To the Editor,
With great interest, we read the recent article by Dhangadamajhi et al. [1], which
identified the possible association of TLR3 exonic variant (rs3775291) with severe
acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and mortality rates
in global populations. The authors included TLR3 rs3775291 polymorphism data of 48,835
healthy individuals from 40 countries and detected an important correlation between
the rs3775291 minor allele and susceptibility to SARS-CoV-2 infection and mortality.
An earlier study [2] used minor allele frequency data from 14 different countries
to show a similar correlation between the rs3775291 variant and COVID-19 susceptibility
and mortality. Although the data investigation and reporting were done elegantly and
scholarly, we found a few minor issues that need to be discussed further.
To draw a firm conclusion in a population-scale analysis, all existing studies must
be included. Genotype data from 48,835 healthy controls from 40 countries were used
by the authors [1]. The minor allele frequency was obtained from 1000 genome projects
and gnomAD, and other databases, such as PubMed and google scholar. After searching
various databases (1000 genomes, gnomAD, dbSNP, PubMed and google scholar) for minor
allele frequency, we found more reports from various populations [Tunisia (n = 5),
Brazil (n = 2), China (n = 13), South Korea (n = 4), Japan (n = 4), Scotland (n = 2),
Spain (n = 3), Denmark (n = 3), Germany (n = 4), Poland (n = 2), Italy (n = 2), Finland
(n = 3), and India (n = 5)]. When compared to the included reports of Dhangadamajhi
et al. [1], we were unable to trace allele frequency data from the Bulgarian population,
and a smaller number of studies from the USA (n = 4) and Sweden cohorts (n = 2) were
identified.
For population-scale correlation analysis, excluding reports that do not obey Hardy–Weinberg
equilibrium (HWE) is critical. In line with this, Dhangadamajhi et al., proposed that
such reports be removed from the correlation analysis. However, they erroneously included
genotype data from Barbados (χ
2 = 4.536, p = 0.033) and Bangladesh (χ
2 = 3.775, p = 0.052) populations, which were deviated or very close to the HWE deviation
score.
Despite the authors' claim that genotype data from 40 countries were used in the study,
the number of countries considered for the population-scale analysis was actually
39. For the correlation study, the authors used minor allele frequency data from Finland
twice. The minor allele data of Finland must be pooled before the correlation analysis.
The Pearson correlation test was used to assess the relationship between the prevalence
of minor allele ‘T’ and the SARS-CoV-2 infection and mortality rate per million subjects
in different populations. The Spearman rank correlation coefficient would be the most
suitable [3] to test the relationship between TLR3 variant and COVID-19 since the
two variables were on different scales and the analysis was not conducted in SARS-CoV-2
infected cases. Using data from Dhangadamajhi et al. [1], a Spearman rank correlation
study showed no significant association between SARS-CoV-2 and the TLR-3 rs3775291
polymorphism (infection: r = 0.244, p = 0.128; mortality: r = 0.247, p = 0.124).
For obtaining minor allele frequency in different populations, the authors used two
different search strategies: (1) genomic databases, such as 1000 Genomes Project and
gnomAD, and (2) literature databases, such as PubMed and Google Scholar. Although
allele frequency and the total number of healthy subjects considered for MAF calculation
have been mentioned in the manuscript’s supplementary table, a piece of additional
information on references and data sources would be more beneficial for the researchers.
On 18 January 2020, the authors collected SARS-CoV-2 data from various countries,
including infections, mortality, and recovery rates. The first cases of SARS-CoV-2
infection were identified in Wuhan, China, in December 2019, and the World Health
Organization declared COVID-19 a pandemic on 11 March 2020. The date listed in the
paper and the supplementary Table 1 (18th January 2020) may be a typographical error.
The infection and mortality status of SARS-CoV-2 on 18 January 2021 were obtained
from Dhangadamajhi et al. supplementary dataset. Data on minor allele frequency were
gathered from a variety of databases, as shown in Table 1. Reports with HWE deviated
genotype distributions were omitted from the present analysis [Barbados (n = 1), Bangladesh
(n = 1), China (n = 2), India (n = 1), Lithuania (n = 2), Nigeria (n = 1)] and a total
of 47,136 healthy subjects from 35 different populations were taken into account.
Using the modified minor allele frequency data, a reanalysis of the association between
rs3775291 minor allele frequency and COVID-19 showed no significant correlation between
rs3775291 minor allele ‘T’ and SARS-CoV-2 infection (Spearman r = 0.181, p = 0.295,
n = 35) or mortality (Spearman r = 0.146, p = 0.402). Up-to-date data of SARS-CoV-2
infection and mortality rate per million were obtained from the Worldometer website
(assessed on 1st April 2021). The spearman rank correlation study of rs3775291 minor
allele frequency (T) with SARS-CoV-2 infection rate (Spearman r = 0.212, p = 0.221,
n = 35) and mortality rate (Spearman r = 0.143, p = 0.412, n = 35) also failed to
show a potential association of rs3775291 polymorphism with COVID-19, bolstering the
absence of an association between rs3775291 and related mortality. However, case–control
studies in different populations are needed to confirm our findings.
Table 1
TLR3 rs3775291 minor allele frequency and SARS-CoV-2 related data of different countries
Population
Dhangadamajhi et al. [1]
Present study
Data assessed on 18th January 2021 (Dhangadamajhi et al. [1])
Data assessed on 1st April 2021
Data sources/remarks
Number of studies
MAF
Total cases
Number of studies
MAF
Total number of Healthy controls
SARS CoV-2 infection /million
SARS-CoV-2 death/million
SARS CoV-2 infection /million
SARS-CoV-2 death/million
Barbados
1
4.2
96
1
4.2
96
3603
24
12,697
146
1000 Genomes, this study was excluded for deviation of genotypes distribution from
HWE
Nigeria
2
0.7
207
1
0.926
108
514
7
776
10
1000 Genomes
Gambia
1
1.8
113
1
1.8
113
1589
52
2213
67
1000 Genomes
Luhya, Kenya
1
3.5
99
1
3.5
99
1821
32
2453
39
1000 Genomes
Sierra Leone
1
1.2
85
1
1.2
85
367
10
492
10
1000 Genomes
Tunisia
1
16.45
158
5
12.0
532
14,729
465
21,327
740
Moumad et al., 2013, Abida et al., 2020
Morocco
1
17.64
204
1
17.7
204
12,319
212
13,323
237
Moumad et al., 2013
USA
7
29.73
3677
4
29.4
3105
72,592
1210
93,747
1700
Dhiman et al., 2008, Edwards et al., 2008, Slattery et al., 2012, Resler et al., 2013
Colombia
2
27.88
459
2
27.9
459
36,544
935
46,920
1237
1000 Genomes, Allikmets et al., 2009
Peru
1
35.3
85
1
35.3
85
31,789
1164
46,491
1561
1000 Genomes
Brazil
1
33.1
299
2
35.4
760
39,340
976
59,682
1506
Assmann et al., 2014, Sa et al., 2015
Nicaragua
1
21.21
132
1
21.2
132
923
25
999
27
Lundkvist et al., 2020
China
8
31.86
3361
13
27.3
5089
61
3
63
3
1000 Genomes, Li et al., 2017, Pang et al., 2014, Chen et al., 2015, Cheng et al.,
2014., Ye et al., 2020, Rong et al., 2013,Chen et al., 2017,Wang et al., 2015, Rong
et al., 2013
South Korea
2
25.85
756
4
29.9
5114
1400
24
2020
34
Korean Genome Projec, KOREAN population from KRGDB, Cho et al., 2017, Hwang et al.,
2009
Taiwan
3
35.31
2281
3
34.6
1713
35
0.3
43
0.4
Yang et al., 2013,Yang et al., 2014
Japan
3
27.27
264
4
32.6
422
2449
34
3741
72
1000 Genomes, Ueta et al., 2007, Ikezoe et al., 2015, Matsuo et al., 2016
Vietnam
1
38.9
99
1
38.9
99
16
0.4
27
0.4
1000 Genomes
Finland
1
33.3
99
DNA
DNA
DNA
7231
111
13,692
152
–
Scotland, UK
1
32.4
91
2
30.3
249
30,330
990
63,766
1859
1000 Genomes, Dwyer et al., 2013,
Spain
2
30.40
472
3
30.1
430
48,160
1140
70,226
1613
1000 Genomes, Sironi et al., 2012, Matas-Cobos et al., 2015
Denmark
2
28.35
1280
3
27.9
1696
32,430
301
39,708
417
Laska et al., 2014, Enevold et al., 2014, Laska et al., 2014
Germany
2
28.96
1034
4
28.8
1291
24,132
555
33,701
917
Yang et al., 2012, Gast et al., 2011,Allikmets et al., 2009,Ye et al., 2020
Poland
1
25.64
78
2
27.0
150
37,796
878
61,396
1403
Studzińska et al., 2017, Grygorczuk et al., 2017
Ireland
1
26.61
263
1
26.2
263
33,527
511
47,371
941
Cooke et al., 2018
Lithuania
1
34.0
135
61,701
894
80,231
1327
Two studies were excluded for HWE deviation
Russia
1
32.6
269
1
34.6
269
24,283
446
31,135
677
Barkhash et al., 2013
Sweden
3
30.43
14,186
2
30.1
1109
51,659
1019
79,328
1327
Günaydın et al., 2014, Svensson et al., 2012
Iceland
1
27.5
169
1
27.5
169
17,393
85
18,096
85
Allikmets et al., 2009
Netherland
2
29.53
1107
2
29.5
1107
52,560
750
74,148
964
Allikmets et al., 2009
Serbia
1
33.17
104
1
33.1
104
42,420
425
68,947
609
Stanimirovic et al., 2013
Italy
1
30.8
107
2
27.1
345
38,939
1346
59,357
1811
1000 Genomes, Sironi et al., 2012
Finland
1
32.68
12,549
3
32.4
16,110
7231
111
13,962
152
1000 Genomes
Estonia
1
32
2412
1
31.9
4480
27,649
241
80,187
680
Genetic variation in the Estonia population
Bulgaria
1
29.94
1335
DNA
DNA
DNA
30,565
1222
49,591
1910
–
Bangladesh
1
27.3
86
1
27.3
86
3183
48
3684
55
1000 Genomes
This study was excluded for deviation of genotypes distribution from
HWE
India
2
24.14
205
5
12.1
766
7600
110
8791
117
1000 Genomes, Alagarasu et al., 2014, Biyani et al., 2015, Meena et al., 2015
Pakistan
1
24
96
1
23.9
96
2315
49
3003
65
1000 Genomes
Sri Lanka
1
31.9
102
1
31.9
102
2420
12
4316
26
1000 Genomes
Iran
1
29.66
118
1
29.6
118
15,660
671
22,237
739
Habibabadi et al., 2020
Australia
1
33.7
163
1
33.7
163
1118
35
1140
35
Allikmets et al., 2009
COVID-19 related data were obtained from article Dhangadamajhi et al and worldometer
assessed on 1st April 2021. TLR3 rs3775291 polymorphism genotype or allele data were
obtained from 1000 genomes, dbSNP, PubMed and Google Scholar
DNA data not available