Main

As of 6 March 2022, the Coronavirus Disease 2019 (COVID-19) pandemic has caused more than 443 million cases and just over 5.9 million deaths globally1. SARS-CoV-2, the causative agent of COVID-19, primarily infects the cells of the upper respiratory tract (URT) where VL increases during the course of infection2.

The two key measurements of VL are RNA levels, often expressed in cycle threshold (Ct) values, and infectious virus that is assessed by virus isolation in cell culture. Although the transmission process is complex, higher VL can serve as a proxy for greater risk of transmission. In several epidemiological studies, higher VL measured by viral RNA was associated with increased secondary transmission in household settings3,4. Infectious SARS-CoV-2 is shed in the URT, starting, on average, from 2 days before symptom onset. In most studies, infectious virus was not detected in respiratory samples collected from non-hospitalized immunocompetent individuals later than 8 days post onset of symptoms (DPOS)5,6,7. Moreover, viral RNA detection did not correlate with infectiousness in an animal model8. Instead, isolation success in cell culture—that is, the ability to replicate the virus in cell culture—was found to correlate with the ability to shed and transmit fully competent viral particles9. Virus isolation success from respiratory tract samples can give information only about the presence or absence of infectious virus but is not able to quantify the infectious viral titer in samples of the URT10.

Since the start of the pandemic, SARS-CoV-2 has constantly evolved, leading to the emergence of new variants. Although most variants vanished quickly, others, such as D614G and the VOCs Alpha, Beta, Gamma, Delta and Omicron, harbor an apparent selection advantage and outcompeted other variants locally or even globally. These VOCs exhibit various mutations11 that lead to immune evasion and/or higher transmissibility, to which increased viral shedding (among other factors, such as environmental stability) may significantly contribute12,13. For Alpha, an approximately ten-fold-higher RNA VL was described compared to pre-VOC viral strains, which was correlated with increased isolation success14,15. Similarly, Delta also showed 10–15-fold-higher RNA levels compared to pre-VOC strains in some studies15,16. In contrast, a study using longitudinal samples did not find a difference among peak RNA VL of pre-VOC, Alpha and Delta17. However, little is known about the quantity of shed infectious viral particles for VOCs, including Omicron.

There is extensive evidence that vaccines against SARS-CoV-2, which target the original strain, reduce infection case numbers and disease severity. However, the effect of vaccination on infectious viral shedding and transmission from vaccinated individuals remains controversial. All currently approved vaccines are administered intramuscularly; thus, the titer of neutralizing antibodies on the mucosal surfaces lining the URT might be limited, and any sterilizing mucosal immunity might be transient18. Epidemiological studies of the secondary attack rate in households of vaccinated versus unvaccinated index cases report contradictory results on the potential effect of vaccination19,20,21. Multiple factors can influence the secondary attack rate in these studies, including patient behavior, age, comorbidities, the infecting variant, time since vaccination and the vaccine used. Therefore, differentiating the effect of vaccination on VL from other factors in purely epidemiological studies is difficult. To our knowledge, no study has directly quantified infectious VL of different VOCs in URT samples of vaccinated and unvaccinated patients with COVID-19.

The dynamics of infectious viral shedding in vaccinated and unvaccinated individuals infected with relevant VOCs require detailed investigation. Understanding of viral shedding in patients would help shape public health decisions to limit community transmission22. Here we compare RNA and infectious VL among pre-VOC strains, Delta and Omicron BA.1 in unvaccinated individuals, as well as in fully vaccinated (two doses) or boosted (three doses) individuals infected with Delta and Omicron BA.1, using respiratory samples from mildly symptomatic patients of different age and sex, sampled in the first 5 DPOS.

Results

In this study, we analyzed the VL characteristics in the URT of unvaccinated pre-VOC-infected individuals, as well as fully vaccinated, boosted and unvaccinated Delta-infected or Omicron BA.1-infected individuals, up to 5 DPOS. We included a total of 565 samples in our cohort, of which 118 originated from individuals infected with pre-VOC SARS-CoV-2; 293 originated from individuals infected with Delta; and 154 originated from individuals infected with Omicron BA.1. Of individuals infected with Delta, 166 were fully vaccinated before infection, and 127 were unvaccinated. Among individuals infected with Omicron BA.1, 91 were fully vaccinated before infection, 30 were boosted and 33 were unvaccinated. None of the individuals infected with pre-VOC SARS-CoV-2 was vaccinated, as vaccines were unavailable at the time of infection. All infected individuals had mild symptoms at the time of sampling, but the further course of the disease is unknown. Individuals with asymptomatic infection at the time of sampling were excluded from the study. All infected individuals except five (two with Delta breakthrough infections and three with Omicron BA.1 breakthrough infections) were immunocompetent. Samples of pre-VOC-infected individuals were collected between 7 April and 9 September 2020, before detected circulation of any VOCs; samples of Delta-infected individuals were collected from 26 June until 13 December 2021; and samples of Omicron BA.1-infected individuals were collected from 11 December 2021 until 19 February 2022. Each infected individual provided only one sample at a single time point. All vaccinated individuals included in this study were diagnosed positive at least 14 days after dose 2 or dose 3, which complies with the vaccination breakthrough definition of the Centers for Disease Control and Prevention (CDC)23. In total, 274 of 287 patients were vaccinated with mRNA vaccines (Comirnaty or Spikevax); one was vaccinated with a non-replicating viral vector vaccine (CoviVac); one was vaccinated with inactivated virus vaccine CoronaVac; one was vaccinated with viral vector vaccine AZD1222; and, for ten patients, the type of vaccine was not reported by the patient. The median time in days between the second dose and breakthrough infection was 69 days (interquartile range (IQR), 38–122), 160 days (IQR, 137–183) and 154 days (IQR, 86–194) for Delta infections titrated on Vero E6 or Vero E6-TMPRSS cells and Omicron BA.1 infections, respectively. All groups of patients (pre-VOC, Delta-unvaccinated, Delta-vaccinated (two doses), Omicron BA.1-unvaccinated and Omicron BA.1-vaccinated (two or three doses)) had a similar age and sex distribution (Table 1).

Table 1 Patient characteristics of the specimens used in this study

We quantified genome copies and infectious viral titers in SARS-CoV-2-positive nasopharyngeal swabs (NPSs) using qRT–PCR and focus-forming assays (FFAs). Only specimens with Ct values below 27 for the E-gene qRT–PCR diagnostic target (Cobas, Roche), as determined by the clinical laboratory at the University Hospital of Geneva (HUG) at the time of diagnosis, were included in our study, as it was shown previously that infectious virus cannot be reliably isolated from samples with higher Ct values9,24. In our hands, no infectious virus was detected in 46 pre-VOC and Delta samples with Ct values ≥27. We also compared overall percentages of samples with a Ct ≥27 for time periods with almost exclusive circulation of pre-VOC, Delta and Omicron BA.1 by analyzing the overall diagnostic dataset from our outpatient testing center and separating patients by vaccination status and DPOS. Among pre-VOC samples, 19.4% had a Ct ≥27, whereas, in the Delta-infected unvaccinated and vaccinated groups, as well as in the Omicron BA.1-infected unvaccinated and vaccinated groups, 21.4%, 17.6%, 21.4% and 20.7% of samples fell into this category, respectively. No major difference was observed between the proportion of Ct value ≥27 when divided by DPOS (Supplementary Table).

To validate our FFA, we compared it to the ability to successfully isolate virus in cell culture. Virus isolation success has been used as a correlate of infectious viral shedding for SARS-CoV-2 (refs. 6,25,26,27) but lacks the ability to differentiate between high and low VL samples. We were able to quantify viral titers using the FFA in 91.9%, 91.7%, 83.8%, 95% and 85.7% of culture-positive samples in the pre-VOC, Delta-unvaccinated, Delta-fully vaccinated (two doses), Omicron BA.1-unvaccinated and Omicron BA.1-fully vaccinated (two doses) groups, respectively, indicating a high sensitivity (Extended Data Fig. 1a). Overall, the Cohenʼs kappa agreement, which measures the level of agreement between two methods, was 0.69, 0.41, 0.51, 0.66 and 0.47 for the five groups, showing a moderate to substantial agreement (Extended Data Fig. 1b).

Low correlation between genome copies and infectious VL

First, we investigated whether RNA genome copies are a good proxy for infectious virus shedding. We observed only a very low correlation (R2 = 0.1476, P = 0.0001) between viral genome copies and infectious virus particles for pre-VOC samples (Fig. 1a). Likewise, low to moderate correlations between RNA genome copies and infectious viral titers were observed for the samples from unvaccinated and vaccinated Delta patients (R2 = 0.3114, P < 0.0001 and R2 = 0.4021, P < 0.0001, respectively) (Fig. 1b,c), as well as unvaccinated and vaccinated Omicron BA.1 patients (R2 = 0.3638, P = 0.0002 and R2 = 0.3055, P < 0.0001, respectively) (Fig. 1d,e).

Fig. 1: Relationship between RNA VLs and infectious viral titers.
figure 1

Linear regression analysis of infectious viral titers in FFU ml−1 and the corresponding RNA VLs in NPSs from the unvaccinated patients infected with pre-VOC (a), unvaccinated patients infected with Delta VOC (b), fully vaccinated patients infected with Delta titrated in Vero E6 cells (c), unvaccinated patients infected with Omicron BA.1 VOC (d) and fully vaccinated or boosted patients infected with Omicron BA.1 titrated in Vero E6-TMPRSS cells (e). Error bars represent 95% confidence bands of the best-fit line. Two-tailed F-test was used to determine statistical significance, and no adjustments were made for multiple comparisons.

Next, we tested if infectious VLs are associated with patient age and sex. We did not observe any correlation between the age and infectious VL for all four groups (Extended Data Fig. 2). Similarly, no significant differences of infectious VLs between male and female patients were detected for pre-VOC, Delta (fully vaccinated or unvaccinated) or Omicron BA.1 (fully vaccinated) samples (Extended Data Fig. 3).

Delta-infected unvaccinated individuals have higher infectious VL

Next, we compared genome copies and infectious VLs in pre-VOC and Delta samples from unvaccinated patients during the first 5 DPOS. Overall, pre-VOC samples had significantly more genome copies (2.98-fold, 0.4744 log10, P = 0.001, t-value = 3.3512, df = 184.48, Cohen's d = 0.44) than Delta samples, but infectious viral titers were significantly higher in Delta-infected individuals (2.2-fold, 0.343 log10, P = 0.0373, t-value = 2.0967, df = 238.82, Cohen's d = 0.27) (Fig. 2a). We found that genome copies for pre-VOC samples were higher at 1 DPOS and 2 DPOS but similar to Delta samples at 0 DPOS, 3 DPOS, 4 DPOS and 5 DPOS (Fig. 2b). Conversely, infectious virus shedding was higher for Delta at 3–5 DPOS but similar at 0–2 DPOS (Fig. 2c). In addition, we observed that genome copies remained largely stable until 5 DPOS, with only a minimal lower number at day 5, whereas infectious VL was significantly lower for pre-VOC (linear model, day 0 versus day 5, slope significantly < 0, P = 0.00036) but not for Delta (linear model, day 3 versus day 5, slope not significantly < 0, P = 0.07741) (Fig. 2b,c).

Fig. 2: RNA VL and infectious viral titers for unvaccinated individuals infected with pre-VOC SARS-CoV-2 versus Delta.
figure 2

a, Genome copies (left) and infectious virus (right) for pre-VOC and Delta unvaccinated patients. Infectious titers (FFU > 0) were detected in 94 pre-VOC and 112 Delta unvaccinated patients; no titers (FFU = 0) were detected in 24 pre-VOC and 15 Delta unvaccinated patients. Error bars indicate mean ± s.d. Two-tailed t-test was used to determined differences of means. *P = 0.0373 and ***P = 0.001. Genome copies (b) and infectious VLs (c) measured for pre-VOC-infected and Delta VOC-infected patients at different DPOS. The solid lines represent the fitted curve calculated using the LOESS method. Error bars represent 95% confidence bands of the best-fit line. LOESS, locally estimated scatterplot smoothing.

The association of the infectious shedding levels with patient age and sex is highly debated14. In this study, we did not detect a correlation between patient age or sex and infectious VL. However, there is increasing evidence of more severe outcomes of COVID-19 disease in older male patients25,27. Thus, to eliminate possible confounders, 83 patients infected with Delta were matched with pre-VOC infected patients in regard to sex, age and DPOS (Extended Data Fig. 4a). Similarly, significantly higher infectious VLs (3.44-fold, 0.5361 log10, P = 0.001, t-value = 3.5261, df = 41, Cohen's d = 0.54) were detected in Delta samples compared to matched pre-VOC samples (Extended Data Fig. 4b).

Fully vaccinated individuals have lower infectious VL in Delta-infected individuals

To determine vaccination’s association with virus shedding, we compared genome copies and infectious VLs in unvaccinated (n = 127) and vaccinated (n = 104) patients infected with Delta for 5 DPOS. Overall, RNA genome copies were significantly lower in vaccinated versus unvaccinated patients (2.8-fold, 0.44 log10, P = 0.0002, t-value = 3.7942, df = 197.07, Cohen's d = 0.51). The decrease in infectious VL was even more pronounced in vaccinated patients (4.78-fold, 0.68 log10, P < 0.0001, t-value = 3.9903, df = 214.85, Cohen's d = 0.53) (Fig. 3a). The kinetics of RNA genome copies were largely similar between vaccinated and unvaccinated patients until 3 DPOS, with a faster decline for vaccinated patients starting at 4 DPOS (Fig. 3b). In contrast, infectious VLs were substantially lower in vaccinated patients at all DPOS, with the biggest effect at 3–5 DPOS (Fig. 3c). Still, at 5 DPOS, infectious virus was detectable in seven of 13 (53.8%) vaccinated patients and 11 of 13 (84.6%) unvaccinated patients. Additionally, 79 Delta-infected unvaccinated individuals were matched with Delta vaccine breakthrough patients in regard to age, sex and DPOS (Extended Data Fig. 4a). Infectious VLs were elevated in unvaccinated patients in comparison to vaccine breakthroughs (8.12-fold, 0.91 log10, P = 0.001, t-value = 3.5789, df = 35, Cohen's d = 0.60) (Extended Data Fig. 4c), confirming a significant reduction of infectious VLs among vaccinated patients. We further analyzed whether infectious VLs correlate with the time interval since administration of the last vaccine dose. A high heterogeneity between patient samples resulted in no significant correlation between the time after vaccination and infectious viral shedding (Extended Data Fig. 5a).

Fig. 3: RNA VL and infectious viral titers for unvaccinated versus vaccinated individuals infected with Delta.
figure 3

a, Genome copies (left) and infectious virus (right) for fully vaccinated and unvaccinated Delta-infected patients. Infectious titers (FFU > 0) were detected in 112 unvaccinated and 75 fully vaccinated Delta-infected patients; no titers (FFU = 0) were detected in 15 unvaccinated and 29 fully vaccinated Delta-infected patients. Error bars indicate mean ± s.d. Two-tailed t-test was used to determined differences of means. ***P = 0.0002 and ****P < 0.0001. Genome copies (b) and infectious VLs (c) measured for vaccinated and unvaccinated Delta-infected patients at different DPOS. The solid lines represent the fitted curve calculated using the LOESS method. Error bars represent 95% confidence bands of the best-fit line. LOESS, locally estimated scatterplot smoothing.

Booster vaccination leads to lower infectious VL in Omicron-infected individuals

Upon the emergence of Omicron BA.1, we analyzed the infectious viral shedding in unvaccinated, fully vaccinated and boosted individuals infected with this variant. We compared RNA and infectious VLs in NPS samples of 91 Omicron BA.1-infected patients and 62 Delta-infected patients who received two doses of vaccine more than 2 weeks before diagnosis. Because Omicron BA.1 can be titrated only on Vero E6-TMPRSS cells, we also titrated another set of samples from vaccinated Delta-infected patients on this cell line to assure comparability between infectious VLs. Omicron BA.1 breakthrough infections in fully vaccinated patients resulted in similar genome copies compared to Delta but significantly lower infectious VLs (14-fold, 1.146 log10, P < 0.0001, t-value = 5.3336, df = 120.2, Cohen's d = 0.90) (Fig. 4a). A significant reduction of infectious VLs was also observed for Omicron BA.1 samples when matching patients for age, sex and DPOS (29.9-fold, 1.476 log10, P = 0.00028, t-value = 4.1887, df = 26, Cohen's d = 0.81) (Extended Data Fig. 4d). Similarly to Delta-infected fully vaccinated individuals, the RNA VLs only slightly decreased over 5 DPOS, whereas infectious VLs declined toward 5 DPOS (Fig. 4b,c). Next, we evaluated whether the vaccination status—that is, unvaccinated, fully vaccinated or boosted—has an influence on RNA or infectious VLs for Omicron-infected individuals. We found no reduction of RNA or infectious VL in fully vaccinated individuals compared to unvaccinated individuals. However, a significantly lower infectious VL, but not RNA VL, was observed for boosted individuals (5.3-fold, 0.728 log10, adjusted P = 0.001325, t-value = 3.635, df = 71.237, Cohen's d = 0.64) (Fig. 4d) compared to fully vaccinated subjects. Similarly to Delta-infected fully vaccinated patients, no significant correlation was found between days after vaccination and infectious VL in fully vaccinated Omicron BA.1-infected patients (Extended Data Fig. 5b).

Fig. 4: SARS-CoV-2 infectious VLs in vaccine breakthrough infections with Omicron or Delta.
figure 4

a, Genome copies (left) and infectious virus (right) for fully vaccinated patients infected with Delta or Omicron BA.1 VOC. Infectious titers (FFU > 0) were detected in 53 Delta-infected and 66 Omicron BA.1-infected fully vaccinated patients; no titers (FFU = 0) were detected in nine Delta-infected and 25 Omicron BA.1-infected patients. Error bars indicate mean ± s.d. Two-tailed t-test was used to determined differences of means. ****P < 0.0001; NS, not significant. Genome copies (b) and infectious VLs (c) measured for fully vaccinated Omicron BA.1-infected and Delta-infected patients at different DPOS. The solid lines represent the fitted curve calculated using the LOESS method. Error bars represent 95% confidence bands of the best-fit line. d, Genome copies (left) and infectious virus (right) for unvaccinated, fully vaccinated or boosted Omicron BA.1 breakthrough cases. Infectious VLs for Delta and Omicron BA.1 samples were determined by FFA on Vero E6-TMPRSS cells. Infectious titers (FFU > 0) were detected in 24 unvaccinated, 66 fully vaccinated and 18 boosted Omicron BA.1-infected patients; no titers (FFU = 0) were detected in nine unvaccinated, 25 fully vaccinated and 12 boosted Omicron BA.1-infected patients. Error bars indicate mean ± s.d. Two-tailed t-tests were applied for each pair. P values were adjusted for multiple testing with the Benjamini–Hochberg method. NS, not significant; * adjusted P = 0.0384 and ** adjusted P = 0.001325. Genome copies (e) and infectious VLs (f) measured for fully vaccinated (two doses) and boosted (three doses) Omicron BA.1-infected patients at different DPOS. The solid lines represent the fitted curve calculated using the LOESS method. Error bars represent 95% confidence bands of the best-fit line. LOESS, locally estimated scatterplot smoothing.

Discussion

In this study, we analyzed virus shedding in COVID-19 patients infected with pre-VOC, Delta and Omicron BA.1 variants and evaluated the effect of vaccination on VL in the URT during the first 5 DPOS. To our knowledge, this is the first study to quantify infectious VLs in individuals infected with different SARS-CoV-2 variants and in vaccination breakthrough cases. We demonstrated a higher infectious VL in unvaccinated Delta-infected individuals compared to pre-VOC-infected individuals and showed a significant reduction of infectious VLs in fully vaccinated Delta-infected individuals. However, only booster vaccination significantly reduced infectious VL in Omicron BA.1-infected individuals. Furthermore, we found a lower infectious VL in Omicron BA.1 breakthrough cases than in Delta breakthrough cases.

The magnitude and timing of infectiousness of patients with COVID-19 is critical information necessary to make informed public health decisions on the duration of isolation of patients and on the need to quarantine contacts. Infectiousness is strongly influenced by VL in the URT of infected patients4. However, VL is often measured as RNA copy numbers and not actual infectious virus. In this study, we could show that RNA copy numbers in NPS samples poorly correlated with infectious virus shedding. This is in line with several other studies that found that RNA is a poor infectiousness indicator, especially in the presence of infection-induced neutralizing antibodies9,26. Nevertheless, in our study, correlation between RNA and infectious VL was equally low between fully vaccinated and unvaccinated Delta-infected patients, indicating that factors other than mucosal neutralizing antibodies may be important for the reduction in infectious VL. In addition, in an animal model, it was demonstrated that infectious virus, but not RNA, is a good proxy for transmission8.

Virus isolation in cell culture is widely used as a proxy for infectiousness6,9,28. Several studies have shown that isolation success significantly drops when RNA VLs are below 6 log10 copies per milliliter in NPS or when samples were collected after 8 DPOS6. Of note, with only a qualitative result, isolation success cannot distinguish between high and low infectious VLs in a patient sample, a key determinant of the potential size of the transmitted inoculum. Differences in infectious VL can affect transmission probability; therefore, we used an FFA that can reliably quantify infectious viral particles from NPSs. FFAs have long been a standard to quantify viral shedding in animal infection models for respiratory viruses, such as influenza, and have recently been used to quantify infectious VL in a SARS-CoV-2 human challenge trial, showing that they are considered as one of the best available proxies for infectiousness29,30,31. However, although we can assume that higher infectious VL leads to higher transmission risk, we currently do not know how many focus-forming units per milliliter (FFU ml−1) are required for a patient to actually transmit the virus.

Within 5 DPOS, we found higher RNA VLs but lower infectious VLs in swabs of unvaccinated patients with pre-VOC infections compared to Delta. These results disagree with other studies that analyzed only nucleic acid detection and found 3–10-fold-higher RNA copy number in Delta-infected patients compared to pre-VOC-infected patients15,32. However, these studies did not control for DPOS, age or sex. Other studies found either no difference in RNA VL between Delta and pre-VOC swabs33 or more than 1,000-fold-higher VL for Delta34, documenting the difficulty of comparing RNA VLs of virus variants during different phases of the pandemic, especially without additional information, such as DPOS. Conversely, in agreement with our results, a higher virus isolation success rate was observed for Delta compared to pre-VOC SARS-CoV-2 or Alpha35.

Vaccines have been shown to tremendously reduce symptomatic SARS-CoV-2 infections. However, vaccination’s effect on breakthrough case infectiousness is unclear. We show that infectious VL and RNA VL is reduced in fully vaccinated Delta patients during the first 5 DPOS. In this time period, approximately 50% of transmissions occur for pre-VOC strains5, indicating that reduced VL could considerably decrease the secondary attack rate. Other studies showed no difference in RNA VL between the vaccinated and unvaccinated early after symptom onset36,37 but found a lower virus isolation rate36. Conversely, another study detected up to ten-fold-reduced RNA VL in vaccinated patients but only for 60 days after full vaccination38. Similarly, two more studies reported decreased RNA VL for vaccine breakthrough infection with pre-VOC and Alpha SARS-CoV-2 (ref. 39) but no effect around 6 months after vaccination when Delta dominated40. Of note, we were still able to detect infectious viral particles in 53.8% of fully vaccinated Delta-infected individuals at 5 DPOS, indicating that shortening of the isolation period to 5 days, as recommended by the CDC, should be carefully evaluated41. Whether lower infectious VL translates into lower secondary attack rates remains controversial and depends on other influencing factors, such as environmental stability of virus particles. Several studies found a correlation between VL and secondary attack rate, with VL of the index case being the leading transmission correlate3,4. In agreement with these findings, epidemiological studies showed reduced transmission from vaccinated index cases, but the effect size depends on the prevalent variant, the vaccine used and the time since vaccination19. In contrast, another study found that the index case vaccination status did not influence the secondary attack rate21. Although VL is a key element of transmission, the process of human-to-human transmission is complex, and other factors, such as varying recommended protection measures, overall incidence, perceived risks and the context of contacts (household versus community transmission), can influence outcomes in the studies reported.

To date, few data exist on VL in vaccine breakthrough infections caused by Omicron, owing to its recent emergence in late November 2021. Reduced neutralization of Omicron by infection-derived and vaccine-derived antibodies was reported in vitro, but the effect was less pronounced for boosted individuals42,43. Furthermore, epidemiological studies show an increased risk of (re-)infection with Omicron in vaccinated and recovered individuals44 with high secondary attack rates among fully vaccinated and boosted individuals45,46,47. Higher RNA VLs as described in some studies were discussed as one potential contributing factor for the emergence of Alpha and Delta, although, for Delta, we could confirm this only for infectious VL in our data. Recent studies have shown that infection with Omicron caused shorter viral RNA shedding and lower peak viral RNA concentrations in comparison to Delta variant48,49. In contrast, other studies found a similar RNA VL for Omicron-infected and Delta-infected patients46,50. These finding are in line with our study, where only infectious VL, but not RNA VL, was significantly lower in Omicron BA.1 breakthrough cases compared to Delta breakthrough cases. In combination, these results indicate that the observed high transmissibility of Omicron BA.1 is not caused by elevated VLs, and the mechanism behind the higher transmissibility remains to be investigated. First, in vitro data hint toward alternative entry mechanisms as well as early replication peaks in cell culture51,52, but no clinical data for these exist so far. Our findings indicate that, with lower infectious VL, the higher transmissibility of Omicron BA.1 seems to be unrelated to an increased shedding of infectious viral particles in vaccinated individuals. Furthermore, we could show that, in the case of Omicron BA.1 breakthrough infections, only boosted individuals had lower infectious VL, but not RNA VL, compared to unvaccinated individuals. These findings are partially in agreement with a recent household transmission study from Denmark where both fully vaccinated and boosted primary cases showed reduced onward transmission46.

Our study has several limitations. We included only samples from symptomatic, but not asymptomatic, infected individuals that were collected ≤5 DPOS with Ct values <27. Therefore, absolute RNA copy numbers are biased toward higher VLs as patients with low VL were not included here. However, patients with low VL have likely little relevance in terms of transmission, and the fraction of patients with Ct values ≥27 was similar for all groups at all DPOS. Furthermore, our focus was on infectious virus shedding, and it has been shown that SARS-CoV-2 culture is unlikely to be successful from samples with higher Ct values24 and that the vast majority of secondary transmission occurs before 5 DPOS, although this requires assessment in Omicron cases5. Other factors, such as poor swab quality, can be a confounding factor leading to low VLs. Also, our results could be affected if the timing between peak VL and the observed onset of symptoms would be considerably different between the variants or between unvaccinated and vaccinated individuals. However, VL trajectories of variants and of vaccinated and unvaccinated individuals run in parallel, indicating that they largely follow similar kinetics. Also, we would like to emphasize that there is currently no agreed cutoff for FFU ml−1 above which a patient could reliably be classified as infectious. In addition, comparisons between variants—that is, between pre-VOC and Delta as well as between Delta and Omicron BA.1—might be affected by differences in adaption of variants to the cell lines used in this study. Lastly, we also would like to mention that almost all individuals in this study were vaccinated with mRNA vaccines that induce high titers of neutralizing antibodies in the blood but relatively low mucosal antibodies. Therefore, our results cannot be generalized to other vaccines—that is, those that are used mainly in low- and middle-income countries.

In conclusion, this study provides significant evidence for higher infectiousness of SARS-CoV-2 Delta as well as a significant effect of full vaccination on infectious VL and its speed of clearance. In addition, we show that Omicron BA.1 has lower infectious VLs compared to Delta in fully vaccinated individuals. Last, after Omicron BA.1 infection, lower infectious VL is observed only in boosted individuals. Our findings highlight the beneficial effect of vaccinations beyond the individual protection from severe disease and underscore the importance of booster vaccination. Thereby, we provide guidance for public health measures, such as shortening of the isolation period and vaccination certificates.

Methods

Participants

Ethical approval

The study was approved by the Cantonal Ethics Committee at the University Hospital of Geneva (CCER no. 2021-01488). All study participants and/or their legal guardians provided informed consent.

Sample collection and setting

NPSs collected from symptomatic (self-reported) individuals by trained professionals in the outpatient testing center of the Geneva University Hospital (HUG), for SARS-CoV-2 qRT–PCR diagnostics, were included in this study. Samples from asymptomatic individuals were not included. Infection with SARS-CoV-2 was diagnosed by qRT–PCR assay (Cobas 6800, Roche). For this study, samples were included between 7 April 2020 and 19 February 2022.

Only specimens with Ct values below 27 for the E-gene qRT–PCR diagnostic target were included in our analyses. All samples originated from the diagnostic unit of the hospital’s virology laboratory and were received for primary diagnosis of SARS-CoV-2. Remaining sample volume was stored at −80 °C, on the same day or within 24 hours. All samples had only one freeze–thaw cycle for the purpose of this study. All specimens from unvaccinated and vaccinated Delta-infected individuals were characterized by full genome sequencing or mutation-specific PCR for their infecting SARS-CoV-2 variant. Initial identification of Omicron BA.1 was done by S-gene target failure of the TaqPath COVID-19 assay (Thermo Fisher Scientific) and confirmed by partial Sanger sequencing of spike53, followed by next-generation sequencing. No sequence information was obtained for samples collected before the first detection of VOCs in Switzerland—that is, pre-VOC samples. Clinical information of the patients was collected by a standardized questionnaire in our testing center and/or through the Cantonal Health Service. The day of symptom onset was defined as day 0 in this study. Only specimens collected within the first 5 DPOS were selected for this study.

Viral load quantification by qRT–PCR

For initial inclusion of samples into this study, Ct values for the E-gene target of the diagnostic qRT–PCR (Cobas 6800, Roche), determined by the diagnostic laboratory of the HUG at the time of sampling, were used. Afterwards, to minimize variability between measurements, all selected samples were re-extracted after thawing, and RNA VL in each sample was determined by E-gene qRT–PCR using SuperScript III Platinum One-Step qRT–PCR Kit (Invitrogen) in our research laboratory. Quantification of genome copy numbers was performed using an in vitro transcribed RNA standard for the E-gene assay as described previously28. Only results obtained from this latter measurement were used for further analysis.

Quantification of SARS-CoV-2 by FFA

Vero E6 and Vero E6-TMPRSS cells were cultured in complete DMEM GlutaMAX I medium supplemented with 10% FBS, 1× non-essential amino acids and 1% antibiotics (penicillin–streptomycin) (all reagents from Gibco). Vero E6-TMPRSS cells were kindly received from the National Institute for Biological Standards and Controls (cat. no. 100978). All infection experiments were performed under Biosafety Level 3 conditions.

FFAs used in this study were adapted from published protocols54. NPS samples were serially diluted and applied on a monolayer of Vero E6 cells in duplicates. After 1 hour at 37 °C, the media were removed, and pre-warmed medium mixed with 2.4% Avicel (DuPont) at a 1:1 ratio was overlaid. Plates were incubated at 37 °C for 24 hours and then fixed using 6% paraformaldehyde for 1 hour at room temperature. Cells were permeabilized with 0.1% Triton X-100 and blocked with 1% BSA (Sigma-Aldrich). Plates were incubated with a primary monoclonal antibody targeting SARS-CoV-2 nucleocapsid protein (Geneva Antibody Facility, JS02, diluted to 0.2 µg ml−1) for 1 hour at room temperature and then with peroxidase-conjugated secondary antibody (Jackson ImmunoResearch, 109-036-09, diluted to 1:2,000) for 30 minutes at room temperature. Foci were visualized using True Blue HRP substrate (Avantor) and imaged on an ELISPOT reader (CTL). We defined a cluster of adjacent cells expressing viral antigen as a foci. Foci were counted and expressed as FFU ml−1. FFAs for comparison of infectious VLs in Delta versus Omicron were performed in Vero E6-TMPRSS cells using the same protocol.

Virus isolation

NPSs were applied on Vero E6 cell monolayers in 24-well plates. Next, 100 µl of each sample was added and incubated for 1 hour at 37 °C. After the incubation, the infectious supernatant was discarded, and virus culture medium was added. Then, 50 µl of the medium was collected to determine VL by qRT–PCR as described above at day 0. Then, 3–4 days after inoculation, the medium was replaced, and 6 days after infection, the infectious medium was collected to determine VL. A genome copy number change of at least 1 log of from day 0 to day 6 indicated a successful isolation.

Statistical analysis

Data collection was done using Excel 2019. All statistical analyses were performed using R statistical software version 4.1.1 (Foundation for Statistical 185 Computing) and Prism version 9.3.1 (GraphPad). All FFU and RNA genome copies were log10 transformed, and samples with no detectable FFU were set to 1 FFU ml−1 for the purpose of analysis. Cohen's d effect size is interpreted as follows: small = 0.2 - <0.5; medium = 0.5 - <0.8; large = ≥0.8

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.