Robotic RNA extraction for SARS-CoV-2 surveillance using saliva samples

Abstract

Saliva is an attractive specimen type for asymptomatic surveillance of COVID-19 in large populations due to its ease of collection and its demonstrated utility for detecting RNA from SARS-CoV-2. Multiple saliva-based viral detection protocols use a direct-to-RT-qPCR approach that eliminates nucleic acid extraction but can reduce viral RNA detection sensitivity. To improve test sensitivity while maintaining speed, we developed a robotic nucleic acid extraction method for detecting SARS-CoV-2 RNA in saliva samples with high throughput. Using this assay, the Free Asymptomatic Saliva Testing (IGI FAST) research study on the UC Berkeley campus conducted 11,971 tests on supervised self-collected saliva samples and identified rare positive specimens containing SARS-CoV-2 RNA during a time of low infection prevalence. In an attempt to increase testing capacity, we further adapted our robotic extraction assay to process pooled saliva samples. We also benchmarked our assay against nasopharyngeal swab specimens and found saliva methods require further optimization to match this gold standard. Finally, we designed and validated a RT-qPCR test suitable for saliva self-collection. These results establish a robotic extraction-based procedure for rapid PCR-based saliva testing that is suitable for samples from both symptomatic and asymptomatic individuals.

Fig 1. Overview of nucleic acid extraction from saliva specimens.

a, Overview of Innovative Genomics Institute’s (IGI’s) specimen processing pipeline for both swab and saliva samples. OP = oropharyngeal. OP-MT = oropharyngeal-mid turbinate. b, Cultured SARS-CoV-2 (1.58x106 TCID50/ml) was mixed 1:1 with OMNIgene solution present in OM-505 collection tubes to test incubation conditions that inactivate viral replication. Samples were either held at room temperature (RT) or incubated at 65°C for the indicated length of time before being applied to Vero-E6 cells. Cytopathic effect (CPE) was quantified at 3 and 7 days post treatment (dpt). c, 3:2 dilution of saliva samples with DNA/RNA Shield improves detection of spiked-in SARS-CoV-2 RNA or MS2 in four saliva donors. Figure made with Biorender.

Fig 2. Validation of the IGI’s assay for detecting SARS-CoV-2 RNA from saliva.

Saliva collected from four unique donors (negative for SARS-CoV-2 by swab) was used to generate a titration curve of ThermoFisher COVID-19 Positive Control SARS-CoV-2 RNA (a) or heat-inactivated virus (b) to determine the assay’s limit of detection (LoD). c, SARS-CoV-2 RNA or heat-inactivated virus was spiked in at 2x or 5x the LoD into 20 unique saliva samples previously determined to be negative for SARS-CoV-2. d, SARS-CoV-2 RNA was spiked into 20 unique saliva samples previously determined to be negative for SARS-CoV-2 at 1x LoD (3x103 copies/ml). A positive extraction control (Pos. Ctrl) negative extraction controls (DNA/RNA Shield, Human RNA) and qPCR controls all returned expected results (c). Ct values >37 are shaded in gray. Undetected Ct values are plotted as zero and designated by “ND”, not detected. e, Limit of detection (RNA copies/μl) comparison of commercial assays and the IGI saliva and IGI swab tests generated from the FDA website [9].

Fig 3. Comparison swab and IGI FAST saliva specimen results over time.

a, Final results for saliva specimens collected from asymptomatic individuals through the IGI FAST study by week. Some samples were collected on week 11 but never processed in the laboratory due to a pipet tip shortage that lasted through weeks 12 and 13. b, Final results for all symptomatic swab-based tests run at IGI during the same timeframe as the IGI FAST study. “*” indicates weeks IGI FAST was suspended due to filter pipet tip shortages.

Fig 4. Four-plex pooling of saliva specimens.

Four-plex pools were generated with saliva samples previously determined to be positive or negative for the presence of SARS-CoV-2 RNA. 81 wells were generated to contain four negative samples and eight contained one positive sample mixed with 3 negative samples. a, All wells containing positive sample pools were called either positive (7/8) or inconclusive (1/8). b, Viral and MS2 Ct values for unpooled and pooled positive saliva samples. Note that unpooled saliva sample 2 was run in two separate positive pools while all other positive pools contained unique positive samples. Samples were sorted by the N gene Ct value in unpooled saliva samples. Ct values >37 are shaded in gray. Undetected Ct values are plotted as zero and designated by “ND”, not detected. c, Results of pooled samples are compared to the final results for unpooled saliva samples run on the finalized saliva extraction protocol. “Specimen Insufficient” pooled samples were subsequently re-run either in new pools or unpooled to obtain a final result.

Fig 5. Clinical concordance between IGI FAST saliva and NP swab samples.

The IGI partnered with Washington Hospital Healthcare System (WHHS) to collect paired nasopharyngeal (NP) swab and saliva specimens to assess the concordance between NP swab and saliva-based tests for detection of SARS-CoV-2. a, A schematic of how paired samples were collected and split for analysis. Note that undiluted NP samples were analyzed at ARUP Laboratories (ARUP NP) whereas NP samples analyzed at IGI (IGI NP) were diluted 1:1 with DNA/RNA Shield prior to extraction. b, Viral Ct values for IGI NP, saliva samples, and ARUP (where available) (top) and the final result of ARUP NP, IGI NP, and saliva samples (bottom). TMA indicates the sample was analyzed using transcription-mediated amplification, thus no Ct values were generated. Samples were sorted by the IGI NP N gene Ct value. As only positive samples are presented, MS2 Ct values were omitted for clarity. Ct values >37 are shaded in gray. Undetected Ct values are plotted as zero and designated by “ND”, not detected. An aliquot of the NP sample for patient 81 was never received by IGI. c, Concordance between the IGI saliva and IGI NP and d, IGI saliva and ARUP NP assays. Note: some individual samples had no result (invalid or no sample/result received) and were thus left out of Positive Percent Agreement (PPA) and Negative Percent Agreement (NPA) calculations. Figure made with Biorender.