COVID-19 and ABO blood group: another viewpoint.

Li et al. have recently published ‘Association between ABO blood groups and risk of SARS‐CoV‐2 pneumonia’, an observation already reported a few weeks ago as a MedRxiv preprint by Zhao et al. and which had a certain impact in the press.

In both studies, the ABO blood groups distribution of patients with coronavirus disease 2019 (COVID‐19) were compared to that of controls from the local populations that showed that blood group A was associated with an increased risk of infection, whereas group O was associated with a decreased risk. Considering this information rather as a working hypothesis, some scientists have called for caution.

However, as already strongly suggested by others, this variable susceptibility to severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) infection could be linked to circulating anti‐A antibodies, which could interfere or even inhibit the virus–cell adhesion process.

We had the idea to analyse these important available data series from the anti‐A or ‐B antibodies viewpoint instead of ABO blood group antigens as the authors did.

In fact, considering the largest series of patients with COVID‐19 (N = 1888) analysed by Zhao et al., we compared the proportion of those possessing anti‐A in their serum (i.e. those of B and O blood groups) and those who did not (i.e. those of A and AB blood groups) to the control cohort (N = 3694; Table I).

Table I

Comparison of subjects with/without anti‐A antibodies in their serum.

	RBC blood group	Control, n (%)	COVID‐19, n (%)	χ2	P	OR (95% CI)
With anti‐A	B and O	2170 (58·7)	927 (52·2)
	A and AB	1524 (41·3)	848 (47·8)	20·74	<0·001	1·30 (1·16–1·46)
Without anti‐A	A	1188 (32·2)	670 (37·7)	19·97	<0·001	1·32 (1·17–1·49)
	AB	336 (9·1)	178 (10·0)	4·58	0·0323	1·24 (1·02–1·51)

The results (Table I) indicate that subjects with anti‐A in serum (i.e. B and O blood groups) are significantly less represented in the COVID‐19 group than those lacking anti‐A whatever the group: A and AB (P < 0·001), A (P < 0·001) or AB (P = 0·0323), whereas there was no significant difference versus circulating anti‐B (data not shown).

We then wondered if there was a difference between anti‐A from O and anti‐A from B, and then we compared the supposed protective effect of anti‐A from O and from B (Table II).

Table II

Comparison of anti‐A from O and from B subjects.

	RBC blood group	Control, n (%)	COVID‐19, n (%)	χ2	P	OR (95% CI)
Anti‐A from O	O	1250 (57·6)	458 (49·4)
Anti‐A from B	B	920 (42·4)	469 (50·6)	17·64	<0·001	1·39 (1·19–1·62)

Whereas both blood group O and B patients possess circulating seric anti‐A, it appears and it is statistically highly significant (P < 0·001) that O group patients are underrepresented (49·4 % vs. 57·6%), whereas B group patients are, on the contrary, overrepresented (50·6% vs. 42·4%), meaning that anti‐A from O is more protective than anti‐A from B.

This latter observation is probably related to the fact that the immunoglobulin predominant isotype of anti‐B/anti‐A is IgM in serum from group A and B individuals, but IgG in O group serum, an already known notion, which has been well documented by flow cytometry.

In conclusion, this way of analysing the data strongly suggests that the presence of anti‐A antibodies in serum and more specifically IgG anti‐A, should be considered as a factor more significant than the blood group itself, as far as the relationship between COVID‐19 susceptibility and ABO blood groups is concerned.

Far from intending to corroborate the authors' conclusions as such, we wanted to show that the resources of immuno‐haematology allow several approaches that could perhaps be useful for the disease follow‐up.