147. Defining the Optimal Serial Testing Interval and Features for Identifying Patients with Early SARS-CoV-2 Infection

Abstract Background Serial testing for SARS-CoV-2 is necessary to prevent spread from patients early in infection. Testing intervals are largely derived from viral kinetic studies performed early in the COVID-19 pandemic. Laboratory and epidemiologic data accrued over the past year present an opportunity to use empiric models to define optimal serial testing intervals and features predictive of early infection. Methods Retrospective analysis of 15,314 inpatients within the Mass General Brigham healthcare system who had two tests within a 36-hour period between May 1 2020 and May 29 2021. Early infection was defined as having a negative test followed by a positive test. Patients with prior positive tests were excluded. The primary outcome was the proportion of patients in early infection over the total number tested serially, stratified by 4-hour testing intervals from the timestamp of the first test. Multivariate modeling was used to identify features predictive of early infection. Covariates included demographics, body site, PCR assay, location, community incidence, percent positivity, and median / skew of Ct value distributions. Results Of 19,971 test pairs, 193 (0.97%) were characterized as a negative followed by a positive within 36 hours. Bivariate analysis showed a close association between negative to positive test pairs during the first surge in spring 2020 that was not present during the winter surge. Negative to positive test pairs were most common in the 12 to 16 hour time interval (51/193, 26%, Figure 1). After controlling for covariates, the Roche cobas assay was more likely to identify patients with a negative to positive test pair relative to the Cepheid Xpert, Hologic Panther Fusion and Roche Liat assays. A second specimen from the lower respiratory tract was more likely to lead to a positive relative to other body sites. Community incidence and Ct value distributions were not predictive and there were no differences between nasal and nasopharyngeal swabs. All 4-hour time intervals from 16 to 36 hours were significant for predicting a negative to positive test pair (Table 1). Figure 1. Distribution of negative to positive test pairs by 4 hour time intervals Table 1. Multivariate regression predicting a negative to positive test pair Conclusion The likelihood of detecting early infection is dependent on PCR platform and body site of sampling. A range of time intervals between 16 to 36 hours after the initial test were likely to identify positive cases. Disclosures Sanjat Kanjilal, MD, MPH, GlaskoSmithKline (Advisor or Review Panel member)


Background.
Although studies show most COVID-19 survivors have post-infection immunity against SARS-CoV-2 that could prevent re-infection, there is still a need to identify the breadth of antibody (Ab) responses associated with clinical phenotypes. We characterized Ab profiles at the estimated peak of Ab diversity among adults with recovered SARS-CoV-2 infections and determined their relationships with clinical factors.
Methods. From April-June 2020, 41 health system employees with PCRconfirmed symptomatic COVID-19 infection enrolled 8-10 weeks after symptom onset. Symptom questionnaires including baseline demographics, COVID-19 symptoms, disease severity, and disease duration were collected and plasma samples were assayed using a custom Luminex Multiplex platform ( Figure 1) to measure the antibody response against 20 COVID-19 related antigens ( Figure 2). Differences in Ab profile titers among different groups were tested using nonparametric t test and Benjamini-Hochberg adjustment for multiplicity. Associations were considered significant at FDR< 0.05.  Results. Mean age was 48 years (range 27-68), with 51% female, 37% White, 32% Black, 29% Asian, and 17% LatinX. Ab profiles (Figure 3) showed 100% cross-reactivity with related alpha and beta coronavirus, and 95% with SARS-CoV-1. 78% had Abs against SARS-CoV-2 nucleocapsid protein (NCP). However, 29% of patients had no immune response against the four spike protein epitopes. These participants also reported fewer symptoms, including no cases of anosmia/ageusia, suggesting mild illness. Anosmia/ageusia, fever, and cough associated significantly with higher Ab titers ( Figure 4).

Conclusion.
Broad immune responses to various SARS-CoV-2 and related antigens were found among a heterogeneous patient population. However, less than 3 months after symptom onset, protective Ab responses to SARS-CoV-2 spike proteins were not detected in nearly one-third of recovered patients, primarily with mild infection. Intact sense of smell and taste demonstrated the greatest association with loss of seroprotective SARS-CoV-2 Ab responses, which may be clinically useful to predict post-infection immunity. Next steps include comparing the magnitude of Ab responses following full series completion with mRNA vaccination among this cohort.

Defining the Optimal Serial Testing Interval and Features for Identifying Patients with Early SARS-CoV-2 Infection
Sanjat Kanjilal, MD, MPH 1 ; 1 Harvard Medical School and Harvard Pilgrim Healthcare Institute, Jamaica Plain, MA

Session: O-30. Research in COVID-19 Diagnostics
Background. Serial testing for SARS-CoV-2 is necessary to prevent spread from patients early in infection. Testing intervals are largely derived from viral kinetic studies performed early in the COVID-19 pandemic. Laboratory and epidemiologic data accrued over the past year present an opportunity to use empiric models to define optimal serial testing intervals and features predictive of early infection.

Methods.
Retrospective analysis of 15,314 inpatients within the Mass General Brigham healthcare system who had two tests within a 36-hour period between May 1 2020 and May 29 2021. Early infection was defined as having a negative test followed by a positive test. Patients with prior positive tests were excluded. The primary outcome was the proportion of patients in early infection over the total number tested serially, stratified by 4-hour testing intervals from the timestamp of the first test. Multivariate modeling was used to identify features predictive of early infection. Covariates included demographics, body site, PCR assay, location, community incidence, percent positivity, and median / skew of Ct value distributions.
Results. Of 19,971 test pairs, 193 (0.97%) were characterized as a negative followed by a positive within 36 hours. Bivariate analysis showed a close association between negative to positive test pairs during the first surge in spring 2020 that was not present during the winter surge. Negative to positive test pairs were most common in the 12 to 16 hour time interval (51/193, 26%, Figure 1). After controlling for covariates, the Roche cobas assay was more likely to identify patients with a negative to positive test pair relative to the Cepheid Xpert, Hologic Panther Fusion and Roche Liat assays. A second specimen from the lower respiratory tract was more likely to lead to a positive relative to other body sites. Community incidence and Ct value distributions were not predictive and there were no differences between nasal and nasopharyngeal swabs. All 4-hour time intervals from 16 to 36 hours were significant for predicting a negative to positive test pair (Table 1).  Table 1. Multivariate regression predicting a negative to positive test pair Conclusion. The likelihood of detecting early infection is dependent on PCR platform and body site of sampling. A range of time intervals between 16 to 36 hours after the initial test were likely to identify positive cases.
Disclosures. Sanjat Kanjilal, MD, MPH, GlaskoSmithKline (Advisor or Review Panel member) Background. Detection and surveillance of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants is of great public health importance. Broadly accessible and inexpensive assays are needed to enhance variant surveillance and detection globally. We developed and validated a single-reaction multiplex real-time RT-PCR (the Spike SNP assay) to detect specific mutations associated with variants of concern (VOC).

Single-amplicon, Multiplex Real-time RT-PCR with Tiled Probes to Detect SARS-CoV-2 spike Mutations Associated with Variants of Concern
Methods. A single primer pair was designed to amplify a 348 bp region of spike. Probes were initially designed with locked nucleic acids (LNAs) to increase probe melting temperature, shorten probe length, and specifically detect 417K, E484K, and N501Y ( Figure). The assay was optimized and evaluated using characterized variant sample pools. Clinical evaluation was performed on a convenience set of residual nasopharyngeal swabs, and variant calls were confirmed by SARS-CoV-2 genomic sequencing in a subset of samples. Following the initial evaluation, unmodified probes (without LNAs) were designed to detect L452R, L452Q, and E484Q. Representative results of variant detection a single Spike SNP run are shown for mutations in the codons for 4177K (A) and mutations that encode 484K (B) and 501Y (C). Curves show dilutions of the following variants: blue, BEI 52286 (wild type); pink B.1.1.7; purple, B1.525; and green, P.1. Variant pools were used for B.1.17, B.1.525, and P.1 strains. Curves are displayed for a given dilution in each channel and result interpretation is shown (D).
Results. The lower limit of 95% detection was 2.46 to 2.48 log 10 GE/mL for the three targets (~1-2 GE/reaction). Among 253 nasopharyngeal swabs with detectable SARS-CoV-2 RNA, the Spike SNP assay was positive in 238 (94.1%), including all samples with Ct values < 30 (220/220) for the N2 target and 18/33 samples with N2 Ct values ≥ 30. Results were confirmed by SARS-CoV-2 genomic sequencing in 50/50 samples (100%). Subsequent addition of the 452R probe did not affect performance for the original targets, and probes for 452Q and 484Q performed similarly to LNAmodified probes.
Conclusion. The Spike SNP assay provides fast, inexpensive and sensitive detection of specific mutations associated with SARS-CoV-2 VOCs, and the assay can be quickly modified to detect new mutations in the receptor binding domain. Similar analytical performance of LNA-modified and unmodified probes presents options for future assay customization that balance the shorter probe length (LNAs) and increased accessibility (unmodified). The Spike SNP assay, if implemented across laboratories offering SARS-CoV-2 testing, could greatly increase capacity for variant detection and surveillance globally.