Diagnostic Performance and Tolerability of Saliva and Nasopharyngeal Swab Specimens in the Detection of SARS-CoV-2 by RT-PCR

ABSTRACT Saliva is a promising alternative for a nasopharyngeal swab (NPS) in specimen collection to detect SARS-CoV-2. We compared the diagnostic performance and tolerability of saliva collection versus NPS in a clinical setting. Paired NPS and saliva specimens were collected sequentially from participants (n = 250) at the Turku University Hospital drive-in coronavirus testing station in the spring of 2022, with Omicron BA.2 as the dominant SARS-CoV-2 variant. Discomfort and preference for the sampling method were assessed. The specimens were analyzed for SARS-CoV-2 using real-time multiplex reverse transcriptase PCR (RT-PCR) with a laboratory-developed test (LDT) and two commercial kits (PerkinElmer SARS-CoV-2 and PerkinElmer SARS-CoV-2 Plus) for several target genes. Among the 250 participants, 246 had respiratory symptoms. With LDT, SARS-CoV-2 was detected in 135 and 134 participants from NPS and saliva, respectively. Of the 250 specimens, 11 gave a discordant outcome, resulting in excellent agreement between the specimen types (Cohen’s kappa coefficient of 0.911; P = 0.763). The cycle threshold (CT) values of LDT and commercial kit target genes were significantly lower from NPS than from saliva. A total of 172 (69%) participants assessed saliva sampling as more tolerable than NPS (P < 0.0001). Our findings present saliva as an applicable alternative for SARS-CoV-2 diagnostics. However, the lower CT values obtained from NPS indicate that NPS may be a slightly more sensitive specimen type. Participants preferred saliva sampling, although delivering an adequate volume of saliva was challenging for some participants. IMPORTANCE The extensive testing of SARS-CoV-2 is vital in controlling the spread of COVID-19. The reference standard for specimen collection is a nasopharyngeal swab (NPS). However, the discomfort of NPS sampling, the risk of nosocomial infections, and global material shortages have accelerated the development of alternative testing methods. Our study demonstrates that patients tolerate saliva sampling better than NPS. Of importance, although the RT-PCR qualitative test results seem to correspond between NPS and saliva, we show significantly lower CT values for NPS, compared to saliva, thus contradicting the suggested superiority of the saliva specimen over NPS in the detection of the Omicron variants of SARS-CoV-2. Future research is still required to enable individual planning for specimen collection and to determine the effects of different SARS-CoV-2 variants on the sensitivity of the saliva matrix.

The manuscript by Ahti and co-workers adds another piece of evidence regarding the diagnostic detection of SARS-CoV-2 Omicron (subvariant BA.2) in NPS versus saliva. The authors well demonstrated that although the coefficient of correlation showed good agreement no matter what RT-qPCR kit used (either lab-based or commercial), the Ct ranges when comparing the two types of samples showed a significative higher viral load on NPS, which demonstrates that saliva can be a less sensitive type of sample for SARS-CoV-2 diagnostics, even considering the Omicron circulation. The data also helps to add to the evidence in the field showing that the nasopharyngeal compartment will harbor higher viral loads during SARS-CoV-2 infection, which was doubted by previous data suggesting higher viral loads for Omicron in the saliva. The study was based on a reasonable number of specimens (n= 250) and both the study design and generated data are sound. Some points need clarification: 1) Authors say that saliva were kept on room temperature for up to 14 days prior to RNA extraction (kit's manufactures' recommendation). It seems quite an extended amount of time to keep saliva containing enveloped viral particles without reducing the amount of viable viral particles. Have authors conducted experiments to assure the viability of these samples? Wouldn't that explain the lower viral load? What was the average time from collection to nucleic acid extraction for samples tested? It was not clear from the sentence that NSP samples were kept at 4C also up to 14 days? Again, seems quite an extensive period to keep viral particles viable. 2) Authors mentioned 9 saliva and NPS samples kept for 3 days at 4C after extraction, since RNA is chemically liable and consequently heat degradable, it should be mentioned how these samples performed in RT-qPCR. 3) Why is the difference between NPS and saliva on the range of 5-6 Cts higher for E and S targeted genes, but only 2 Cts for the Sdel? is the Median for Sdel in saliva calculated correctly, since the IQR had such a high range? Could authors comment on this data? Also, what was the relevance of using S as target with primers/probe sets to distinguish the del69-70 if this specific wasn't shown? 4) Authors should provide a figure or table showing the discordant samples in each RT-qPCR assay they performed, were they the same? Did it vary according to the assay or the sample? How was the internal control performance for discordant samples? 5) In lanes 208-211 authors mention that Ct values in saliva were lower for drinking, eating, etc before sample collection, but the difference was not significant, thus they shouldn't say it was different! The sentence should be rephrased to better describe the finding. 6) Authors affirm that the time-of-day saliva was collected make no difference for diagnostics, however, no data was shown, specifically, how many periods were evaluated? how were they determined? how many samples foreach period were analyzed? This set of data is of relevance and should be shown.
As a minor observation, the mean age and standard deviation of participants should not be mentioned on the abstract.  There are several strengths to our study. The participants were prospectively 296 enrolled in a clinical setting at a drive-in testing station and paired specimens were collected. 297 Saliva collection was supervised by the study nurse to minimize incorrect collection procedures, 298 and to ensure that an adequate volume of saliva was provided. All the paired specimens were 299 analyzed with three tests, each of which provided consistent results. Additionally, several target 300 Please create a separate paragraph of clinical recommendations from this study: Regarding use of saliva as specimen, timing of sample collection, preference of patients, genes of SARS-CoV-2 were analyzed. We also evaluated the discomfort associated with both 301 sampling methods and which sampling method the participant preferred. 302 There are also several limitations to our study. The nucleic acid extraction and RT-303 qPCRs were not performed on the same day but within a few days of specimen collection. 304 Another limitation is that some saliva specimens did not fulfill the minimum requirement of 2 305 ml of saliva volume. Additionally, people who experience nasopharyngeal sampling more 306 painful than others were probably more likely to participate in a study evaluating an alternative 307 sampling method, thus causing a potential selection bias on the discomfort evaluation. On the 308 other hand, we did not specify that the selection of the preferred sampling method should only 309 be based on the sampling experience, therefore some participants might have considered the 310 traditional sampling method more reliable and thus have preferred NPS even though it caused 311 more discomfort.

312
In summary, our study shows excellent sensitivity and specificity of saliva 313 specimens in the detection of SARS-CoV-2 by RT-qPCR. However, the CT values of all the 314 diagnostic target genes were significantly lower from NPS than saliva, indicating that NPS may 315 be slightly more sensitive in the detection of SARS-CoV-2 compared with saliva. In terms of 316 tolerability, saliva sampling was highly preferred, yet delivering an adequate volume of saliva 317 was challenging for some participants.      Figure 3. The discomfort associated with nasopharyngeal (NPS) and saliva sampling. The 508 participants assessed discomfort with a discomfort scale from 1 (no discomfort) to 5 (extreme 509 discomfort). The evaluation is missing from one participant. 510 1 March 31st, 2023

Spectrum05324-22 (Diagnostic Performance and Tolerability of Saliva and Nasopharyngeal Swab Specimens in Detection of SARS-CoV-2 by RT-PCR)
Response to Reviewers:

Reviewer #1
The manuscript by Ahti and co-workers adds another piece of evidence regarding the diagnostic detection of SARS-CoV-2 Omicron (subvariant BA.2) in NPS versus saliva. The authors well demonstrated that although the coefficient of correlation showed good agreement no matter what RT-qPCR kit used (either lab-based or commercial), the Ct ranges when comparing the two types of samples showed a significative higher viral load on NPS, which demonstrates that saliva can be a less sensitive type of sample for SARS-CoV-2 diagnostics, even considering the Omicron circulation. The data also helps to add to the evidence in the field showing that the nasopharyngeal compartment will harbor higher viral loads during SARS-CoV-2 infection, which was doubted by previous data suggesting higher viral loads for Omicron in the saliva.
The study was based on a reasonable number of specimens (n= 250) and both the study design and generated data are sound.
-We thank the Reviewer for their appraisal. We did our best in preparing point-by-point responses (found below) to the raised issues. We hope that these responses fulfill Your expectations.
Some points need clarification: 1) Authors say that saliva were kept on room temperature for up to 14 days prior to RNA extraction (kit's manufactures' recommendation). It seems quite an extended amount of time to keep saliva containing enveloped viral particles without reducing the amount of viable viral particles. Have authors conducted experiments to assure the viability of these samples? Wouldn't that explain the lower viral load? What was the average time from collection to nucleic acid extraction for samples tested? It was not clear from the sentence that NSP samples were kept at 4C also up to 14 days? Again, seems quite an extensive period to keep viral particles viable.
-We thank the Reviewer for this important comment. The time from collection to nucleic acid extraction for the paired saliva and NPS samples of each patient were: 0 days for 12 paired samples, 2 days for 13 paired samples, 3 days for 47 paired samples, 4 days for 42 paired samples, 5 days for 56 paired samples, 6 days for 53 paired samples, 7 days for 18 paired samples, and 16 days for 9 paired samples. The extended amount of time before extraction of 9 (3.6%) paired samples was caused by the shortage of laboratory personnel during the COVID-19 pandemic. The median time from collection to nucleic acid extraction was 5 (IQR 3-6) days. We have added the following sentence to the Results section (lines 177-178 of the Marked-Up Manuscript): "The median time from specimen collection to nucleic acid extraction was 5 (IQR 3-6; range 0-16) days." According to the CE-IVD validated DNA/RNA Shield™ Saliva/Sputum Collection Kit manufacturer´s protocol saliva can be stored at room temperature for up to 28 days. The maximum time from sample collection to RNA extraction was 16 days instead of 14 days due to a weekend but still well within the manufacturer specified 28 days. We apologize the error and we have corrected the following sentence in the Diagnostic methods paragraph, line 118: "NPS specimens were stored at +4 °C and saliva specimens at room temperature according to instructions of the specimen collection kit for a maximum of 16 days." Considering that the saliva samples were stored at the validated specimen collection kit and the paired saliva and NPS samples had the same storage time before extraction, it would seem unlikely that the lower viral load of saliva samples was caused by the storage time. 2) Authors mentioned 9 saliva and NPS samples kept for 3 days at 4C after extraction, since RNA is chemically liable and consequently heat degradable, it should be mentioned how these samples performed in RT-qPCR.
-We thank the Reviewer for the comment. Of these 9 samples, concordant qualitative test results were obtained from both saliva and NPS (4 samples were positive with all three tests from both specimen types 3) Why is the difference between NPS and saliva on the range of 5-6 Cts higher for E and S targeted genes, but only 2 Cts for the Sdel? is the Median for Sdel in saliva calculated correctly, since the 3 IQR had such a high range? Could authors comment on this data? Also, what was the relevance of using S as target with primers/probe sets to distinguish the del69-70 if this specific wasn't shown?
-We thank the Reviewer for lifting this up. In the LDT, E gene served as the diagnostic target. The qualitative analysis of two S gene targets (S gene and Sdel) was used for epidemiologic surveillance of new SARS-CoV-2 variants in Southwest Finland. As we have described in the Materials and Methods section (lines 138-141), varying background due to cross-reactivity of S gene with the Sdel probe was observed in the Sdel channel (TexasRed/orange). The threshold level was manually set above the background fluorescence and the threshold was at a higher level compared to the other gene targets, resulting in the higher CT values for the Sdel analysis as compared to other gene targets. We checked the calculation and found no errors and no obvious explanation for the small CT difference between NPS and saliva for Sdel. 6) Authors affirm that the time-of-day saliva was collected make no difference for diagnostics, however, no data was shown, specifically, how many periods were evaluated? how were they determined? how many samples for each period were analyzed? This set of data is of relevance and should be shown.
-We thank the Reviewer for this valuable comment since we were unclear in explaining how we interpreted the association between the sample collection time and diagnostic sensitivity. First, the time of the sample collection was recorded with one-minute precision. In the statistical analysis, binary logistic regression was used to evaluate the association between the time of day of saliva sample collection and the qualitative test result or CT values from saliva. In these calculations, time is a continuous variable, which is compared to the qualitative test results of the saliva samples (categorical variable) and to the CT values from saliva (continuous variable). That is to say, the time of sample collection was not divided into different periods (e.g., morning or afternoon) and statistical calculations were performed only using time as a continuous variable.
We have now also inspected our results using time of collection as a categorical variable. The time of collection was divided into two periods: before 12.00 (morning; number of collected samples, 67) and from 12.00 onward (afternoon and evening; number of collected samples, 177); data missing from 6 participants. The time of collection ranged from 9.00 am to 6.39 pm. Fisher's exact test (X 2 test) was used to compare categorical time and qualitative results of the LDT E gene (P = 0.39), PerkinElmer N gene (P = 0.56), and PerkinElmer Plus N/E genes (P = 0.56). Two sample ttest was used to compare categorical time and CT values from saliva. No significant difference was seen for the mean CT values of the LDT E gene (P = 0.20), PerkinElmer N gene (P = 0.22), and PerkinElmer Plus N/E genes (P = 0.32). We have clarified the use of time as a continuous variable to the Statistical analysis paragraph of the Materials and Methods section (lines 166-167): "The sample collection time was treated as a continuous variable, and binary logistic regression was used to evaluate the association between the time of day of saliva specimen collection and the test result from saliva." We have also added the range of sample collection time, and the number of collected samples per period (morning vs afternoon and evening) to the Results section (lines 227-228): "The majority of saliva specimens (177/244, 73%; data missing from six participants) were collected after 12 am (IQR,11.00 am to 1.55 pm; range, 9.00 am to 6.39 pm)." We have not added the categorical calculations of time to the findings to our revised manuscript, as the evaluation of the time of saliva sample collection was not the main focus of this manuscript and we think that our calculation using time as a continuous variable would describe the finding well enough.
As a minor observation, the mean age and standard deviation of participants should not be mentioned on the abstract.
-We thank the Reviewer for the observation. We have removed the mean age and standard deviation of participants from the abstract. 4) Please create a separate paragraph of clinical recommendations from this study: Regarding use of saliva as specimen, timing of sample collection, preference of patients.
-We thank the Reviewer for this valuable remark since we were insufficient in explaining the generalizability of this research. As a response, we have added the following paragraph to the Discussion section of the manuscript (lines 313-323): "Based on the results of this study, NPS should remain to be the standard specimen type for the detection of the SARS-CoV-2 Omicron variant since the lower CT values of NPS indicate that saliva may be a slightly less sensitive specimen type for SARS-CoV-2 diagnostics. On the other hand, the sensitivity of the saliva specimen in this study was similar to NPS, which supports the use of saliva as an alternative to NPS in certain situations. Saliva could reasonably be used for patients that are exposed to complications of NPS collection, are unwilling to sample with NPS, or are particularly sensitive to the discomfort caused by NPS collection. Regarding our observations, saliva could also be collected at any time of day and without strict restrictions to oral hygiene, thus making the collection at home easy to perform. Sample collection at home might reduce risk of nosocomial infections, especially in lowincome countries where a lack of protective equipment might be more common."