Centenarians exposed to the Spanish flu in their early life better survived to COVID-19

Centenarians and COVID-19 pandemic

Although widely emphasized by the media [5], the resilience of centenarians to COVID-19 remains a controversial issue that is now increasingly attracting researchers. Couderc et al. [6], in a study on 321 nursing home residents in Southern France, including 12 centenarians, reported a higher mortality rate among these centenarians (50% vs 24.6%) compared with other younger residents, corresponding to one of the lowest survival rates compared with other published series [7]. Similarly, Marcon et al. [8], in a study on 42 centenarians from North-Eastern Italy as part of the CaT (Centenari a Trieste) project, observed that COVID-19-related mortality among long-lived individuals was higher than that of the population between 50 and 80 years of age, and that the mortality rate among the oldest women exceeded that of men. Overall, these data suggest that despite their ability to reach the extremes of human lifespan, centenarians are not particularly resistant to COVID-19 compared with the general population. In this context, a first research question arises: Does the oldest old, and more specifically the centenarians, die more than usually during the COVID-19 pandemic?

The impact of the end of WWI

WWI (1914-1918) led to a period of deprivation for the Belgian population suffering under German occupation. Especially in 1917 and 1918, the nutritional status and sanitary conditions of the population had further worsened until 1920 [10]. The impact of nutrient deficit and low-quality foods during wartime may have had devastating effects on both the short- and long-term health of the new-born. Protein-energy malnutrition may heavily affect the course of pregnancies, causing increased stillbirth rates [11]. Unfortunately, no data on the number of stillborn and only little data on the level of infant mortality during WWI are available in Belgium [12]. An increase in infant mortality was observed in 1917 and 1918, a similar trend also reported in the neighboring countries, e.g., France and the Netherlands [13]. Such an increase may have resulted in a stronger selection against the weakest babies, born in this period, considering that harsh privations persisted more than a year after the end of the war. Research on French children born during WWI revealed that, in general, parental socioeconomic conditions were a strong predictor of new-borns’ longevity [14]. Children whose mothers had faced famine during pregnancy tended to have a shorter lifespan [15]. These long-term effects have been ascribed to a direct fetal distress or mediated by epigenetic mechanisms [16]. Nevertheless, it is important to mention the strong support by the U.S. Commission for Relief in Belgium that improved the food supply for babies, thereby ensuring them better early-life conditions despite the poor socio-economic context lasting until the end of 1919 [17]. Although some kind of selection early in the life of these generations can hardly be excluded, its impact on their survival into old ages is unknown, and there is no consensus on the existence of its favorable or detrimental significance [18]. If such a selection was real and supposed to be favorable for survival, babies born during WWI or immediately after would be expected to have different mortality risks along their whole lifespan and not exclusively at the extreme end of life. To test the selection hypothesis, we disaggregated the mortality risks of babies born before and after the end of WWI, considering their month and year of births for the periods 1991-2000 and 2001-2010. The corresponding curves presented in Figure 4 do not show any evidence of a significant variation of mortality for people born around the end of WWI. We cogently conclude that a selection in terms of WWI can hardly be responsible for the relative better survival of ‘older’ centenarians compared with ‘younger’ ones against 2020 COVID-19, despite both groups likely experiencing similarly threatening early-life conditions during the three years around the end of WWI.

Figure 4. Probability of dying from 1991 until 2000 and from 2001 to 2010, calculated by year and month of births (STATBEL and data from Centre de Démographie, UCLouvain, Belgium).

The impact of “Spanish flu”

The second important event that may potentially explain our results is the outbreak of the Spanish flu that caused about 50,000 deaths by the fall of 1918, i.e., more than the estimated 44,000 victims of WWI [19]. As explained by Brulard in his dissertation [20], the Belgian press covering the timeline of the Spanish flu mentioned the latter for the first time on July 7, 1918. The overall death rate in the city of Brussels increased: from July 1 to July 20, 68 civilians of the age 10–40 years died, while the number of deaths for the same age group in June was 35. In the first half of August, the whole country was facing the virus but, very quickly, the pandemic subsided, only to re-emerge in October. This second wave was by far the most devastating as more than half of the victims attributed to the Spanish flu died between mid-October and mid-November. Thereafter, it gradually lessened in intensity, although many deaths linked to the Spanish flu were still recorded in December. A third wave resurfaced in January 1919, culminating in February. Although it was less deadly than the second wave, it testified to the persistent circulation of the virus, which did not disappear until the end of May 1919.

In our analysis, we detect a synchronic association between the timing of the Spanish flu pandemic and the switch in centenarians’ EMR. To better outline this potential association, we group the number of observed and expected deaths of centenarians during the 2020 pandemic into three periods of birth, namely, before, during and after the Spanish flu pandemic (Table 1). A significant difference in the EMR of centenarians in 2020 is evident between the “young” centenarians born after August 1, 1918 and the “older” ones born before that date.

The near-perfect synchronism between the EMR gap during the 2020 pandemic for cohorts born before and after August 1, 1918 and the surge of the influenza pandemic naturally suggests a causal effect, which becomes plausible as we excluded the main alternative explanation based on the impact of WWI ending, especially considering the persistence of poor living conditions after the conflict.

For the exposure to the Spanish flu pandemic, three groups of newborns could be distinguished. The first group included babies who were already born before the pandemic broke out, and who faced the virus in their early life. The youngest among them could have been protected by maternal antibodies transferred through breastmilk during the first months of their life. Nonetheless, as the duration of the pandemic exceeded that of maternal protection, the majority of them had to face the virus. In contrast, babies born after August 1, 1918 but before the pandemic completely disappeared by the end of May 1919, could have been protected by maternal immunity, with a variable risk of coming in contact with the virus. Babies born after May 1919 might have been exposed to the H1N1 virus but only in utero and, if they survived, were protected by maternal immunity in their first months of life, while those born after January 1920 were not exposed to the virus at all.

Only the first group faced the virus expressing a significant survival advantage later in life during the 2020 COVID-19 pandemic. Our speculative hypothesis is that most of them could have developed immune memory cells capable of recognizing epitopes antigenically related to the H1N1 virus potentially even a century later. This hypothesis is supported by some scientific evidence. For instance, individuals born before 1957 and exposed to the H1N1 influenza A virus were better protected from the 2009 pH1N1 [21]. Even though the influenza and SARS-CoV-2 viruses are different, some structural homology between them has been reported [22] and a subset of the T cell repertoire capable of cross-reacting with both influenza virus and SARS-CoV-2 virus epitopes has been identified [23]. More specifically, SARS-CoV-2 T-cell repertoires with specificity to both the coronavirus and the M1 immunodominant epitope of influenza virus are more frequent that expected, which may have relevant implications in the response to COVID-19 for those individuals previously exposed to some influenza strains. Indeed, this raises the possibility that cohorts exposed to the Spanish flu in 1918 were capable to mount an effective response also against COVID-19 in 2020. Efficient memory cells can persist for many decades, as demonstrated by the study of Yu et al. [24] wherein individuals exposed to the H1N1 pandemic in 1918/1919 were still able to produce neutralizing antibodies a century later, confirming that a centenarian's immune system can successfully react to pathogens to which they were exposed early in life. However, this hypothesis is in contrast with that of Gagnon et al. [25] that exposure to the influenza pandemic early in life is a risk factor for dying during subsequent heterosubtypic pandemics. For this reason, a putative influenza / coronavirus cross-response should be viewed an interesting research hypothesis that will require further investigations aimed at better clarifying the underlying molecular mechanisms acting in the host.

Strengths, limitations, and future research

The main strength of the presented study is the direct link established between the two greatest pandemics of the past one hundred years. Our investigations are innovative in that we study the only persons exposed to both pandemics, i.e., today’s centenarians. Fortunately, reliable and exhaustive data on centenarians and their survival is available in Belgium, and the sufficient number of people involved in our analysis allow a high level of statistical significance. Data concerning people born at the time of the Spanish flu and are still alive at the onset of the 2020 COVID-19 pandemic on March 10 are examined in great detail (according to month of birth), something that hardly occurs in such kind of research, taking into account the size of each birth cohort. Nevertheless, the study has several limitations. First, due to the small number of male centenarians, a separate analysis of males and females could not be performed. Another limitation is linked to the difficulty in distinguishing COVID-19-related deaths and those due to other causes. Further, only the overall mortality, including that due to COVID-19, is addressed as during the early stage of the pandemic, some cases are probably not attributed to COVID-19. Third, it cannot be ruled out that the susceptibility to COVID-19 among the oldest old is strongly influenced by the specific living conditions of this age group, as most centenarians were confined to nursing homes. Further research is on-going to assess the impact of these collective living arrangements on mortality during the pandemic. Finally, we do not carry out direct blood testing in the oldest people experiencing COVID-19, therefore our hypothesis of protection provided by a cross-reactive immunity elicited during a previous H1N1 pandemic remains entirely speculative. Nevertheless, the possibility that those exposed to the Spanish flu have developed immune mechanisms capable of inducing a more effective anti-COVID-19 response than non-exposed people born after it is intriguing and deserves further investigation across different populations.