prepared for the Coalition for Epidemic Preparedness Innovations
(CEPI) posits that technologies with ‘‘no licensure track-record,”
have a substantially higher risk of failure than ‘‘well-established”
technologies [22]. This is consistent with the suggestion that the
high failure rate of vaccines in development prior to 2013 was
due, in part, to the large number of candidate vaccines using unpro-
ven, nucleotide (DNA or RNA) technologies [13].
Research on the basic science and technologies underlying new
drugs or vaccines, and the maturation of these technologies to the
point that they can support efficient development, is funded pri-
marily by the public sector, principally governments. We have pre-
viously examined the scale of NIH funding for the basic research
and technologies underlying new drug approvals by identifying
NIH funding cited in published research papers [21,23]. These
studies show that the NIH contributed more than $170 billion for
research related to the 356 drugs approved from 2010–2019 or
their drug targets, with more than 85% of this funding involving
research on the drug targets rather than the drugs themselves.
We have also observed that the majority of this research is funded
through investigator-initiated research projects [21].
The present work examines the maturation of research on the
technologies being used in candidate COVID-19 vaccines as well
as the NIH funding supporting this research over the past twenty
years. Specifically, we examine published research on vaccines
incorporating attenuated or inactivated viruses, synthetic (recom-
binant) proteins, DNA or mRNA, or recombinant viral vectors as
well as research on formulations including conventional adjuvants,
virus-like particles, nanoparticles and toll-like receptor 9 (TLR9)
agonists. We also examine research and NIH funding for vaccines
aimed at coronaviruses and three unrelated viral pathogens that
have been associated with epidemic transmission: Zika, Ebola,
and dengue. We consider the impact of this prior research on accel-
erated efforts to develop a vaccine for COVID-19, and the impor-
tance of sustained public-sector funding in establishing a
foundation for responding to pandemic outbreaks.
2. Materials and methods
Technologies utilized in candidate vaccines against COVID-19
were identified from the World Health Organization (WHO)
‘‘DRAFT Landscape of Candidate COVID-19 vaccines” [24]. Viruses
with epidemic potential were identified from Plotkin [25] or the
WHO Blueprint [26].
Searches were performed using PubMed (www.pubmed.gov;
accessed June 3, 2020), with the updated Automatic Term Mapping
(May, 2020) optimized with Medical Subject Headings (MeSH)
terms or Boolean modifiers to increase specificity (Supplemental
Table 1). PubMed Identifiers (PMIDs) with their respective publica-
tion year were recorded.
PMIDs acknowledging NIH funding were identified in NIH
RePORTER data tables (https://exporter.nih.gov/ExPORTER_Cata-
log.aspx; accessed June 3, 2020) as described previously [23]. Each
PMID was associated with a project year corresponding to the pro-
ject number and year of publication using the ‘‘Link Tables for Pro-
ject to Publication Associations.” Costs (since 2000) were derived
from the ‘‘Project Data Table.” Publications occurring before the
first year of the grant award or more than four years after the last
year of the grant were excluded. Publications 1–4 years after the
last year of the grant were associated with the project costs of
the last year. All values are described after eliminating duplicate
PMIDs, NIH-funded PMIDs, project years, and project costs arising
from the identification of PMIDs in multiple searches, multiple
sources of funding for some PMIDs, and multiple PMIDs related
to many projects. ‘‘Unique” values across technologies are
described after eliminating duplicates across technologies. Activity
codes and the funding institute were determined from the project
codes. Grant categories were derived from ‘‘NIH Types of Grant
Programs 2020” (https://grants.nih.gov/grants/funding/funding_
program.htm; accessed May, 2020). Costs are given in constant
dollars inflation-adjusted to 2018 using the U.S. Bureau of Labor
Statistics All Urban Consumer Prices (Current Series) (https://
www.bls.gov/data/; accessed May, 2020). All analyses were per-
formed in PostgreSQL and Excel.
The bibliographic Technology Innovation Maturation Evaluation
(TIME) model assesses the maturation of a body of published
research by modeling the rate of research accumulation, as
described previously [20]
. The model quantifies the ‘‘S-curve” of
technology maturation described in other sectors [14,15,18,19].
The TIME model fits an exponentiated logistic function to the accu-
mulation of PubMed identified publications over time. The equa-
tion has the form:
N ¼ L
1
1þe
rðtt
0
Þ
where N is the cumulative number of publications, L is the upper
limit of publications, r is the growth rate, t is time, and t
0
is the mid-
point of exponential growth. This asymmetric, sigmoidal function
exhibits the common logistic sigmoid function (‘‘S-curve”) over
log scales. The established (Te) point is defined as the point of min-
imum acceleration or logN’’(t)
min
. Results are visualized as annual
publications, cumulative publications, or log cumulative publica-
tions to assess the suitability of the calculated curve fit.
3. Results
3.1. Research on vaccine technologies
As of July 31, 2020, the WHO listed 165 candidate vaccines
against COVID-19 [24]. PubMed searches were performed for ten
of the technologies utilized in this portfolio of products. The ten
technologies and the number of publications through 2019 are
shown in Table 1. The time course of publications on the ten tech-
nologies together is shown in Fig. 1A. The time course of publica-
tion on individual technologies is shown in the interactive
graphic at https://tabsoft.co/31EkYeK.
Technologies involving whole virus preparations, including live
attenuated and inactivated viral vaccines, have been widely used in
vaccines for polio, influenza, MMR, and other products since the
1950s. Research on these technologies has continued to accumu-
late since 1980 without evident acceleration or deceleration
(Fig. 1A). These technologies were used in four of the first ten can-
didate vaccines to enter clinical trials against COVID-19, but only
12 (7.3%) candidate vaccines through July 31, 2020.
Synthetic vaccines incorporating recombinant proteins
emerged with biotechnology in the 1980s. Exponential growth of
research on synthetic vaccines was evident after 1985, but growth
has slowed since the late 1990s (Fig. 1A). While only one of the first
ten candidate COVID-19 vaccines to enter clinical trials employed
synthetic vaccine technologies, they are used in 66 (40%) candidate
vaccines through July 31, 2020.
Vaccines employing recombinant viral vectors emerged in the
mid-1990s as an outgrowth of research on gene therapy. A period
of exponential growth was evident in the 1990s, followed by slow-
ing from the early 2000s to the present (Fig. 1A). Two of the first
ten candidate COVID-19 vaccines in clinical trials used recombi-
nant adenoviral vectors. Overall, recombinant viral vectors are
employed in 31 (19%) candidate vaccines through July 2020.
DNA-based vaccine technologies also emerged in the mid-
1990s as an outgrowth of research on gene therapy and exhibited
a similar pattern of exponential advance and slowing. A related
A.E. Kiszewski, Ekaterina Galkina Cleary, M.J. Jackson et al.
Vaccine 39 (2021) 2458–2466
2459