Skip to main content

ORF3a mutation associated with higher mortality rate in SARS-CoV-2 infection

Our take —

This study explored the question of whether mutations in the SARS-CoV-2 genome were associated with higher mortality at a global scale. While this is an important question, there are several strong methodological flaws in this study that make it difficult to support the authors’ conclusion that mutations in ORF3a are associated with higher mortality. This does not mean that the mutation cannot be associated with higher mortality, only that this study does not provide evidence for this.

Study design

Ecological

Study population and setting

This study included COVID-19 incidence and mortality data from 23 countries with at least 20,000 cases as of May 2, 2020 (downloaded from www.worldmeters.info/coronavirus). The authors selected 15 of these countries for further analysis by identifying all countries in the bottom half of both infection and mortality rates (Saudi Arabia, Portugal, India, Peru, Russia, Turkey, Germany), and all countries in the top half of both infection and mortality rates (Mexico, Sweden, Netherlands, Belgium, Brazil, Iran, UK, Spain). The authors used publicly available genetic sequence data from these 15 countries (218 sequences in total) and compared observed mutations in the low and high infection/mortality groups.

Summary of Main Findings

The authors claimed an association of mutations in ORF3a in the SARS-CoV-2 with mortality during the pandemic. They did this by identifying several ORF3a mutations that occur only in sequences from high infection/mortality countries and not in the low infection/mortality group. They then performed epitope and structure prediction on the ORF3a protein with and without these mutations and found that the mutations lead to loss of B cell epitopes and altered protein structure.

Study Strengths

This study had many strong limitations that prevent valid conclusions.

Limitations

First, the researchers compared only low infection and low mortality countries to high infection and high mortality countries. To really understand mortality separately from infection, the authors would also have needed to consider the other combinations (low infection, high mortality and high infection, low mortality). Second, the authors assumed that the infection and mortality data is accurate and unbiased and does not suffer from selection bias or confounding such as quality of healthcare for infected individuals. Infection rates were calculated simply as total cases/total tests, which can be vastly skewed by testing rates in these countries. Without more information about testing availability in each country, the infection rate cannot be assumed to be correct. Third, the authors claimed that SARS-CoV-2 genetic diversity lead to varied clinical outcomes and used this to connect ORF3a mutations to mortality. While it is possible that some variants may have lead to varied outcomes, it is faulty to assume that all changes lead to changed clinical outcomes. Indeed, several large studies (e.g., published analyses from the COG-UK consortium) have so far shown no clear correlation between already-observed diversity and clinical manifestation of SARS-CoV-2 infection. Finally, this analysis was based on only 218 publicly available genomes spanning an unknown timeframe. Biases and discrepancies in the amount of sequence data available from each country (and when these sequences were collected) may have affected the number of mutations accumulated, so it is unfair to state that not observing a mutation in this small data set means it was never observed in a particular country.

Value added

This paper is problematic in its potential for bias and as such, should be approached with skepticism.

This review was posted on: 3 January 2021