clinical utility. 433

This study utilised high numbers of combined naso/oropharyngeal swabs (n= 559) and saliva samples (n 434 = 86,760) for the evaluation of this novel sample preparation method in combination with the Direct RT-435 LAMP assay. RNA RT-LAMP was also performed on >25,000 samples for comparison, providing updated 436 values for the performance of the assay reported previously 3, 19, 36 Diagnostic sensitivity for RNA RT-LAMP on swab and saliva samples was improved when compared to a 448 previous report utilising this method 3 , with values of >96% and >80%, respectively when considering all 449 CT values, and 100% for both sample types when considering CT <25 with these samples having a high 450 probability of containing replicating virus for over 24,000 samples tested. Direct RT-LAMP sensitivity on 451 swab samples was also improved from the previous method with 100% sensitivity for CT <25, 77.78% for 452 CT <33 and 70.35% for CT <45 across 559 samples used for this evaluation. In contrast, sensitivity for Direct 453 RT-LAMP on saliva was in general higher than that determined for swabs (CT <33 = 87.61%, CT <45 = 454 84.62%), apart from the group with CT values below <25, which had a reported sensitivity of 99.01%. 455

These results support previous reports which demonstrate comparable performance when comparing 456 paired swabs and saliva samples 41, 42 , and that one sample type is not superior to the other. Interestingly, 457 the diagnostic sensitivity for RNA and Direct RT-LAMP for saliva samples was almost equivalent (80.65% 458 and 84.62%, respectively) suggesting that RNA extraction may not even be required when performing 459 testing on saliva samples. Direct RT-LAMP also demonstrates a higher sensitivity than a wide variety of 460 lateral flow tests (LFTs) in the CT < 25, CT ≥ 25 and overall categories, with the overall sensitivity of Direct 461 RT-LAMP on saliva samples achieving a higher overall sensitivity than 94 out of 96 LFTs previously 462 evaluated 43 . We found that the correlation between PCR CT value and the Direct RT-LAMP Tp was weaker recognised time course differences between initial viral infection of the salivary glands and later infection 465 of the respiratory tract 26, 30 . 466

Previous studies have described the importance of identifying asymptomatic individuals, particularly 467 those with high viral loads 28,44-48 . The ability of the Direct RT-LAMP assay to reliably detect individuals 468 with medium-high viral loads in a simple to collect, non-invasive sampling process highlights the suitability 469 of this assay for both symptomatic and asymptomatic population screening. This is particularly important 470 in healthcare and care home staff where the use of asymptomatic COVID-19 screening would reduce the 471 risk of onward transmission of SARS-CoV-2, consequently maintaining NHS capacity and Social Care 472 capacity and more importantly, reducing the risk to vulnerable individuals present within those 473 environments 36 . 474

It is important to note that when designing surveillance strategies for asymptomatic infection testing as 475 an intervention to reduce transmission, frequency of testing and result turnaround time may be 476 considered more significant than diagnostic sensitivity 49 . 'Gold standard' tests with high sensitivity such 477 as RT-qPCR generally need to be performed in centralised testing facilities, often resulting in increased 478 reporting times, leading to a less effective control of viral transmission 49 . In contrast, point of care tests 479 such as Lateral flow tests (LFT) 50,43 or those requiring only a basic/mobile laboratory set-up such as Direct 480 RT-LAMP, which have the ability to produce rapid results, can be performed frequently e.g., daily or 481 multiple times per week. Consequently, the likelihood of sampling an individual when their viral load is 482 highest as seen in the early, often pre-symptomatic stages of infection increases, maximising the 483 probability of rapidly detecting infectious cases, allowing prompt isolation. In this use case sampling and 484 testing frequency using a rapid assay with suitable accuracy in detection of medium-high viral loads, but 485 not necessarily optimal sensitivity over the whole range including low to very low viral loads, is desirable 486 or necessary 49,51 . Frequent on-site testing of asymptomatic NHS healthcare workers using Direct RT-LAMP 487 has been successfully implemented in the pilot study described here; and continues to be utilised. Direct-488

RT-LAMP has also been used in a mass community based pilot in school and higher education settings 36 , 489 to identify those individuals who may have been missed when surveillance relies only on symptomatic 490 individuals coming forward for testing. With the use of mobile or pop-up laboratories, Direct RT-LAMP 491 could also be used for risk-based mass testing, for example, targeting specific geographical areas or 492 vulnerable groups. The potential also exists for lyophilisation of the Direct RT-LAMP reagents reported in 493 other studies 52,53 , which would minimise the necessity for trained personnel by reducing pipetting steps 494 and the requirement for a cold chain, allowing greater capacity of the assay in multi-use case scenarios 495 including point-of-care and in low-and middle-income countries (LMICs). 496 which when tested against a panel of respiratory pathogens causing indistinguishable clinical signs to 499 COVID-19, demonstrated a high level of analytical specificity (100% in this case) 3 and analytical sensitivity 500 of the Direct RT-LAMP, which is reported to detect 1000 cp/ml 3, 36, 41 . Additionally, the RNA and Direct RT-501 LAMP assays evaluated as part of this study have been shown to reliably detect the emerging variants of 502 concern (VOC) including the B.1.1.7 alpha variant, the 501Y.V2 beta variant, the P1 gamma variant and 503 the new rapidly spreading B.1.617.2 delta variant 54,55 (https://www.gov.uk/government/collections/new-504 sars-cov-2-variant (accessed June, 2021) . The emergence of further VOC could lead to a criticism of the 505 RT-LAMP assay due its reliance on a single target, ORF1ab, where mutations in the target region in a 506 sample could lead to false negatives. For RT-qPCR this has been observed during the current pandemic 56-507 58 where at least a dual target assay is recommended 59 . However, this is less likely to occur for the RT-508 LAMP assay used in this pilot evaluation. Firstly, due to the multiple sets of primer pairs utilised, three 509 pairs, with two pairs within the target region. This builds in redundancy to mutation not unlike a duplex 510 RT-qPCR. Secondly, the ORF1ab region is highly conserved and crucial for viral replication and fitness in 511 SARS-CoV-2. As a result, these regions are well maintained using a proofreading system via the nsp14 512 protein 60 resulting in a more stable genome compared to many other RNA viruses. 513

The authors highlight the importance of incorporating an inhibition control into the next iteration of the 514 RT-LAMP assays. Although the paired RT-LAMP and RT-qPCR data from this study show a good correlation 515 and any false negative results were likely due to the analytical sensitivity of the RT-LAMP assay, not 516 sample driven inhibition. To this end, a control primer set by OptiGene Ltd was evaluated (PS-0010), 517 targeting the human ribosomal protein LO gene. Preliminary analysis of the inhibition control primers 518 showed consistent detection across 279 saliva and 381 combined naso/oropharyngeal swab samples 519 using both RNA and Direct RT-LAMP (manuscript in preparation). Incorporation of this inhibition control 520 into the RT-LAMP assays would alleviate a potential limitation of the current assays and further support 521 quality assurance for use within a clinical diagnostic setting. One further limitation to LAMP assays is the 522 potential for contamination from assay product which can be significant. LAMP assays produce vast 523 amounts which can persist in the environment not only causing potential false positive results in 524 subsequent testing but also anomalous results in laboratory workers who are part of a SARS-CoV-2 testing 525 programme 61 . Therefore, it is crucial that appropriate waste streams are in place to mitigate this risk. 526

This study demonstrated high sensitivity and specificity for a novel sample preparation method used for 527 SARS-CoV-2 Direct RT-LAMP, particularly in samples from which the individual would likely be considered 528 infectious, highlighting the usefulness of saliva as a simple to collect, non-invasive sample type. The highly 529 sensitive RNA RT-LAMP assay provides a rapid alternative with a reliance on differing reagents and 530 equipment to RT-qPCR testing, thus providing additional diagnostic capacity and redundancy through 531 diversity. Direct RT-LAMP may complement existing surveillance tools for SARS-CoV-2 testing including such as frequent, interval-based testing of asymptomatic individuals that may be missed when reliance is 534 on symptomatic testing alone. However, care should be taken when considering frequency of testing, 535 messaging around the role and interpretation of asymptomatic rapid tests, integration of data storage 536 and access, and the challenges faced when scaling up surveillance to large populations. 537

The role out of a new testing strategy can often throw up interesting and unexpected experiences. These 538 collective experiences and lessons learnt from setting up an NHS asymptomatic staff testing programme 539 using Direct RT-LAMP will be shared in a future publication. 540 541

Rapid diagnostic testing at scale to identify and isolate symptomatic and asymptomatic individuals 543 potentially transmitting infectious SARS-CoV-2 is an essential part of the response to the COVID-19 544 pandemic. RT-LAMP on both extracted RNA and directly on crude samples potentially provides faster 545 turnaround times than reverse-transcriptase quantitative real-time PCR testing, with a higher sensitivity 546 and specificity than antigen lateral flow devices. Increasing evidence points to potential benefits of SARS-547

CoV-2 testing using saliva rather than nasopharyngeal/oropharyngeal swabs, therefore a multi-site 548 evaluation of an improved simple sample preparation method for Direct SARS-CoV-2 RT-LAMP was 549 undertaken. This study demonstrated that the RNA RT-LAMP assay has high sensitivity and specificity, 550 providing a rapid alternative to RT-qPCR testing with a reliance on differing reagents and equipment. The 551 simple SARS-CoV-2 Direct RT-LAMP preparation method also demonstrated high sensitivity and specificity 552 for detecting SARS-CoV-2 in saliva and naso/oropharyngeal swabs from asymptomatic and symptomatic 553 individuals, notably in saliva samples from which the individual would likely be considered infectious. The 554 findings highlight the usefulness of saliva as a simple to collect, non-invasive sample type, potentially 555 applicable for interval-based testing of asymptomatic individuals. 

Clinical characteristics of 140 561 patients infected with SARS-CoV-2 in Wuhan, China

Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT-567

A highly effective reverse-transcription loop-mediated isothermal amplification

LAMP) assay for the rapid detection of SARS-CoV-2 infection

Diagnosis of SARS-Cov-2 Infection by RT-PCR Using Specimens Other Than Naso-and 575

Self-collected unstimulated saliva, oral swab, and nasopharyngeal swab 578 specimens in the detection of SARS-CoV-2

Saliva is more sensitive than nasopharyngeal or nasal swabs for diagnosis of asymptomatic 582 and mild COVID-19 infection

Stability 586 of SARS-CoV-2 RNA in nonsupplemented saliva

Practical challenges to the clinical implementation of saliva for SARS-CoV-2 detection

SARS-CoV-2 viral load is associated with increased 10

Testing Saliva to Reveal the Submerged Cases of the 596 COVID-19 Iceberg

Performance of Saliva

Oropharyngeal Swabs, and Nasal Swabs for SARS-CoV-2 Molecular Detection: a Systematic 599 Review and Meta-analysis

Posterior Oropharyngeal Saliva for the 602

Detection of Severe Acute Respiratory Syndrome Coronavirus 2 ( SARS-CoV-2 )

Comparison of SARS-CoV-2 605 detection in nasopharyngeal swab and saliva

May viral load detected in saliva in the early stages of infection be a 607 prognostic indicator in COVID-19 patients?

Viral dynamics of SARS-CoV-2 in saliva from infected patients

Rapid Salivary Test suitable for a mass screening program to 614 detect SARS-CoV-2: A diagnostic accuracy study

Mass Screening of Asymptomatic Persons for Severe Acute 617

Respiratory Syndrome Coronavirus 2 Using Saliva

Are posterior oropharyngeal saliva specimens an acceptable alternative to 620 nasopharyngeal sampling for the monitoring of sars-cov-2 in primary-care settings? Viruses

Preliminary optimisation of a simplified sample preparation method to permit direct detection of 626

SARS-CoV-2 within saliva samples using reverse-transcription loop-mediated isothermal 627 amplification (RT-LAMP)

Bivariate analysis 629 of sensitivity and specificity produces informative summary measures in diagnostic reviews. J 21. Bhargava A. Dynamic aspects of ORF1ab and N RNA cycle threshold values among COVID-19 632 patients in China

SARS-CoV-2 genomic and subgenomic RNAs in diagnostic 634 samples are not an indicator of active replication

Clinical features and dynamics of viral load in 636 imported and non-imported patients with COVID-19

SARS-CoV-2 infection of the oral 638 cavity and saliva

Temporal profiles of viral load in posterior 642 oropharyngeal saliva samples and serum antibody responses during infection by SARS-CoV-2: an 643 observational cohort study

Saliva as a gold-standard sample for SARS-645

CoV-2 detection

viral-load curves in paired saliva and nasal swabs inform appropriate respiratory sampling site 648 and analytical test sensitivity required for earliest viral detection

Just 2% of SARS-CoV-2−positive individuals carry 90% of the virus circulating in 653 communities

Accuracy and stability of saliva as a sample for reverse 657 transcription PCR detection of SARS-CoV-2

Saliva or Nasopharyngeal Swab Specimens for Detection of SARS-CoV-2

SARS-CoV-2 detection by a 668 clinical diagnostic RT-LAMP assay

Head-to-head comparison of direct-input RT-PCR and RT-LAMP against RT-qPCR on 672 extracted RNA for rapid SARS-CoV-2 diagnostics

Inhibitory effect of salivary fluids on 674 PCR: Potency and removal

Department of Health and Social Care. Rapid evaluation of OptiGene RT-LAMP assay (direct and 678 RNA formats). Department of Health and Social Care. United Kingdom

Viral 680 RNA load as determined by cell culture as a management tool for discharge of SARS-CoV-2 681 patients from infectious disease wards

Predicting infectious severe 684 acute respiratory syndrome coronavirus 2 from diagnostic samples

Virological assessment of hospitalized patients with COVID-2019

Quantifying the relationship between SARS-CoV-2 viral load and infectiousness

Critical evaluation of the methodology used by

Concerning the Optigene Direct LAMP assay, and it`s use in at-risk groups and hospital 696 staff

Self-collected 698 anterior nasal and saliva specimens versus healthcare worker-collected nasopharyngeal swabs 699 for the molecular detection of SARS-CoV-2

Asymptomatic SARS Coronavirus 2 infection: 705 Invisible yet invincible

Transmission of COVID-19 in 282 clusters in Catalonia

Spain: A cohort study

Estimating the extent of 710 asymptomatic COVID-19 and its potential for community transmission: Systematic review and 711 meta-analysis

Occurrence and transmission potential of asymptomatic and presymptomatic SARS-CoV-2 714 infections: A living systematic review and meta-analysis

SARS-CoV-2 Transmission from People without COVID-19 Symptoms

Test sensitivity is secondary to frequency and turnaround time for COVID-19 screening

infectivity by viral load, S gene variants and demographic factors and the utility of lateral flow 724 devices to prevent transmission

Viral dynamics of acute SARS-CoV-2 infection and applications to diagnostic and public health 728 strategies

Molecular Assays to Rapidly Detect Foot-and-Mouth Disease Virus Directly from Clinical Samples 732 in Field Settings

The development of two field-734 ready reverse transcription loop-mediated isothermal amplification assays for the rapid 735 detection of Seneca Valley virus 1

737 rapid spread of a new severe acute respiratory syndrome-related coronavirus

lineage with multiple spike mutations in South Africa

Identification of a Polymorphism in the N Gene of SARS-CoV-2 That Adversely Impacts 745

Detection by Reverse Transcription-PCR

Mutation in the N Gene of SARS-CoV-2 May Affect the Detection of the Virus by Reverse 748

Transcription-Quantitative PCR

A Recurrent Mutation at Position 26340 of SARS-CoV-2 Is Associated with 751

Failure of the E Gene Quantitative Reverse Transcription-PCR Utilized in a Commercial Dual-752

Mutations on COVID-19 diagnostic targets

Low genetic diversity may be an 756

Achilles heel of SARS-CoV-2

Laboratory-Generated DNA Can Cause Anomalous Pathogen Diagnostic Test Results