Written Evidence Submitted by Dr Derek Gatherer
(CLL0006)
Fig.1: Deaths per million of population (dpm) from COVID-19. Black: less than 100 dpm; blue: 100-250 dpm; red: greater than 250 dpm. Black ovals – low impact central & southern zones. Blue ovals – intermediate impact “buffer zones”. CC-BY-SA 3.0 (Map by “San Jose”, annotated by author).
Title of submission: Over several decades, serology on respiratory viruses has been neglected, thereby seriously handicapping our ability to understand mortality patterns for COVID-19
Summary: The pattern of deaths per million from COVID-19 shows a striking geographical feature. Central Europe and the southern Mediterranean coast have been comparatively least affected, whereas the Atlantic coast and northern Mediterranean coast have been worst affected. To the west and east of the low affected area, are “buffer zones” of intermediate severity. Various reasons for this pattern may be suggested, but I propose that the most likely is cross-over immunity due to previous localised epidemics of one of the milder coronaviruses. However, serology - the discipline producing the studies that would enable us to answer this question – petered out in the 1980s as research funds drifted into other areas, notably genome sequencing. COVID-19 has highlighted the consequences of this neglect (it was already apparent with regard to Ebola virus in Africa). The lesson we should learn from this is to restore the serology of viral infections as a research priority.
Main text
Figure 1 shows the deaths per million of population (dpm) from COVID-19 in Europe and immediate vicinity, given by the WHO in mid-October 2020. There is a striking geographical pattern, with two strips of relatively lightly affected countries – one extending southwards from Finland to the northern Balkans and the other on the southern shores of the Mediterranean. Some of these countries have been very lightly affected indeed: Slovakia, for instance, has a dpm comparable to that of New Zealand. The low impact zone in central Europe is flanked by intermediate zones to the east and west. In the west, the intermediate zone is narrow and dpm sharply rises as we pass from Germany into the Low Countries and France.
Comparison of mortality rates between countries has, of course, become a major sport in the media. The most common suggestions advanced are, not necessarily in order of popularity:
1) Countries have different medical cultures surrounding classification of causes of death and conventions regarding the writing of death certificates.
2) Some countries are simply better governed than others and these tend to be the ones with lower mortality rates.
3) Some countries are simply more isolated than others or have a low population density, both of which limit the opportunities for infection.
Explanation 1 has some credibility, for instance we had a major correction in the way we counted COVID-19 deaths in the UK by removing those who died 28 days after diagnosis. However, it would be strange indeed if mere medical convention showed such clear geographical patterns. If explanation 1 was the major reason, we would expect to see a far more sporadic pattern.
Explanation 2 is popular in the media, with Germany, Norway, Finland, Slovenia, Austria, Denmark etc. often given as examples. However, this is rather compromised by the failure to explain why counties like Hungary and Belarus are always omitted from such lists, when indeed they are doing better than Germany and Denmark.
Explanation 3 has its most obvious example outside of Europe, in New Zealand. In Europe, there may be analogous situations in Iceland and the Faroes. However, it is scarcely possible to maintain that Austria or Slovakia are similarly low population density islands under effective quarantine.
The decline in mortality as we move from west to central Europe and then its increase again as we move yet further east cannot be explained by the above three explanations and is indicative of some intrinsic factor at work. There are therefore three further possibilities:
4) Respiratory viruses spread better in cool, damp climates and these climates also foster the development of complicating symptoms.
5) Genetic variation provides some innate protection to some populations.
6) Some populations have some pre-existing partial immunity, obviously not to COVID-19 but acquired via exposure to another coronavirus.
Explanations 4 and 5 fall down almost immediately because the correlations proposed do not match the pattern of the mortality data. Explanation 4 may appear initially tempting given the highest mortality on the Atlantic seaboard, but Italy and the southern Balkans are counter examples. Likewise, explanation 5 falls because there is no known pattern of genetic variation in white-European populations that matches the pattern of COVID-19 mortality, nor is the distribution of BAME populations in Europe congruent with the pattern observed, despite the established greater mortality rates in BAME individuals.
This leaves explanation 6. In this scenario, one of the known four milder coronaviruses that seasonally infect humans (229E, NL63, HKU1 and OC43) would have caused an outbreak in central Europe in a recent winter. The four milder coronaviruses occur in irregular seasonal cycles, often at intervals of two or three years. Most winters, a subset of the four will be in circulation. In rare instances, all four will be present at once, as happened in Australia in 2004. Immunity raised during such outbreaks may be partially cross-protective against other coronaviruses. This theory could account for the low mortality zone in central Europe and for the mortality gradients as we move west and east – out of the presumed epicentre of the previous coronavirus outbreak.
This theory ought to be easy to test. One would simply review the medical literature for data on coronavirus antibodies in European populations over the last few decades, looking for an immunological signal of a central European outbreak. However, this is not possible, as the required data simply does not exist. My own efforts to retrieve such data have yielded only two publications: one-third of residents of Hamburg had antibodies to coronavirus OC43 in 1975 and 58% of Hungarians sampled five years later. I was not surprised by this lack of evidence. The decline of serology as a discipline since the 1980s, underfunded by comparison with more glamorous new technologies such as genome sequencing, has meant that we are lacking data on exposure of populations to many pathogens. Genome sequencing is good at telling us how many people have a disease at the time of testing, but it tells us nothing about those who have recovered. Since recovered patients may no longer be secreting virus, the only trace of their past infection is in their antibodies. Interest in serology declined because it was regarded as old fashioned, doubt was sown about its sensitivity and specificity (we have seen such doubt expressed again with regard to antibody tests against COVID-19), research funding drifted away and many clinical units simply stopped performing it.
The consequences of the neglect of serology have already produced grave consequences with respect to another important disease – Ebola virus in Africa, consequences about which I have written previously, both in the scientific literature and in more popular form. Doubt cast on older serological surveys for Ebola virus antibodies in Africa and the virtual abandonment of such surveys from the early 1990s onwards was a direct contributor to our failure to recognise the Ebola outbreak that began in Guinea in late 2013. The delay in action allowed the virus to spread and become the first wide-area Ebola epidemic in history. Serology for coronaviruses has also suffered as a consequence of this trend to move to genome sequencing. Additionally, those respiratory viruses regarded as “mild” or simply the causes of “the common cold” have not been seen as worthy of investigation. The MRC Common Cold Unit closed in 1989 and attempts at a partial revival through the Cardiff Common Cold Centre also ended in 2017.
I should stress that I am not against testing by genome sequencing. We have learned many useful things about COVID-19 from its genome data, which could not have been discovered using any other method. However, data on antibodies are as important as genome data. Genome sequencing tells us what COVID-19 is doing now, but antibodies tell us what has happened, either very recently or at longer distances of time. Because of a neglect of antibody testing in the past, probably compounded by a neglect of “common cold”, we are unable to interpret the geographical pattern of mortality from COVID-19 infection. Of the six possible explanations considered here, only the immunological theory would appear to be plausible, but we have neglected over the years to collect the evidence that could address it. It is not too late to start, but one lesson we take from COVID-19 (as previously from Ebola) must be that we need to restore serology to our list of research priorities. COVID-19 has also reminded us (although SARS in 2003 also ought already to have prompted this) that viruses regarded as trivial – “common cold” etc. – may turn out to have considerably nastier relatives, whose arrival leaves us wishing we had paid more attention to, and learned more from, their milder cousins.
(October 2020)