Comparison of Severe Acute Respiratory Syndrome Coronavirus 2 Screening Using Reverse Transcriptase–Quantitative Polymerase Chain Reaction or CRISPR-Based Assays in Asymptomatic College Students

Key Points Question Are CRISPR-based methods a reliable and accessible option to capture severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) outbreaks in a college population? Findings In this cohort study, 1808 asymptomatic college students were screened for SARS-CoV-2 status using reverse transcriptase–quantitative polymerase chain reaction (RT-qPCR) and CRISPR-based assays. Nine samples positive for SARS-CoV-2 were detected by RT-qPCR, and 8 were confirmed by CRISPR-based assay and clinical laboratory diagnostic testing, uncovering a change in viral prevalence that coincided with the relaxation of lockdown measures and the rise of coronavirus disease 2019 cases in the community. Meaning CRISPR-based methods appear to offer reliable SARS-CoV-2 testing for virus screening and allow capture of the leading edge of an outbreak.


Introduction
The coronavirus disease 2019 (COVID- 19) pandemic has claimed hundreds of thousands of lives and has disrupted life in countless communities. To control this pandemic, communities worldwide closed businesses, prohibited large social gatherings, and adopted nonpharmacological intervention measures. [1][2][3] Initial restrictions were successful in several countries where COVID-19 cases, hospitalizations, and deaths declined. 2,3 However, as communities relaxed social distancing and restrictions, COVID-19 cases returned, often with exponential growth. Several metrics, including percentage of positive test results, hospitalizations, and death rates, have been used to gain insights into epidemic trends in specific populations. Prevalence among asymptomatic persons is an important but more elusive metric, primarily because of test scarcity and prioritization of symptomatic patients or contacts with confirmed cases. Nevertheless, understanding both asymptomatic prevalence and the effect of nonpharmacological intervention measures on infection rates has tremendous potential to inform vital public health decisions.
A fundamental aspect of pandemic control is careful planning for the reopening of college campuses. Although COVID-19 testing has focused on individuals with increased risk of infection and mortality, an increasing disease burden has emerged in those aged 19 to 30 years, many of whom attend colleges and universities. 4 Every year since 2017, more than 15 million students attend colleges in the US. 5 Many students reside in dormitories and off-campus housing, frequently in crowded conditions, sharing restrooms, kitchens, and common areas. 6 These living conditions are associated with high morbidities of diseases such as meningococcal meningitis, influenza, mumps, and measles. [7][8][9][10] Respiratory pathogens, such as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), are easily transmitted among individuals living in college dormitories and during social contact by exposure to live virus in aerosol droplets. [11][12][13] Further complicating SARS-CoV-2 transmission in university settings is the well-documented infectivity of asymptomatic persons, many of whom are likely to be presymptomatic with high viral loads. [14][15][16][17][18][19][20] Those without symptoms are likely to be responsible for as many as 44% of new infections. 21 Recent examples of colleges reopening and promptly closing or implementing drastic quarantine measures for their students after the detection of COVID-19 outbreaks illustrate the challenges of safely bringing academic activities back to campus during a pandemic. The upsurge of cases within college populations also presents a risk beyond campus walls because infections can spill over to neighboring communities. 22 The early identification of infected individuals through expanded and frequent surveillance testing is essential to curb disease spread. However, before undertaking such large-scale surveillance testing, the prevalence of asymptomatic infection must be ascertained to inform decisions regarding the utility of expanded testing in a university population. 23 Several methods are currently available for COVID-19 diagnosis, with reverse transcriptasequantitative polymerase chain reaction (RT-qPCR) assays being most commonly used. 24 The high demand for COVID-19 testing has overwhelmed supply chains, limiting the availability of critical reagents and specialized equipment necessary for RT-qPCR. CRISPR-based assays provide a robust and sensitive alternative for the detection of SARS-CoV-2 genomes. These assays use common and widely available reagents and are adaptable to minimal instrumentation and infrastructure. Although CRISPR-based tests have been validated for the detection of COVID-19 in clinical samples, no information is available about the performance of these assays for SARS-CoV-2 screening in asymptomatic individuals. [25][26][27][28] To understand viral prevalence in the university community and to assess the potential of a CRISPR-based test to screen for SARS-CoV-2 in asymptomatic persons, we enrolled healthy volunteers from the University of California, Santa Barbara (UCSB) in a virus screening study. We obtained self-collected oropharyngeal swab samples, processed for SARS-CoV-2 testing using 2 methods: CREST (Cas13-based, rugged, equitable, scalable testing), a newly developed CRISPRbased assay, 25 and the Centers for Disease Control and Prevention (CDC)-recommended RT-qPCR assay, 29 which we used as a point-of-reference test. We compared the results obtained from May 28 to June 11, 2020, approximately 2 months into a statewide stay-at-home mandate, and June 23 to July 2, 2020, approximately 3 weeks after easing local restrictions for isolation in the community. Our results revealed no COVID-19 cases in the study population during the May-June collection period.
Using the same methods, we demonstrated a substantial shift in prevalence approximately 1 month later, which coincided with changes in community restrictions and public interactions. Notably, the CRISPR-based assay performed as well as the CDC-recommended RT-qPCR assay. Our study substantiates the utility of self-collected oropharyngeal swabs and CRISPR-based testing as valuable alternatives for large-scale surveillance sampling of SARS-CoV-2 in asymptomatic individuals.

Study Population
The population of UCSB includes 26 134 students (82.2%) and 5668 staff and faculty (17.8%). Among the students, 38.2% live in university housing, and 33.6% in the nearby community of Isla Vista (23 096 residents; 1.866 square miles, 12 377 people/square mile). This cohort study was open to all symptom-free individuals 18 years or older who were affiliated with UCSB (students, faculty, staff, and direct relatives). Individuals who exhibited a fever (38.0°C), cough, or shortness of breath in the 2 weeks before or on the day of sample collection were excluded from the study. Only 5 participants were excluded owing to presenting symptoms at the time of collection and were referred to local health care resources. Preanalytical and postanalytical protocols were reviewed and approved by the institutional review board of Santa Barbara Cottage Hospital. All participants provided written informed consent. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.

Sample Collection
Health care professionals at UCSB collected written, informed consent and demographic data (age, address, telephone, sex, and UCSB affiliation) at the sampling locale. Samples were assigned a numeric code for deidentification purposes. Samples were acquired as self-collected oropharyngeal swabs stored in phosphate-buffered saline, with surveillance by a health care professional (H.S., B.M., and L.P.). Samples were inactivated at 56°C for 30 minutes, and RNA was extracted using 1 of

SARS-CoV-2 Detection by RT-qPCR
We performed RT-qPCR following the procedures in the emergency use authorization granted by the US Food and Drug Administration. 29 Viral RNA was reverse transcribed and amplified using the 1-step complementary DNA master mix kit (TaqPath; Thermo Fisher Scientific 501148245) following the manufacturer's recommendations. Reactions were prepared as previously described. 25 Briefly, a 15-μL master mix reaction was prepared using the established CDC protocol, 29 and 5 μL of RNA were added into the reaction with each of the target-specific RT-qPCR primers and probes. For no-template controls, 5 μL of nuclease-free water were used. Positive control reactions used 10 6 copies of in vitro transcribed RNA encoding the SARS-CoV-2 nucleocapsid sites N1 and N2. Reactions were run in a qPCR instrument (CFX96 Touch; Bio-Rad Laboratories, Inc) using the following thermal profile: 25°C for 2 minutes; 50°C for 15 minutes; 45 cycles of 95°C for 5 seconds followed by 55°C for 30 seconds and plate read; and hold at 4°C. Data were analyzed using the manufacturer's software (CFX Maestro; Bio-Rad Laboratories, Inc) with a single threshold for determination of quantification cycle (Cq) value. We prepared standard curves of in vitro transcribed RNAs, ranging from 10 6 to 10 0 copies/μL, to determine detection limits. One-way analysis of variance with a post hoc Dunnett test was used to determine the Cq value significance from no-template controls using Prism software, version 8 (GraphPad Software Inc). The limit of detection for N1 and N2 is 10 2 copies/μL (Cq, 32.59 and 34.405, respectively); for ribonuclease P (RNaseP), 10 3 copies/μL (Cq, 34.328). Samples were considered positive if the signal for both N1 and N2 was above the limit of detection. Samples were processed in-house with a turnaround time from 12 to 30 hours from the moment of collection.

CREST Assay
CREST reactions were performed as described. 25

Estimation of Viral Load
To estimate the viral load in the asymptomatic or presymptomatic participants confirmed as having positive test results, the genome equivalents per microliter were calculated based on the Cq values for N1 and N2 from the RT-qPCR assay. The calculation used linear regression on a standard curve ranging from 10 0 to 10 6 gene copies/μL.

Statistical Analysis
Correlations between N1 and N2 and between our CRISPR-based assay and the RT-qPCR assay were calculated using the Pearson correlation coefficient, assuming data are from a bivariate normal distribution, using the R function cor.test() (R Program for Statistical Computing). Percentage of positive rates were fit using a logistic growth model where current P = KP [P + (K -P)e −rt ], with K = 100%, P = .03, r fit by minimizing the error found to be r = 0.101, and rt indicating rate of maximum population growth.

Results
A total of 1808 healthy volunteers were screened for SARS-CoV-2. All participants were asymptomatic for COVID-19 at the time of sample collection. Samples were collected from May 28 to their place of residence. The study population's mean (SD) age was 28.4 (11.7) and 26.6 (10.5) years for cohorts 1 and 2, respectively, with a minimum age of 18 years and a maximum of 75 years (Table).
SARS-CoV-2 genomes were detected using CREST, the CRISPR-based method recently developed by Rauch et al, 25 and the RT-qPCR test recommended by the CDC was used as the point of reference 29 (eFigure 1 in the Supplement). Both methods detected 2 sites in the nucleocapsid gene, N1 and N2, and 1 site in the host RNaseP transcript, which ensured consistency in the analyses.
All samples collected in cohort 1 (n = 732) had negative results by both tests (Figure 1, right side). In contrast, 8 positive samples were detected by the CRISPR-based assay and 9 by RT-qPCR in cohort 2 (n = 1076) (Figure 1). There was a good correlation in detecting the nucleocapsid gene using the N1 and N2 probes (CRISPR-based assay, Pearson correlation coefficient r = 0.872) ( Figure 1A) and primers (RT-qPCR assay, Pearson correlation coefficient r = 0.566) ( Figure 1B). The participants with positive results had a mean (SD) age of 21.7 (3.3) years, and all self-identified as UCSB students (Table).  (Table). None of the participants reported fever as a symptom.
The estimated viral loads for the positive samples ranged from 286 to 510 000 copies/μL (Figure 3 and eTable 2 in the Supplement). These viral load levels were not significantly different from those detected in a control set of deidentified residual nasopharyngeal swab samples obtained from symptomatic patients in the local community provided to us by collaborators at the Santa Barbara County Public Health Department (eTable 2 in the Supplement and Figure 3). Notably, the quality of the self-collected specimens using oropharyngeal swabs was not significantly different from those collected using nasopharyngeal swabs as measured by the detection of RNaseP transcripts (Figure 3).
The prevalence of SARS-CoV-2 in the study population in cohort 1 was 0, whereas that of cohort 2 was 0.8%, with a daily incidence ranging from 0 to 1.7% (Figure 4 and eTable 3 in the Supplement).
The prevalence dynamics in the study population reflect the increase in COVID-19 cases diagnosed in the UCSB neighboring communities of Goleta and Isla Vista, where most of our participants reside ( Figure 4). The increase in the number of infections detected in this study-and those in Santa Barbara County-coincided with the reopening of personal care and recreation venues (restaurants and bars) in Santa Barbara County (Figure 4). Laboratory Improvement Amendments-certified laboratory.

Discussion
As colleges and universities through the US struggle to recover from the academic, social, and economic effects of months of remote learning, a pressing trial remains: how to reopen campuses safely? A primary challenge for university communities is the potential for covert infections promoted by social and academic gatherings, which are unavoidable in the context of a vibrant university campus. Recent evidence indicates that asymptomatic and presymptomatic individuals can unknowingly transmit the virus and fuel covert outbreaks. 19,30,31 The early detection of asymptomatic infections-particularly those with high SARS-CoV-2 loads, such as those detected in our analyses that may underlie superspreader events-is vital for mitigating viral transmission and containing outbreaks. This information is also essential to guide university directives to make decisions regarding campus openings across the country and ensure superior education continuity.
Epidemiological models support this notion and suggest that universal and frequent SARS-CoV-2 testing is necessary for efficient disease containment. 23 However, the economic effects of providing reliable and regular testing for thousands of students, faculty, and staff may prohibit larger campuses from closely monitoring their communities.   25 This CRISPR-based assay is as efficient at detecting SARS-CoV-2 infections in asymptomatic participants as the CDC-recommended RT-qPCR, which is considered the criterion standard testing method. It also has the added benefit of enabling an easy-to-interpret and dependable binary readout: fluorescence vs no fluorescence. The CRISPR-based assay showed perfect concordance with positive cases diagnosed in a Clinical Laboratory Improvement Amendments-certified laboratory (Pacific Diagnostics Laboratory), further corroborating its robustness. Because CREST was designed to be a low-cost and accessible method, it offers a muchsought alternative for communities where resources are limited and where access to testing is      difficult. This CRISPR-based method is scalable, enabling high-throughput testing, and it uses laboratory-generated or off-the-shelf commercially available reagents, thus eliminating the restriction of limiting supply chains. For these reasons, we surmise that CREST can offer a solution for places where access to professional laboratories is restrictive and instances in which a high volume of repetitive sampling is necessary, including the university setting.

JAMA Network Open | Infectious Diseases
One of our most significant observations is the difference in SARS-CoV-2 prevalence between the 2 cohorts we analyzed. We did not detect any infections in the 732 people tested in late May and early June.

Limitations
This study has some limitations. The reported analytical sensitivity of oropharyngeal swab samples for SARS-CoV-2 detection is lower than that of nasopharyngeal swab samples, particularly when samples are collected 8 to 15 days after onset of illness. [32][33][34][35] Despite this limitation, we selected selfcollected oropharyngeal swabs as the sampling method for SARS-CoV-2 screening. Our goals were to minimize the effect of this study on the limited availability of nasopharyngeal swabs for clinical purposes and reduce the viral exposure of health care personnel who supervised sample collection.
The low number of samples with positive results detected herein limits the interpretation of the data.
The results presented reflect the low prevalence of SARS-CoV-2 in Santa Barbara County at the time of this study.