Detection of three pandemic causing coronaviruses from non-respiratory samples: systematic review and meta-analysis

SARS-CoV-2 has posed an unprecedented challenge to the world. Pandemics have been caused previously by viruses of this family like Middle East Respiratory Corona Virus (MERS CoV), Severe Acute Respiratory Syndrome Corona Virus (SARS CoV). Although these viruses are primarily respiratory viruses, but they have been isolated from non-respiratory samples as well. Presently, the detection rate of SARS‐CoV‐2 RNA from different clinical specimens using Real Time Reverse Transcriptase Polymerized Chain Reaction (qRT‐PCR) after onset of symptoms is not yet well established. Therefore, the aim of this systematic review was to establish the profile of detecting SARS‐CoV‐2, MERS CoV, SARS CoV from different types of clinical specimens other than the respiratory using a standard diagnostic test (qRT‐PCR). A total of 3429 non-respiratory specimens were recorded: SARS CoV (total sample—802), MERS CoV (total sample—155), SARS CoV-2 (total sample—2347). Out of all the samples studied high positive rate was seen for saliva with 96.7% (14/14; 95% CI 87.6–100.0%) for SARS CoV and 57.5% (58/250; 95% CI − 1.2 to 116.2%) for SARS CoV-2, while low detection rate in urine samples for SARS CoV-2 with 2.2% (8/318; 95% CI 0.6–3.7%) and 9.6% (12/61; 95% CI − 0.9 to 20.1%) for SARS CoV but there was relatively higher positivity in urine samples for MERS CoV with detection rate of 32.4% (2/38; 95% CI − 37.3 to 102.1%). In Stool sample positivity was 54.9% (396/779; 95% CI 41.0–68.8%), 45.2% (180/430; 95% CI 28.1–62.3%) and 34.7% (4/38; 95% CI − 29.5 to 98.9%) for SARS CoV-2, MERS CoV, and SARS CoV, respectively. In blood sample the positivity was 33.3% (7/21; 95% CI 13.2–53.5%), 23.7% (42/277; 95% CI 10.5–36.9%) and 2.5% (2/81; 95% CI 0.00–5.8%) for MERS CoV, SARS CoV-2 and SARS CoV respectively. SARS‐CoV‐2 along with previous two pandemic causing viruses from this family, were highly detected stool and saliva. A low positive rate was recorded in blood samples. Viruses were also detected in fluids along with unusual samples like semen and vaginal secretions thus highlighting unique pathogenic potential of SARS‐CoV‐2.

The search terms were combined using Boolean logic "or" and "and". Filters were set to exclude non-human studies. Finally, the article that was directed to relevant categories as indicated in the PRISMA flow diagram (Fig. 1).

Eligibility criteria.
Clinical trial and observational studies including cross-sectional studies, retrospective studies and prospective studies were selected for review. Case series and case study including even one patient per study was considered. This review is limited and focused to isolation of the Human Corona virus specifically SARS CoV, MERS CoV, SARS CoV-2 by real time Reverse Transcriptase Polymerase Chain Reaction i.e., qRT-PCR. The studies that had targeted non-respiratory sample positivity for the qRT-PCR was taken in consideration. The studies which were conducted to determine diagnostic accuracy, reviews and non-human articles were excluded.
Data extraction. The selection of the articles was done by 4 independent reviewers (SM, CM, JKM and ST) who evaluated the articles for the potential inclusion by screening the titles and abstract followed by full-text screening to determine eligibility of inclusion of the article for final study. Any discrepancies in inclusion of the articles were resolved by SM. The single specimen positive by qRT-PCR was taken into study. The number of patients tested were counted for each type of the specimen rather than the number of samples.
Result analysis. Among 80 articles included the all the positive cases were added for the study. Information was recorded if there at least one specimen tested positive (confirmed cases). Different types of non-respiratory clinical samples were recorded for SARS CoV, MERS CoV, SARS CoV-2. Various type of clinical samples like stool/faecal, anal/rectal swab, urine, semen, testes tissue, vaginal secretions/mucus, vaginal swabs, placental specimen, amniotic fluid, whole blood, serum, plasma, cord blood, ascitic fluid, peritoneal fluid, gastric fluid, pericardial fluid, pleural fluid, breast milk, saliva, ocular surface samples, ocular swab sample, conjunctival swab, tears, CSF were taken in consideration. The rectal/anal swab specimen in the studies that also included stool/ faecal specimen, the sample with maximum positivity was taken into account i.e. the stool/faecal specimen, in following studies by Huang et al. 5  www.nature.com/scientificreports/ estimate (test positivity rate/proportion) by a random-effects model using Open Meta Analyst software 85 . The results were expressed using by pooled effect estimates and their 95% confidence intervals (CIs) and a correction factor of 0.5. Heterogeneity is commonly observed within diagnostic test accuracy reviews due to the differences in study quality, sample size, method, and different outcomes for which random effects models were kept default. Heterogeneity in the analysed studies was determined using Cochrane Q-statistic test and I 2 statistic with subgrouping performed based on the types of clinical specimens. (Fig. 2). Between-study heterogeneity assessed using and I 2 statistic. I 2 ranges from 0 and 100% and larger values represent increasing heterogeneity (Table 1).
Sensitivity analysis. Sensitivity analysis was done to determine the effect of individual studies on the pooled estimates. Univariate meta-regression was performed from primary studies using the random-effects model. A leave-one-out sensitivity analysis was done to assess heterogeneity between the study results. It was performed by iteratively removing one study at a time to confirm that our findings were not driven by any single study (Fig. 3a,b). Quality assessment. The assessment was performed by four independent reviewers and further checked by two additional reviewers. As recommended by Cochrane Handbook for Systematic Reviews of Diagnostic Accuracy 86 , we adopted QUADAS-2 (Quality Assessment of Diagnostic Accuracy Studies -2) 87 to evaluate the bias and quality of selected studies. The following four domains were considered for risks of bias and application concerns as depicted in the assessment tool: (1) participant selection; (2) index test; (3) reference text; and (4) flowing and timing. Studies with more than one "high risk of bias" were excluded (Fig. 4).
Publication bias. Output created by Open Meta Analyst software through a cumulative forest plot was indirectly used to assess the publication bias 88 . Funnel Plot and and Egger's and Begg tests were conducted to assess the publication bias (P > 0.05) of the enclosed literature ( Table 2). As seen in (Fig. 5), the funnel plot appearing symmetric, indicates absence of substantial publication bias.
Positive detection rate from non-respiratory samples. A total of 3429 non-respiratory specimens were recorded having SARS CoV (total sample-802), MERS CoV (total sample-155), SARS CoV-2 (total sample -2347). Here number of samples corresponds to number of patients. As we observed high rates of data heterogeneity further sensitivity analysis was undertaken to identify its potential source (study

Discussion
This systematic review and meta-analysis provide comprehensive data on detection of pandemic causing coronaviruses from non-respiratory samples. Our findings suggest that all the viruses are mainly respiratory, but they can cause multisystem involvement. They are detected from stool, urine, semen, testes tissue, vaginal secretions, placental specimen, amniotic fluid, whole blood, serum, plasma, cord blood, ascitic fluid, peritoneal fluid, gastric fluid, pericardial fluid, pleural fluid, breast milk, saliva, ocular specimen/tears and CSF. This finding is supported by several studies demonstrating detection of these viruses from various non-respiratory samples enlisted above. These findings show that in clinical practice non-respiratory samples are also important. SARSCoV-2 seems to have a strong predilection for the angiotensin two converting enzyme (ACE-2): the wide expression in different human tissues (as well as in the lung also in the intestine, testicle, kidney, etc.,) would also justify different theoretical modes of transmission of the virus in addition to respiratory route.
Out of all the samples studied maximum positivity was seen for saliva. This study revealed that saliva specimen had positivity of 57.5% for SARS CoV-2 detection and 96.7%for SARS CoV 40,48,68,71 . Saliva has been accepted as reliable non-invasive specimen with high sensitivity in comparison to NPS and throat swab 48,71 . Now that FDA approved kits are also available for diagnosis of covid-19 from saliva, this can be a safe alternative to respiratory samples 89 . Saliva has the advantage of being easiest to collect amongst all samples. Although positivity rate of saliva for SARS CoV-2 seems lower in our analysis as compared to previous study 3 . It could be explained by the sample size of the study by Pasomsub et al. 2020 was 200 out of which only 19 patients were positive for COVID-19 by NPS of which 18 tested positive for the saliva specimen causing rest to be outliers leading to decreased percentage of positivity. SARS CoV and SARS CoV-2 both has equally detected from saliva but there were no studies on detection of MERS.
The stool sample is well established for high sensitivity for the corona viruses 29 . This study also had same interpretation though the positivity rate for MERS CoV was lower than that of the other two viruses. Nevertheless, stool sample can be an alternate sample for diagnosis, but its clinical relevance and infectiousness needs to All the three viruses have been detected from both serum and blood during early period of viremia 28,44,70 It is highest for SARS-CoV-2 followed by MERS-CoV and SARS CoV. But the detection rate is highly variable depending on the timing of sample collection and significantly less as compared to respiratory sample. Nevertheless, it can complement diagnosis in absence of positivity from respiratory samples 90 .
Another interesting finding was detection of these viruses from genito-urinary specimen. There was 2.2% positivity in urine samples for SARS CoV-2,32.4%for MERS-CoV and 9.6%for SARS CoV. Detection of SARS CoV-2 from semen 9,33,82 sample was 9.8% and vaginal 26 swabs were 21.2%. Detection from these samples highlight unique pathogenic potential of this virus which needs to be evaluated further. www.nature.com/scientificreports/ There were studies that arises question about vertical transmission of corona viruses. Positivity in placental samples23.4% and in amniotic fluid 8.1%for SARS CoV-2 was seen in post-delivery women, but none were found for the SARS CoV. No study for the MERS CoV was found. The post-delivery cord blood sample was not positive in SARS CoV-2 and SARS CoV. Similarly, none of the three corona viruses were not detected in the breast milk samples from pregnant woman.
To our knowledge, this is the first systematic review to comprehensively examine and compare SARS-CoV-2, SARS-CoV, and MERS-CoV detection in non-respiratory samples. Our study has limitations. First, almost all patients in the included had detection from non-respiratory samples at various time points which might have affected the results. Second, our meta-analysis identified substantial study heterogeneity, probably due to differences in study population, different extraction methods and kits used in RT-PCR. Thirdly, most of the studies published are case reports or case series where patient selection may have been biased. As a result, our analyses were based on varying methods, sample timing, sample frequencies and study endpoints differing widely between the included studies impairing contrasts and strong conclusions.
We recognized a systematic review and meta-analysis published on detection of SARS-CoV-2 from different clinical specimen using RT-PCR that included studies published up until may2020. The study include 7 studies and also analysed all samples including respiratory samples 3 . Our study on the other hand has analysed 80 studies wherein a total of 3429 samples were included. Besides, we have also compared the positivity in non-respiratory specimen of the three pandemic causing viruses which has not been attempted elsewhere.
This review provides detailed understanding about the evidence available so far on detection of SARS-CoV-2, SARS-CoV, and MERS-CoV from non-respiratory samples. It has implications in understanding viral dynamics and possible transmission routes to determine preventive steps that need to be taken. Various mitigation and prevention strategies for the current ongoing pandemic should be designed and implemented keeping in mind that this virus could have non-respiratory route of transmission.

Conclusion
In this study, SARS-CoV-2 along with previous two pandemic causing viruses from this family, were highly detected stool and saliva. A low positive rate was recorded in blood samples. Also, viruses were also detected in fluids along with unusual samples like semen and vaginal secretions thus highlighting unique pathogenic potential of SARS-CoV-2. Thus, mitigation and prevention strategies should also focus on non-respiratory routes of transmission for the current pandemic.     Table 3. Detection of SARS, MERS, SARS CoV-2 in different non-respiratory samples.