tetano
Editor, Senior Moderator
Elife
. 2026 Jan 16:13:RP96032.
doi: 10.7554/eLife.96032.
Stratification of viral shedding patterns in saliva of COVID-19 patients
Hyeongki Park[SUP] 1 2 [/SUP], Yoshimura Raiki[SUP] 1 [/SUP], Shoya Iwanami[SUP] 1 [/SUP], Kwangsu Kim[SUP] 1 3 4 [/SUP], Keisuke Ejima[SUP] 5 6 [/SUP], Naotoshi Nakamura[SUP] 1 [/SUP], Kazuyuki Aihara[SUP] 7 [/SUP], Yoshitsugu Miyazaki[SUP] 8 [/SUP], Takashi Umeyama[SUP] 8 [/SUP], Ken Miyazawa[SUP] 8 [/SUP], Takeshi Morita[SUP] 9 [/SUP], Koichi Watashi[SUP] 9 [/SUP], Christopher B Brooke[SUP] 10 11 [/SUP], Ruian Ke[SUP] 12 [/SUP], Shingo Iwami[SUP] #[/SUP][SUP] 1 7 13 14 15 16 17 [/SUP], Taiga Miyazaki[SUP] #[/SUP][SUP] 18 [/SUP]
Affiliations
Living with COVID-19 requires continued vigilance against the spread and emergence of variants of concern (VOCs). Rapid and accurate saliva diagnostic testing, alongside basic public health responses, is a viable option contributing to effective transmission control. Nevertheless, our knowledge regarding the dynamics of SARS-CoV-2 infection in saliva is not as advanced as our understanding of the respiratory tract. Here, we analyzed longitudinal viral load data of SARS-CoV-2 in saliva samples from 144 patients with mild COVID-19 (a combination of our collected data and published data). Using a mathematical model, we quantified individual-level viral dynamics and stratified them into three groups using a clustering approach. Notably, the three groups exhibited distinct differences in viral RNA detection durations: 11.5 days (95% CI: 10.6-12.4), 17.4 days (16.6-18.2), and 30.0 days (28.1-31.8), respectively. Surprisingly, this stratified grouping remained unexplained despite our analysis of 47 types of clinical data, including basic demographic information, clinical symptoms, results of blood tests, and vital signs. Additionally, we quantified the expression levels of 92 micro-RNAs in a subset of saliva samples, but these also failed to explain the observed stratification, although the mir-1846 level may have been weakly correlated with peak viral load. Our study provides insights into SARS-CoV-2 infection dynamics in saliva, highlighting the challenges in predicting the duration of viral RNA detection without indicators that directly reflect an individual's immune response, such as antibody induction. Given the significant individual heterogeneity in the kinetics of saliva viral shedding, identifying biomarker(s) for viral shedding patterns will be crucial for improving public health interventions in the era of living with COVID-19.
Keywords: SARS-CoV-2; infectious disease; machine learning; mathematical model; microbiology; viral dynamics; viruses.
. 2026 Jan 16:13:RP96032.
doi: 10.7554/eLife.96032.
Stratification of viral shedding patterns in saliva of COVID-19 patients
Hyeongki Park[SUP] 1 2 [/SUP], Yoshimura Raiki[SUP] 1 [/SUP], Shoya Iwanami[SUP] 1 [/SUP], Kwangsu Kim[SUP] 1 3 4 [/SUP], Keisuke Ejima[SUP] 5 6 [/SUP], Naotoshi Nakamura[SUP] 1 [/SUP], Kazuyuki Aihara[SUP] 7 [/SUP], Yoshitsugu Miyazaki[SUP] 8 [/SUP], Takashi Umeyama[SUP] 8 [/SUP], Ken Miyazawa[SUP] 8 [/SUP], Takeshi Morita[SUP] 9 [/SUP], Koichi Watashi[SUP] 9 [/SUP], Christopher B Brooke[SUP] 10 11 [/SUP], Ruian Ke[SUP] 12 [/SUP], Shingo Iwami[SUP] #[/SUP][SUP] 1 7 13 14 15 16 17 [/SUP], Taiga Miyazaki[SUP] #[/SUP][SUP] 18 [/SUP]
Affiliations
- PMID: 41543519
- PMCID: PMC12810952
- DOI: 10.7554/eLife.96032
Living with COVID-19 requires continued vigilance against the spread and emergence of variants of concern (VOCs). Rapid and accurate saliva diagnostic testing, alongside basic public health responses, is a viable option contributing to effective transmission control. Nevertheless, our knowledge regarding the dynamics of SARS-CoV-2 infection in saliva is not as advanced as our understanding of the respiratory tract. Here, we analyzed longitudinal viral load data of SARS-CoV-2 in saliva samples from 144 patients with mild COVID-19 (a combination of our collected data and published data). Using a mathematical model, we quantified individual-level viral dynamics and stratified them into three groups using a clustering approach. Notably, the three groups exhibited distinct differences in viral RNA detection durations: 11.5 days (95% CI: 10.6-12.4), 17.4 days (16.6-18.2), and 30.0 days (28.1-31.8), respectively. Surprisingly, this stratified grouping remained unexplained despite our analysis of 47 types of clinical data, including basic demographic information, clinical symptoms, results of blood tests, and vital signs. Additionally, we quantified the expression levels of 92 micro-RNAs in a subset of saliva samples, but these also failed to explain the observed stratification, although the mir-1846 level may have been weakly correlated with peak viral load. Our study provides insights into SARS-CoV-2 infection dynamics in saliva, highlighting the challenges in predicting the duration of viral RNA detection without indicators that directly reflect an individual's immune response, such as antibody induction. Given the significant individual heterogeneity in the kinetics of saliva viral shedding, identifying biomarker(s) for viral shedding patterns will be crucial for improving public health interventions in the era of living with COVID-19.
Keywords: SARS-CoV-2; infectious disease; machine learning; mathematical model; microbiology; viral dynamics; viruses.