Association of the IFNG +874T/A Polymorphism with Symptomatic COVID-19 Susceptibility

Round 1

Reviewer 1 Report

The authors present a study comparing the prevalence of 2 well-known polymorphisms (one in TNF gene, the other in IFNg gene) in a control population of 397 (or 497 for the 2nd polymorphism) subjects and in a case population of 222 symptomatic (non severe and severe) subjects, all from the Belem region in Brazil. The analyses were performed by Taqman on DNA extracter fom PBMCs. The results are clearcut : there is no difference of minor allelic frequency in the TNF polymorphism (12% vs 10% in the case group) while there is a significant difference for the IFNg polymorphism (25% vs 33% in the case group).

The authors provide a molecular mechanism explaining the symptomaticity associated with the IFNg variant and indicate that the IFNg SNP could be used as a marker for prognosis in Covid patients.

This study is clean and properly performed, and the results are clear. 

Here are my comments to the authors :

The paper is proper and could be slightly improved if they give the rsXXX number attributed to the 2 SNPs they are studying and on which chromosome the genes are located.

They indicate that the control populations come from 3 studies for TNF and from 2 studies for IFNg, could they confirm that the minor allelic frequencies are similar in all the control populations ? It would reinforce the association they observe.

In Table 2 they provide only frequencies, they should also provide odds ratios. These odds ratios indicate the odds of being symptomatic (and/or severe) when carrying the genotype (or allele) compared to controls.

I believe that concluding that the IFNg SNP could be used as a predictive marker for developing symptomatic Covid is a bit short, since we don’t have the odds ratios and it may not be cost-effective or even useful. They have not shown that most individuals carrying INFg TT would become symptomatic and should see how they can evaluate this. On the other hand, most individuals, even with IFNg TT, have a non severe Covid and it is not sure such a test would be useful. These points should be discussed and if the TT genotype itself is not sufficient to allow a predictive test for symptomatic Covid, they can rather mention that the marker could contribute to be a part of a predictive test involving other risk criteria. 

Author Response

The same text below is attached.

Response to Reviewer 1

The authors present a study comparing the prevalence of 2 well-known polymorphisms (one in TNF gene, the other in IFNg gene) in a control population of 397 (or 497 for the 2nd polymorphism) subjects and in a case population of 222 symptomatic (non severe and severe) subjects, all from the Belem region in Brazil. The analyses were performed by Taqman on DNA extracter fom PBMCs. The results are clearcut: there is no difference of minor allelic frequency in the TNF polymorphism (12% vs 10% in the case group) while there is a significant difference for the IFNg polymorphism (25% vs 33% in the case group).

The authors provide a molecular mechanism explaining the symptomaticity associated with the IFNg variant and indicate that the IFNg SNP could be used as a marker for prognosis in Covid patients.

This study is clean and properly performed, and the results are clear. 

Here are my comments to the authors:

The paper is proper and could be slightly improved if they give the rsXXX number attributed to the 2 SNPs they are studying and on which chromosome the genes are located.

R = Thank you! We inserted the rsxxx information in tables and text, as well as the chromosome localization of both SNPs in Material and methods section.

They indicate that the control populations come from 3 studies for TNF and from 2 studies for IFNg, could they confirm that the minor allelic frequencies are similar in all the control populations? It would reinforce the association they observe.

R = The frequencies obtained from previous studies in the same population (Belém) showed that for the minor allele *A of the SNP rs1800629 (TNF -308G/A) ranged from 8.3% to 14.5%, given that the frequency for COVID-19 patients in our study is 9.9%, thus, in this range. Neverthless, the frequencies of the minor allele *T of the SNP rs2430561 (IFNG +874T/A) were 23.7% and 28.4% in previous studies, then, lower than in our COVID-19 sample (33.5%).

We Included this information in the text of the manuscript and believe that could reinforce our conclusions. Thank you very much.

In Table 2 they provide only frequencies, they should also provide odds ratios. These odds ratios indicate the odds of being symptomatic (and/or severe) when carrying the genotype (or allele) compared to controls.

R = It was done.

I believe that concluding that the IFNg SNP could be used as a predictive marker for developing symptomatic Covid is a bit short, since we don’t have the odds ratios and it may not be cost-effective or even useful. They have not shown that most individuals carrying INFg TT would become symptomatic and should see how they can evaluate this. On the other hand, most individuals, even with IFNg TT, have a non severe Covid and it is not sure such a test would be useful. These points should be discussed and if the TT genotype itself is not sufficient to allow a predictive test for symptomatic Covid, they can rather mention that the marker could contribute to be a part of a predictive test involving other risk criteria. 

R = We agree. Thank you! We rewrote the text accordingly.

Author Response File: Author Response.pdf

Reviewer 2 Report

In this study, Lima de Sarges et al. report an association between the *T allele of IFNG +874T/A and susceptibility to symptomatic COVID-19 by comparing patients recruited at the Evandro Chagas Institute and at the Belem Adventist Hospital and healthy controls recruited in the Belem population.

A key question is to know whether the genetic background of both populations is similar or not. If the ethnic origin of patients recruited in the area of the Evandro Chagas Institute and of the Belem Adventist Hospital is not representative of the multiplicity of ethnicities in Belem, the results can not be interpreted. Such a discrepancy may explain for instance why, as noticed by the authors, in the reference 19 conducted in Hong Kong the control population presented a lower frequency of *T allele of IFNG +874T/A than in the general Chinese population.

The authors explain the link between the *T allele of IFNG +874T/A and susceptibility to symptomatic COVID-19 by the fact that IFNg induces ACE2 expression. Yet, if this was the case, *T allele of IFNG +874T/A should be also associated with disease severity. Another explanation that the authors should discuss is the possibility that IFNg overexpression paves the way for clinical symptoms. Yet, here again, IFNg overexpression should be associated with disease severity.

The authors claim that « there is good agreement between our results and the trend observed in the meta-analysis by Leite et al. ”, yet, in contrast with their results, in this review IFNG +874 (T) correlated positively with COVID-19 mortality.

In the discussion the authors should state that the absence of link between TNF polymorphism and COVID-19 they observed might be due to a power deficiency, given the relatively low number of participants they recruited.

The p values should be indicated in the Table 2, not in its legend.

Author Response

The same text below is attached.

Response to Reviewer 2

In this study, Lima de Sarges et al. report an association between the *T allele of IFNG +874T/A and susceptibility to symptomatic COVID-19 by comparing patients recruited at the Evandro Chagas Institute and at the Belem Adventist Hospital and healthy controls recruited in the Belem population.

A key question is to know whether the genetic background of both populations is similar or not. If the ethnic origin of patients recruited in the area of the Evandro Chagas Institute and of the Belem Adventist Hospital is not representative of the multiplicity of ethnicities in Belem, the results can not be interpreted. Such a discrepancy may explain for instance why, as noticed by the authors, in the reference 19 conducted in Hong Kong the control population presented a lower frequency of *T allele of IFNG +874T/A than in the general Chinese population.

R= This question addresses the major point of case-control studies: the assumption that subsamples of a given sample are panmictic and not genetically differentiated. This issue was first approached by Wahlund in 1929 and is nowadays frequently referred to as population stratification or substructuring. Such effects must be avoided, as they compromise the statistical results, interpretation, and conclusions.

Our group has a long history of studies with case-control approaches, and, like many others, we are concerned about this issue and employ two strategies.

The first strategy is demographic. In the present study, we ensure that all sampled patients reside in the same metropolitan area, namely the Greater Belém, comprising several closely located municipalities with high mobility among them. It is important to note that during the pandemic, patients were not able to choose hospitals or care units close to their homes, seeking available healthcare anywhere.

Another strategy involves the use of additional genetic markers. Substructuring would affect all genetic loci. We analyzed two loci, and statistically significant differences were observed in only one. A similar pattern was observed in our previous study (Rodrigues et al. 2023; https://doi.org/10.3390/v15051197), where only one SNP locus was found to be associated, while the remaining two displayed similar genotype and allele frequencies.

In a second study, approaching long COVID in a different sample of Belém, our group also compared allele and genotype frequencies with those obtained from the literature for Belém (Silva et al, 2023; https://doi.org/10.3390/v15040885). The same pattern was found. An interesting point is that we used ten SNPs and most of them displayed no differentiation among Belém and subsamples.

Therefore, despite the importance of ethnicity in COVID-19, our study provides no evidence of sample substructuring biases introduced by potential ethnic differences among our sub-samples.

We agree that this issue must be better addressed in the manuscript and include much of the discussed above in the manuscript text.

The authors explain the link between the *T allele of IFNG +874T/A and susceptibility to symptomatic COVID-19 by the fact that IFNg induces ACE2 expression. Yet, if this was the case, *T allele of IFNG +874T/A should be also associated with disease severity. Another explanation that the authors should discuss is the possibility that IFNg overexpression paves the way for clinical symptoms. Yet, here again, IFNg overexpression should be associated with disease severity.

R = Indeed, the association with symptomatic COVID-19 and the absence of an association signal for the *T allele of IFNG +874 with severity appears to be paradoxical. It is possible that a lack of statistical power due to small sample size plays a role. However, it is also plausible to suppose that after the development of symptoms, influenced by IFNG polymorphism, the outcome of the disease, in terms of severity, can be more strongly determined by other factors, like genetic ones that are still unknown, or environmental and biological, like age, comorbidities, sex, nutritional status and access to the healthcare system.

We thank you for instigating such discussion and we included it in the manuscript text.

The authors claim that « there is good agreement between our results and the trend observed in the meta-analysis by Leite et al.”, yet, in contrast with their results, in this review IFNG +874 (T) correlated positively with COVID-19 mortality.

R = The paper of Leite et al found a positive correlation of the allele T with the mortality rate worldwide. Our results suggest that the same allele is associated with the development of symptoms. If the *T allele populational frequency influences the proportion of symptomatic individuals that can evolve to a worse disease outcome, this allele can indirectly influence the mortality rate. However, we agree that the sentence must be improved. Thank you.

In the discussion the authors should state that the absence of link between TNF polymorphism and COVID-19 they observed might be due to a power deficiency, given the relatively low number of participants they recruited.

R = Of course we cannot rule out the possibility of lack of power. However, the differences in allele and genotype frequencies are really very small. It would be necessary thousands of patients to reach statistical significance considering differences lower than 2%. Alternatively, replicate the study in other population with different ethnic and demographic backgrounds would help to confirm the lack of association. As referred in the manuscript Chong et al did not observe any association between TNFA polymorphism and COVID-19 severity or symptoms. Thus, we think the lack of power possibility is less probable. We improve the manuscript text to address this issue.

The p values should be indicated in the Table 2, not in its legend.

R = It was done, thank you.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Thanks for your answers.