PNAS Nexus
, 2023,
2
, 1–16
https://doi.org/10.1093/pnasnexus/pgad033
Advance access publication 14 March 2023
Research Report
Extreme differences in SARS-CoV-2 viral loads
among respiratory specimen types during presumed
pre-infectious and infectious periods
Alexander Viloria Winnett
a
,1
, Reid Akana
a
,1
, Natasha Shelby
a
,1
, Hannah Davich
a
, Saharai Caldera
a
, Taikun Yamada
b,c
,
John Raymond B. Reyna
b
, Anna E. Romano
a
, Alyssa M. Carter
a
, Mi Kyung Kim
a
, Matt Thomson
a
, Colten Tognazzini
d
,
Matthew Feaster
d
, Ying-Ying Goh
d
, Yap Ching Chew
b
,c
and Rustem F. Ismagilov
a
,
*
a
California Institute of Technology, 1200 E. California Blvd, Pasadena, CA 91125, USA
b
Pangea Laboratory LLC, 14762 Bentley Cir, Tustin, CA 92780, USA
c
Zymo Research Corp., 17062 Murphy Ave, Irvine, CA 92614, USA
d
Pasadena Public Health Department, 1845 N. Fair Oaks Ave, Pasadena, CA 91103, USA
*To whom correspondence should be addressed: E-mail:
rustem.admin@caltech.edu
1
A.V.W., R.A., and N.S. authors contributed equally to this report. Order of co-first authorship was determined by extent of contributions; see detailed Author
Contributions statement in the supplement.
Edited By:
Amalio Telenti
Abstract
SARS-CoV-2 viral-load measurements from a single-specimen type are used to establish diagnostic strategies, interpret clinical-trial
results for vaccines and therapeutics, model viral transmission, and understand virus–host interactions. However, measurements
from a single-specimen type are implicitly assumed to be representative of other specimen types. We quantified viral-load
timecourses from individuals who began daily self-sampling of saliva, anterior-nares (nasal), and oropharyngeal (throat) swabs before
or at the incidence of infection with the Omicron variant. Viral loads in different specimen types from the same person at the same
timepoint exhibited extreme differences, up to 10
9
copies/mL. These differences were not due to variation in sample self-collection,
which was consistent. For most individuals, longitudinal viral-load timecourses in different specimen types did not correlate. Throat-
swab and saliva viral loads began to rise as many as 7 days earlier than nasal-swab viral loads in most individuals, leading to very low
clinical sensitivity of nasal swabs during the first days of infection. Individuals frequently exhibited presumably infectious viral loads
in one specimen type while viral loads were low or undetectable in other specimen types. Therefore, defining an individual as
infectious based on assessment of a single-specimen type underestimates the infectious period, and overestimates the ability of that
specimen type to detect infectious individuals. For diagnostic COVID-19 testing, these three single-specimen types have low clinical
sensitivity, whereas a combined throat–nasal swab, and assays with high analytical sensitivity, was inferred to have significantly
better clinical sensitivity to detect presumed pre-infectious and infectious individuals.
Keywords:
COVID-19, testing strategies, viral loads
Significance Statement
In a longitudinal study of SARS-CoV-2 Omicron viral loads in three paired specimen types (saliva, anterior-nares swabs, and oropha
-
ryngeal swabs), we found extreme differences among paired specimen types collected from a person at the same timepoint, and that
viral loads in different specimen types from the same person often do not correlate throughout infection. Individuals often exhibited
high, presumably infectious viral loads in oral specimen types before nasal viral loads remained low or even undetectable.
Combination nasal–throat swabs were inferred to have superior clinical sensitivity to detect infected and infectious individuals.
This demonstrates that single-specimen type reference standard tests for SARS-CoV-2, such as in clinical trials or diagnostics eval
-
uations may miss infected and even infectious individuals.
Competing interest:
R.F.I. is a co-founder, consultant, and director for and has stock ownership in Talis Biomedical Corporation. All
other authors report no potential conflicts.
Received:
November 22, 2022.
Revised:
January 18, 2023.
Accepted:
January 24, 2023
© The Author(s) 2023. Published by Oxford University Press on behalf of National Academy of Sciences. This is an Open Access article
distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits
unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
Introduction
Measurements of viral load in respiratory infections are used to
establish diagnostic strategies, interpret results of clinical trials
of vaccines and therapeutics, model viral transmission, and
understand virus–host interactions. But how viral loads change
across multiple specimen types early in SARS-CoV-2 infection is
not well understood. Specifically in the context of diagnostics, as
new SARS-CoV-2 variants-of-concern (and new respiratory
2 |
PNAS Nexus
, 2023, Vol. 2, No. 3
viruses) emerge with different viral kinetics (1
), it is imperative to
continually re-evaluate testing strategies (including specimen
type and test analytical sensitivity) for detecting pre-infectious
and infectious individuals. Early detection can reduce transmis
-
sion within communities (
2, 3
) and the global spread of new var
-
iants, and enable earlier initiation of treatment resulting in
better outcomes (
4
–6).
Selecting testing strategies to achieve detection in the pre-
infectious and infectious periods requires filling two critical
knowledge gaps: (i) Which respiratory specimen type accumulates
virus first? (ii) What is the appropriate test analytical sensitivity to
detect accumulation of virus in the pre-infectious and infectious
stages? These two gaps must be filled in parallel. Commonly, an
individual’s infection is described by the viral load sampled
from a single-specimen type, which is appropriate when there is
one principal specimen type (e.g. HIV in blood plasma).
However, some respiratory pathogens, including SARS-CoV-2,
can infect multiple respiratory sampling sites (7–9).
Nasopharyngeal swabs have been the gold standard for
SARS-CoV-2 detection but are poorly tolerated and challenging
for serial sampling and self-collection. Many alternate specimen
types are now widely used. Some of these are suitable for routine
testing, and are approved for self-collection (e.g. saliva, anterior-
nares [nasal] swabs, and oropharyngeal [throat] swabs) in some
countries. Cross-sectional studies comparing paired specimen
types from the same person have shown that cycle threshold
(Ct, a semi-quantitative proxy for viral load) values can differ sub
-
stantially between specimen types (
10), and the clinical sensitivity
of different specimen types is not equivalent (11
). Sometimes, vi
-
ral loads in one specimen type are low or even absent while viral
loads in another type are high (
12
–14). Nasal swabs (including
those used for rapid antigen testing) are the dominant specimen
type used in the United States for workplace screenings and
at-home testing. However, several studies (15–18), news media
(19
), and social-media posts have speculated that in Omicron in
-
fections, viral load accumulates in oral specimens before the na
-
sal cavity. Formal investigations of specimen types from single
timepoints and cross-sectional studies have been contradictory,
potentially due to when individuals were sampled; viral loads
from individuals sampled after symptom onset may not reflect vi
-
ral loads from earlier in the infection. Rigorous, longitudinal com
-
parisons of paired specimen types starting from the incidence of
infection are needed to fill this gap.
The second knowledge gap is the analytical sensitivity needed for
reliable detection of pre-infectious and infectious individuals. The
assay analytical sensitivity is described by the limit of detection
(LOD); generally, the LOD of an assay describes its ability to detect
and quantify target at or above a certain concentration in that spe
-
cimen type with
>
95% probability (
20
). Assays with high LODs (low-
analytical sensitivity) require a high concentration of virus to reli
-
ably yield positive results, whereas assays with low LODs (high ana
-
lytical sensitivity) can reliably detect much lower concentrations of
virus. For example, in early SARS-CoV-2 variants, some studies
showed that saliva accumulated virus earlier than nasal swabs,
but at low levels (
14, 21, 22), thus saliva required a
high-analytical-sensitivity (low LOD) assay (14,
23
). However,
low-analytical-sensitivity tests (including rapid antigen tests) are in
-
creasingly authorized and used globally (
24, 25
). Which of these tests
can detect pre-infectious and infectious individuals requires quanti
-
tative, longitudinal measurements of viral concentration in multiple
specimen types starting from the incidence of infection.
Early detection, in the pre-infectious period, is ideal to prompt
infection-control practices (e.g. isolation) before transmission
occurs, and detection during the infectious period is critical to
minimize outbreaks. Replication-competent (i.e. infectious) virus
has been recovered from saliva (9), oropharyngeal swabs (26), and
nasal swabs (27
), but it is impractical and infeasible to perform vi
-
ral culture on each positive specimen to determine if a person is
infectious. However, studies that performed both culture and
RT-qPCR found that low Ct values (high viral loads) are associated
with infectious virus. Specific viral loads likely to be infectious for
each specimen type have not been established (
28), partly because
Ct values are not comparable across assays (29,
30) and culture
methods differ. However, as a general reference, viral loads of
>
10
4
–10
7
RNA copies/mL are associated with the presence of
replication-competent virus (17,
31–41), and these values have
been used in outbreak simulations (35,
39, 42–44
). The enormous
range (
>
4 orders of magnitude) in observed viral loads that corres
-
pond with infectiousness emphasizes why quantitative measure
-
ments of loads in different specimen types are needed to make
robust predictions about tests that will detect the pre-infectious
and infectious periods.
The assumption made early in the COVID-19 pandemic that vi
-
ral load always rises rapidly from undetectable to likely infectious
(
45) has been challenged by numerous longitudinal studies of viral
load in different specimen types that show early SARS-CoV-2 viral
loads can rise slowly over days (14,
17, 18, 21, 27, 41, 46
–49
), not
hours. These findings are encouraging because a longer window
provides more time to identify and isolate pre-infectious individu
-
als. However, making use of this opportunity by selecting an opti
-
mal diagnostic test requires a thorough understanding of how
viral load changes in each specimen type early in infection.
Moreover, to reliably detect an infectious person, the infectious
specimen must be tested with an assay that has an LOD below
the infectious viral load for that specimen type. However, many au
-
thorized COVID-19 tests (including rapid antigen tests) have LODs
well above the range of reported infectious viral loads (
50, 51).
Filling the two critical and inter-related knowledge gaps about
specimen type and assay LOD requires high-frequency quantifica
-
tion of viral loads, rather than semi-quantitative Ct values, in
multiple specimen types starting from the incidence of infection,
not after a positive test or after symptom onset, as is commonly
done. Moreover, quantification must be performed with a
high-analytical-sensitivity assay to capture low viral loads in the
first days of detectable infection. It is challenging to acquire
such data. Individuals at high risk of infection must be prospect
-
ively enrolled prior to detectable infection and tested longitudin
-
ally with high-frequency in multiple paired specimen types.
To our knowledge, four studies have reported longitudinal
viral-load timecourses in multiple, paired specimen types from
early infection. A university study (27
) captured daily saliva and
nasal-swab samples for 2 weeks from 60 individuals, only 3 of
whom were negative for SARS-CoV-2 upon enrollment. In our pri
-
or study, we captured twice-daily viral-load timecourses from 72
individuals for 2 weeks (
52
), 7 of whom were negative upon enroll
-
ment (
14). In six of seven individuals, we inferred from viral-load
quantifications that a high-analytical-sensitivity saliva assay
would detect infections earlier than a low-analytical-sensitivity
nasal-swab test. In a SARS-CoV-2 human challenge study (17
),
10 of 18 infected participants had detectable virus by PCR in throat
swabs at least 1 day prior to nasal swabs, and replication-
competent virus was recovered from throat swabs before nasal
swabs in at least 12 of 18 participants. Participants in these three
studies were infected with pre-Omicron variants. One longitudin
-
al study (
15) analyzed viral loads in saliva, nasal swabs, and throat
swabs in Omicron; however, daily measurements in all three
Viloria Winnett et al.
| 3
specimen types were captured for only two individuals, both of
whom were already positive upon enrollment. A separate
case-ascertained household-transmission study with prospective
daily sampling also captured viral-load measurements from the
incidence of infection using a combination nasal–throat swab spe
-
cimen type (
41
). In the United Kingdom, where this study was per
-
formed, a combination nasal–throat swab specimen type is
regularly used for diagnostic testing (
53, 54). However, the rise
and fall of Omicron viral loads in multiple paired single-specimen
types from the incidence of infection has not been characterized,
despite these data being necessary to define the appropriate test
analytical sensitivity and specimen type to best detect pre-
infectious and infectious individuals.
Here, we measured and analyzed the viral-load timecourses of
the Omicron variant in three specimen types appropriate for self-
sampling (saliva, nasal swabs, and throat swabs) by individuals
starting at or before the incidence of infection as part of a
household-transmission study in Southern California. We then
utilized these data to determine which specimen type and analyt
-
ical sensitivity would yield the most reliable detection of pre-
infectious and infectious individuals. A separate paper reports
the results of daily rapid antigen testing in this study (
55).
Materials and methods
Study design
This case-ascertained study of household transmission (approved
under Caltech IRB #20-1026) was conducted in the greater Los
Angeles County area between November 23, 2021, and March 1,
2022. All adult participants provided written informed consent; all
minor participants provided verbal assent accompanied by written
permission from a legal guardian. Children aged 8–17 years old add
-
itionally provided written assent. See Supplemental Information for
details.
A total of 228 participants from 56 households were enrolled; 90
of whom tested positive for SARS-CoV-2 infection during enroll
-
ment (Fig.
1). We limited our analyses to 14 individuals (Tables
S1 and S5, Fig.
2
) who enrolled in the study at or before the inci
-
dence of acute SARS-CoV-2 infection. To be included in the cohort,
a participant must have had at least one specimen type with viral
loads below quantification upon enrollment, followed by positiv
-
ity and quantifiable viral loads in all three specimen types.
Each day, participants reported symptoms, then self-collected
saliva, anterior-nares (nasal) swab, and posterior oropharyngeal
(hereafter throat) swab specimens for RT-qPCR testing in Zymo
Research SafeCollect devices (CE-marked for EU use), following
manufacturer’s instructions (56,
57
). Participants collected speci
-
mens immediately upon enrollment, then daily upon waking, as
morning sample collection has been shown to yield higher viral
loads than evening collection (
52).
RT-qPCR testing for SARS-CoV-2
Extraction and RT-qPCR were performed at Pangea Laboratories
(Tustin, CA, USA) using the FDA-authorized
Quick
SARS-CoV-2
RT-qPCR kit, with results assigned per manufacturer criteria
(58). Additional details in Supplemental Information. This assay
has a reported LOD of 250 copies/mL of sample.
Quantification of viral load from RT-qPCR result
To quantify viral load in RT-qPCR specimens, contrived specimens
across a 13-point standard curve (dynamic range from 250 to
4.50
×
10
8
copies/mL) for each specimen type was generated at
Caltech and underwent extraction and RT-qPCR as described
above. All three replicates at 250 copies/mL of specimen were de
-
tected, independently validating the reported LOD for the assay.
For each specimen type, the standard curve generated an equa
-
tion to convert from SARS-CoV-2
N
gene Ct values to viral loads
in genomic copy equivalents (hereafter copies) per mL of each spe
-
cimen type. See Supplemental Information for additional details
and equations. Positive specimens with viral loads that would be
quantified below the assay LOD were considered not quantifiable.
Viral sequencing and lineage/variant
determination
Viral sequencing of at least one specimen for each participant
with incident infection was performed on nasal or throat speci
-
mens with moderate to high viral loads by Zymo Research at
Pangea Lab. See Supplemental Information for details.
Defining pre-infectious and infectious periods
The pre-infectious period is all SARS-CoV-2-positive timepoints
prior to the first timepoint in which any specimen type contains
viral load greater than the indicated infectious viral-load thresh
-
old. There are three main methods for defining the infectious pe
-
riod for an individual based on viral loads. First, the infectious
@ssessedYforYeligibilityYg]YZY0-40fz
–nrolledYgnYZY22Lz
–xcludedYfromY
analysesYg]YZYU5z
RT-qPCR posive in all 3
specimen types on
enrollment (n = 64)
Enrolled late in infecon
(n = 13)
*Withdrew (n = 2)
Excluded from
enrollment
(N = 3,703)
Did not meet
inclusion criteria
(n = 3,636)
Declined to parcipate
(n= 67)
Enrolled before
or at the
incidence of
infecon
(n = 14)
Uninfected
(n = 138)
SARS-CoV-2 infected (n = 90)
Fig. 1.
A CONSORT diagram shows participant recruitment, eligibility,
enrollment, and selection for inclusion in the study cohort.