Altitude conditions seem to determine the evolution of COVID-19 in Brazil

COVID-19 is spreading rapidly in Brazil, a country of continental dimensions, but the incidence of the disease is showing to be very heterogeneous, affecting cities and regions differently. Thus, there is a gap regarding what factors would contribute to accentuate the differences in the incidence of COVID-19 among Brazilian cities. This work aimed to evaluate the effect of altitude on the incidence of COVID-19 in Brazilian cities. We analyzed the relative incidence (RI), the relative death rate (RDR) of COVID-19, and air relative humidity (RH) in all 154 cities in Brazil with a population above 200 thousand inhabitants, located between 5 and 1135 m in altitude. Pearson's correlation analysis was performed to compare a relationship between altitude with RI and RDR, and between RH with RI and RDR. Altitudes were classified into three classes [low (altitude ≤ 97 m a. s. l), middle (97 < altitude ≤ 795 m a. s. l), high (795 < altitude ≤ 1135 m a. s. l)] for the RI, RDR, and RH variables. To compare the three classes of altitude, analysis of variance (ANOVA) and Tukey test were used to compare averages (p < 0.05). Our epidemiological analysis found that the RI, RDR, and RH were lower in cities located in high altitudes (795 < altitude ≤ 1135 m a. s. l) when compared to the middle (97 < altitude ≤ 795 m a. s. l) and low (altitude ≤ 97 m a. s. l) cities altitudes. Furthermore, our study shows that there is a negative correlation between the incidence of COVID-19 with altitude and a positive correlation with RH in the cities analyzed. Brazilian cities with high altitude and low RH have lower RI and RDR from COVID-19. Thus, high altitude cities may be favorable to shelter people at risk. This study may be useful for understanding the behavior of SARS-CoV2, and start point for future studies to establish causality of environmental conditions with SARS-CoV2 contributing to the implementation of measures to prevent and control the spread of COVID-19.

In December 2019, a new coronavirus was identified-initially called 2019-nCoV and later renamed SARS-CoV-2-in Hubei province, People's Republic of China 1 . The pathogen causes coronavirus-19 disease (COVID- 19), which spread rapidly, reaching the pandemic level on March 11, 2020 2 . The incidence of cases and deaths caused by COVID-19 in the world increased at different rates from the first cases. In Brazil, a country of continental dimensions, the spread of the disease is very heterogeneous, affecting cities and regions differently.
Efforts to minimize its spread are announced every minute in the media. Also, thousands of searches worldwide are focused on the various nuances of COVID-19. However, studies that relate to the incidence of the disease to environmental factors are incipient, especially in countries like Brazil, which stands out for its great socioeconomic and environmental diversity 2 . As the behaviour of COVID-19 concerning climatic attributes is still poorly understood, investigating the influence of altitude and environmental characteristics of cities on the incidence and deaths caused by COVID-19 can generate results that contribute to the development of public policies that minimize the spread of the disease.
In Brazil, COVID-19 began to be disseminated from large cities located in the coastal zone, that have high air relative humidity (RH), and quickly reached the interior, but the contamination speed of the population is not homogeneous 2 . Some cities have higher incidence and deaths from COVID-19 than other cities with similar numbers of inhabitants. About that, some questions were raised about the differences between the records of infection and deaths between cities. What factors would be able to contribute to accentuate these differences Figure 1. Distribution of 154 Brazilian cities with a population above 200 thousand inhabitants selected to analyse the relative incidence and relative deaths per 100.000 inhabitants of Coronavirus Disease-19 (COVID-19) cases according to three classes of altitudes. Low (red), middle (orange) and high (blue). The information is available in free spatial data DIVA-GIS (http://www.diva-gis.org/Data). The map was made using Quantum geographic information system (QGIS) version 3.14.15 6 1) and was higher in the north, followed by the northeast, southeast, central-west and, south (Table 2). On June 01, 2020, the RI varied between 12.10 (Sete Lagoas) and 1063.49 (Sobral) ( Table 1) and was higher in the north, followed by the northeast, southeast, central-west and, south (Table 2). On June 16, 2020, the RI changed between 18.78 (Sete Lagoas) and 2998.95 (Parauapebas) ( Table 1) and was higher in the north, followed by the northeast, southeast, central-west, and south (Table 2). On July 01, 2020, the RI changed between 64.98 (Montes Claros) and 4758.18 (Parauapebas) ( Table 1) and was higher in the north, followed by the northeast, central-west, southeast, and south (Table 2).
In Brazil, between May 17 and June 01, between June 02 and June 16, and between June 17 and July, 01, 2020, the RI increased 92.3% 64.8%, 63.20%, respectively. The largest increase in RI was observed in the central-west region and the smallest in the northeast region (Table 2), and the rate of RI growth decelerated in all regions, except in the southern region (Fig. 2).
Between May 17 and June 01, between June 02 and June 16, and between June 17 and July 01, the RDR increased respectively 78.4%, 45.5% and 34.6% in Brazil. The largest increase in RDR was observed in the centralwest region and the smallest in the north region (Table 2), and the rate of RDR growth decelerated in all regions, except in the southern region (Fig. 2).
Between March 01 and May 17, 2020, the average RH data of 63 cities were analysed, which varied between 52 and 89.4% (Table 1) with an RH average of 74% (  Table 1). The first confirmed case in Brazil occurred on February 26 of 2020 in the city of São Paulo, and until April 8 of 2020, all cities in Brazil with a population of up to 200 thousand inhabitants confirmed at least one case of COVID-19 (Table 1). There were no significant differences between cities concerning the period of accelerated dissemination. Would, then, the altitude and the RH have contributed to accelerate the spread of the virus in Brazil?

Discussion
Our epidemiological analysis of the COVID-19 pandemic in Brazil indicates a direct association between the incidence of COVID-19 with altitude and RH in Brazilian cities with a population above 200 thousand inhabitants. The low RI and RDR in cities with higher altitudes may be related to environmental factors, which influence the spread of the virus and the physiology of human beings.
Climatic factors can be determinant for transmission by some viruses 10,11 . Thus, a study performed in Russia assessed possible climatic predictors in the rise of COVID-19 intensity. The results showed that temperature seasonality influenced COVID-19 spread in humid continental Russia, whereas temperature (diurnal range and seasonality) was influential in sub-artic Russia 10 . In addition, another study aimed to examine the linkage between climatic variables and COVID-19 in National Capital Territory (NCT) of Delhi, India. The results of that study suggested that climatic conditions in NCT of Delhi were favourable for COVID-19 and the disease may spread further with the increasing temperature, relative humidity, evaporation and wind speed 11 . However, it is necessary to consider comprehensive biophysical assessments of altitude, humidity, UV radiation in the maintenance and transmission of SARS-CoV-2 12,13 . Recently studies have been published about temperature and UV index, and that heating, and UV radiation can eliminate the viral infectivity 12,13 . These findings could help to understand the relationship of COVID-19 cases in different Brazilian cities due to altitude.
In contrast to what other studies report regarding the greater spread of the influenza virus in environments with lower RH 14 , our data show that COVID-19 RI is higher in cities where RH is highest. These results are in line with those obtained by C Arias-Reyes, N Zubieta-DeUrioste, L Poma-Machicao, F Aliaga-Raudan, F Carvajal-Rodriguez, M Dutschmann, E Schneider-Gasser, G Zubieta-Calleja and J Soliz 3 , which lists air dryness as one of the factors that control the spread of the virus at high altitudes. In working with climatic data from 5 large Brazilian cities (Manaus, Fortaleza, Brasília, Rio de Janeiro, and São Paulo) 15 , showed that moderate relative air humidity (averages between 77.7 and 81.6%) favor the spread of this disease. In the present study, 5 cities with the highest RI, in the four periods analyzed, had an average RH between 79.3 and 84.8%, values similar to those found by A Auler, F Cássaro, V da Silva and L Pires 16 .
Another point that should be considered is that the higher altitude, by itself, favors a higher incidence of ultraviolet (UV) radiation, especially in the UVA and UVB spectra, which can produce a bactericidal effect due to changes in the molecular chains of DNA and RNA. Thus, as a hypothesis, UV radiation would shorten the virus half-life, thereby reducing the virus's ability to survive in Brazilian cities located at higher altitudes and, consequently, the survival of the COVID-19 virus. M Blumthaler, W Ambach, and R Ellinger 17 observed an increase of 9 ± 2% in UV radiation for every 1000 m of altitude, under clear sky conditions. In addition, the increase in UV radiation can reach 11.3% (5 to 1135 m altitude) between the Brazilian cities studied in this work. Besides, considering that vitamin D production is dependent on exposure to UV radiation and that vitamin D levels positively modulate the immune system 18 , the hypothesis of higher immune defense against SARS-CoV-2 is plausible in cities with higher altitude. Future studies should investigate this hypothesis.
The low air density and greater distance between molecules in Brazilian cities located at higher altitudes could also reduce the inoculation of airborne viruses compared to sea level. C Arias-Reyes, N Zubieta-DeUrioste, L Poma-Machicao, F Aliaga-Raudan, F Carvajal-Rodriguez, M Dutschmann, E Schneider-Gasser, G Zubieta-Calleja, and J Soliz 3 suggested that inhabitants of cities with altitudes above 3000 m a.s.l are less susceptible to developing effects caused by COVID-19 due to ultraviolet radiation and thinner air.
Physiological factors can influence the pathogenicity of SARS-CoV-2 at high altitudes. The barometric reading varies with changing weather conditions and becomes lower as the altitude increases. Thus, the volumes of inspired air, which require humidification, are much higher than at sea level, and the air density is lower at high altitudes 19 . As a result, compensatory adjustments to facilitate the release of oxygen to cells occur in individuals living at higher altitudes, such as the increase in the levels of 2,3-diphosphoglycerate (2,3-DPG), a chemical compound found inside the red blood cell, whose function is to reduce hemoglobin's affinity for oxygen in order to facilitate its release into tissues 20 . Given the above, a probable hypothesis for less severity in individuals infected with COVID-19 living in cities with higher altitude could be due to adaptations in these compensatory adjustments to increase the bioavailability of peripheral oxygen. This adaptation is proven in individuals who live at altitudes above 3000 m a.s.l.
Other studies have also investigated the relationship between altitude, infection, and case fatality by COVID-19. In Peru, infection by COVID-19 reduced with increasing altitude. However, case-fatality rate was not dependent of altitude 20 . In addition, case studies in the USA and Mexico showed that mortality due to COVID-19 was greater in cities with altitude higher than 2000 m versus located lower than 1500 m 21 . A publication from Italy showed no association of COVID-19 with altitude 22 . In our study, RI and RDR became lower as the altitude increase (Fig. 4). In addition, although the case fatality rate (deaths/cases) decreased over the period analysed for all altitude classes, the reduction was greater at high altitudes (Fig. 5). In this way, such adaptations, even to a lesser extent, could contribute to minimizing the severity of infection in cities located at higher altitudes. www.nature.com/scientificreports/ The casual movement of dissolved oxygen molecules establishes the PO 2 of plasma and tissue fluids. The pressure of oxygen in the solution helps to regulate breathing, particularly at higher altitudes, when the ambient PO 2 decreases considerably; it also determines the uptake of oxygen by haemoglobin in the lungs and the subsequent release into the tissues. However, haemoglobin saturation with oxygen changes very little until the oxygen pressure has decreased to about 60 mmHg. Even when alveolar PO2 drops to 75 mmHg, as it does at high altitudes, haemoglobin saturation decreases by only approximately 6%. At 60 mmHg alveolar PO2, haemoglobin is still 90% saturated with oxygen. Below that pressure, the volume of oxygen combined with haemoglobin decreases more quickly.
As exposed, the S shape of the oxyhaemoglobin dissociation curve indicates that there is only a small change in the percentage saturation of haemoglobin with oxygen up to an altitude of approximately 3,048 m. At 1,981 m, for example, the alveolar PO2 falls from its value at sea level from 100 to 78 mmHg 3 . However, haemoglobin remains 90% saturated with oxygen.
In addition, regarding the possible physiological factors, the receptor-binding domain (RBD) in the SARS-CoV-2 protein was recently identified and that the RBD protein-bound firmly to the receptors of the human www.nature.com/scientificreports/ angiotensin-converting enzyme 2 (ACE2) 23 . Human ACE2 is part of the renin-angiotensin system (RAS), an essential hormonal system for controlling blood pressure and fluid and electrolyte balance. In the classical view, RAS peptides are generated from a single precursor protein called angiotensinogen (ATG). After being cleaved by the protease renin, this protein forms the inactive decapeptide angiotensin I, which is hydrolysed by the  www.nature.com/scientificreports/ angiotensin-converting enzyme (ACE) and forms the octapeptide angiotensin II (Ang II), the principal peptide in the system. ACE2 cleaves a single residue of angiotensin I (Ang I) that generates the Ang-(1-9) peptide and degrades Ang II to the Ang-(1-7) vasodilator. Current data obtained during the pandemic suggest that the use of ACE inhibitors and angiotensin type I receptor blockers increase the expression of ACE2. Consequently, increased expression of ACE2 would facilitate infection by COVID-19 24,25 . Studies show that RAS elements are modulated at high altitudes [26][27][28] . In this way, it is possible that the expression of ACE2 can be down-regulated due to the high altitude and favour a lower incidence of COVID-19 infection. Other studies speculate in this direction and suggest different paths for an ACE2 down-regulation 3,29 , but future studies need to clarify this hypothesis. Given the above, considering that the Brazilian cities studied at a higher altitude are around 1100 m 5 , probably the hypothesis of compensatory adjustments related to ACE2 in pulmonary epithelial cells, i.e., a protective factor for virus penetration and evolution of severe pulmonary edema, should be studied with caution in residents of Brazilian cities with higher altitudes. Furthermore, this population is also not exposed to conditions of chronic hypoxia. The results presented in this work can be useful for the implementation of public policies for prevention, control of the dissemination of COVID-19 in Brazil and the world. In addition, it can contribute to future studies, including other zoonotic viruses that cause respiratory diseases, as well as allowing the recommendation of changing the environment for people at risk in COVID-19.
Our findings identified that virulence by SARS-CoV-2 is lower in Brazilian cities with a population above 200 thousand inhabitants, located at relative high altitudes, and where the RH is lowest (Fig. 3). These findings are in line with the physiological compensatory adjustments of the inhabitants of cities located at higher altitudes, as well as with the common characteristics. Thus, our study starts point for future studies to establish causality of environmental conditions with SARS-CoV2, contributing to the implementation of measures to prevent and control the spread of COVID-19. It is, however, important to note that the information presented here clearly lacks any physiological evidences, which may merit further investigation. In addition, we cannot assume uniform density population in every city studied, as well quality of care, probability of infection, among other ecological variables, that may affect the outcome. In other words, future studies should consider those variables. As prospective, longitudinal studies are needed to confirm whether these associations remain over time.

Data availability
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.