Associations between solar and geomagnetic activity and peripheral white blood cells in the Normative Aging Study

Abstract

It has been hypothesized that solar and geomagnetic activity can affect the function of the autonomic nervous system (ANS) and melatonin secretion, which both may influence immune response. We investigated the association between solar geomagnetic activity and white blood cell counts in the Normative Aging Study (NAS) Cohort between 2000 and 2013. Linear mixed effects models with moving day averages ranging from 0 to 28 days were used to evaluate the effects of solar activity measures, Interplanetary Magnetic Field (IMF), and Sunspot Number (SSN), and a measure of geomagnetic activity, K_p Index (K_p), on total white blood cell (WBC), neutrophil, monocytes, lymphocyte, eosinophil, and basophil concentrations. Even after adjusting for demographic and health related factors, there were consistent significant associations between IMF, SSN, and K_p index, with reduced total WBC, neutrophils, and basophil counts that were stronger with longer moving averages. The associations were similar after adjusting for ambient air particulate pollution. Our findings suggest that periods of increased solar and geomagnetic activity result in lower WBC, neutrophil and basophil counts that may contribute to slight immune suppression.

1. Introduction

Solar activity encompasses the quasi-periodic oscillation of all electromagnetic radiation emitted from the Sun to the space environment and Earth (Mendoza & Sánchez de la Peña, 2010). Alterations in solar intensity, magnetic field fluctuation, and solar winds modulate disturbances of the Earth’s magnetic field, influencing biological and physiological responses of terrestrial life (Anderson, 1992; Barkhatov et al., 2008). Studies suggest variations in solar activity and Earth’s magnetic field disturbances (or geomagnetic disturbances) may influence the function of the autonomic nervous system (ANS) through the disruption of the circadian rhythms (Breus et al., 1995; McCraty et al., 2017). The ANS is tied to immune signaling as neurotransmitters released by the sympathetic and parasympathetic systems induce peripheral immune modulatory responses through receptors on immune cells (Kenney & Ganta, 2014). ANS dysregulation also plays a role in peripheral white blood cell count changes suggesting that solar and geomagnetic activity may influence circulating white blood cell and differential count (Scheiermann et al., 2013).

Circulating leukocytes include neutrophils, lymphocytes, monocytes, eosinophils, and basophils that are derived from hematopoietic stem cells found in bone marrow. Depending on cell type, peripheral leukocytes may reflect bone marrow and systemic response to infection or response to an allergen or medication. Hematopoietic malignancies, pre-malignant conditions, and viral infections can disrupt bone marrow function that may be reflected in circulating white blood cell and differential counts (Chandran et al., 2015). Few studies have investigated associations between solar or geomagnetic activity and immune response. Stoupel et al. (1995) compared immunoglobulin (IgA, IgG and IgM) levels during solar minimum and maximum, noting a decrease in all three immunoglobulin levels as solar maximum was reached. A follow up study showed variations in autoimmune disease biomarkers including anticardiolipin and lupus anticoagulant with changes in geomagnetic activity, showing possible clinical implications related to geomagnetic storms (Stoupel et al., 2006). Interactions with solar and geomagnetic activity also occur beyond the boundaries of Earth’s atmosphere. Astronauts experience higher exposures to solar activity from space weather and the responses observed provide insight for potential exposure relationships. Stowe et al. (1999) explored the relationship between short-term spaceflight and changes in neutrophil and lymphocyte counts. Leukocyte subset changes attributable to sympathetic nervous system effects were also observed among astronauts, even after adjusting for length of spaceflight (Stowe et al., 2003).

In this study we investigated the effects of solar [interplanetary magnetic field (IMF), sunspot number (SSN)] and geomagnetic activity [K_p Index] on immune dysregulation by evaluating changes in peripheral leukocytes and differential counts in a cohort of elderly healthy males living in the Greater Boston Area, Massachusetts, USA. We also assessed these effects after adjustment for fine particulate matter, particle number, black carbon, and particle radioactivity (i.e. radioactivity associated with ambient particles). To the best of our knowledge, this is the first epidemiologic investigation to provide a comprehensive assessment of the association between solar activity and peripheral white blood cell and differential count in an aging cohort.

2. Methods

2.1. Study Populations

Study participants were members of the Normative Aging Study (NAS) cohort, a closed cohort established by the U.S. Department of Veterans Affairs in 1963 (Bell et al., 1966). The NAS longitudinal study recruited men from the greater Boston area. At the beginning of the study, the participants’ mean age was 42 years (range 21–81 years). All participants were free of chronic conditions at baseline and attended clinical physical examinations every 3 to 5 years. The study was approved by the Institutional Review Boards of Harvard School of Public Health and the VA Boston Healthcare System. All cohort members provided written informed consent prior to the start of the study. Participants underwent clinical examinations under conditions of abstaining from smoking and an overnight fast. At each visit occurring between 2000 and 2013, patients had a full physical and laboratory analysis. Participants with an ever diagnosis or cause of death from any hematologic cancers (ICD-9 Codes 200 to 208) were excluded from the study, as these cancer diagnosis or treatment methodologies may effect WBC differential values (de Jonge et al., 2011). In the final analysis, 728 subjects with 2,048 total visits were eligible for inclusion.

2.2. Questionnaire assessment

Standardized questionnaires were administered to NAS cohort members at each visit to collect information about their health and behaviors. The comprehensive questionnaire assessed the participant’s complete medical history, family medical history, social behaviors (smoking and alcohol consumption), medications, food consumption frequency and quantitative health evaluation measures (height and weight). A full list of parameters has been previously described elsewhere (Bell et al., 1966; Hu et al., 1996)

2.3. Blood measurements

Following an overnight fasting period, venous blood samples were collected from all participants attending clinical visits. Differential and total WBC counts were analyzed using a Sysmex XN-10^™ system in the VA Boston hematology laboratory (De Labry et al., 1990). High sensitivity CRP was determined using an immunoturbidimetric assay on the Roche Cobas 6000 system (Roche Diagnostics - Indianapolis, IN), using reagents and calibrators from Roche. This high-sensitivity assay has a limit of detection of 0.03 mg/L.

2.4. Solar activity exposures

Measures of SSN, IMF, and K_p are proxy measures to quantify solar activity and geomagnetic activity. IMF captures the intensity of the sun’s magnetic field, a component of solar activity. SSN provides insights on the timepoint in the solar cycle, another aspect of solar activity. K_p represents disturbances is Earth’s magnetic field and geomagnetic activity. The Earth’s and Sun’s magnetic fields merge at a point called the magnetopause, where under specific conditions the IMF can affect the Earth’s magnetic field potentially generating geomagnetic disturbances (Lugaz, 2019). These three variables together constitute a larger picture of the interactions between the Sun and the Earth, providing additional data to quantify human health impacts.

SSN shown on the National Aeronautics and Space Administration (NASA) Goddard Space Flight Center’s OMNIWeb² website was obtained from the Belgium SILSO Center. SSN is a parameter used as a proxy for solar activity (Hathaway, 2015). Sunspots are dark areas with strong magnetic fields on the Sun’s surface and have a diameter ranging from 3,500 to 6,000 km (Solanki, 2003). Over ~11-year solar cycle periods, SSN rapidly increases and slowly falls throughout the solar cycle (Hathaway, 2015). The solar maximum is defined as the point of highest sunspot activity.

IMF is a direct measure of the portion of the Sun’s magnetic field that is carried by solar wind into interplanetary space. IMF measures presented on the NASA Goddard Space Flight Center’s OMNIWeb website were obtained using numerous spacecraft measurements. Total IMF intensity [expressed in nanotesla (nT)] measures the combined IMF strength for all directions.

K_p index is a parameter used to quantify geomagnetic activity. High fluctuations in geomagnetic activity are called geomagnetic disturbances (GMD), which mostly result from increases in solar activity. K_p index is measured on a 0 to 9 scale, where zero is interpreted as very little activity and 9 is extreme geomagnetic storms. K_p index was calculated as the average of minute-by-minute data samples taken over a three-hour period. K_p index values presented on the NASA Goddard Space Flight Center’s OMNIWeb website were obtained from the German Research Centre for Geosciences. The K_p index was measured using NASA’s mapping of standard daily K_p index.

2.5. Air pollution data

Daily concentrations of PM_2.5 in μg/m³ (particulate matter ≤2.5 μm in diameter) were measured using a Harvard impactor (Blomberg et al., 2020; Vieira et al., 2020). Continuous measurements of Black Carbon (BC) in μg/m³ were analyzed using an Aethalometer (Magee Scientific Company, model AE-16, Berkeley, CA). Continuous measurements of particle number (PN) concentrations were analyzed using a condensation particle counter (CPC, TSI Inc. Model 3022a, Shoreview, MN). All measurements were taken at the Harvard Supersite, located on the roof of the Harvard Medical School’s Countway Library of Medicine. This site is approximately three miles from downtown Boston and approximately 1 km from the Hospital were NAS cohort members were examined.

2.6. Particle Radioactivity

Particle gross β-activity levels were obtained from the U.S. EPA RadNet monitoring network. The network includes 135 sampling stations, which are equipped with high-volume air samplers. Total Suspended Particles (TSP) were collected on filters (10-cm-diameter synthetic fiber). Particle gross β activity was measured 5 hours after the end of sampling to allow for decay of short-lived radon progeny. Moving averages of samples over multiple days were calculated. Particle gross β-activity exposures (mBq/m³) were estimated using data from the Boston, MA, RadNet site. These values were then imputed to generate a daily Log(β) activity variable (Blomberg et al., 2020; Nyhan et al., 2018).

2.7. Weather parameters

Daily local weather data, including ambient temperature (°C) and relative humidity (%) were obtained from the National Oceanic Atmospheric Administration’s, National Climatic Data Center (NOAA).

2.8. Exposure Windows

Short- and medium-term exposure windows were calculated for SSN, IMF, and K_p index for using moving averages for day 0 (the day of visit), and up to 28 days after the visit. As solar activity occurs far from the Earth’s surface there is an expected lag between the solar event occurrence and the exposure metrics. The exposure window was limited to one month to eliminate potential confounding by seasonal changes in solar exposures or additional exposures.

2.9. Statistical Analysis

The effects of solar activity parameters and immune outcomes were estimated using mixed-effects linear models with a random intercept for each participant to account for longitudinal correlation and multiple measurements for each participant at different visits. All models were adjusted for the following variables: day of year (DOY), day of the week (DOW), ambient temperature (° C), relative humidity (%), and seasonality³. We also included the following individual predictors: age at time of sample, estimated glomerular filtration rate (eGFR) based on the chronic kidney disease epidemiology collaboration equation (CKD-EPI)⁴, and pack years of smoking cigarettes. Categorial variables included in the model consisted of: race (white or non-white), lung diseases⁵ (yes versus no), diabetes mellitus (yes versus no), history of any chronic heart disease (yes versus no), smoking history (categorical: never, current and former), and two or more alcoholic drinks per day (yes versus no) (Levey et al., 2009). A sensitivity analysis was performed excluding samples with CRP levels greater than 10 mg/L and samples where a CRP measurement was not performed to assess if observed associations remained consistent. An additional sensitivity analysis was performed excluding any participant visits that took place outside of Massachusetts. To ensure consistency of results an additional sensitivity analysis was performed using two categorical solar variable assignments. The first category consisted of solar exposures below the median for each moving day average and the second category consisted of solar exposures above the median for each moving day average. Each of the 28 moving day averages had a different median values so categorical assignments were conducted based on moving day averages for each of the solar variables

In post-hoc analysis, we additionally adjusted our primary model for air pollution and radioactivity variables, one at a time to understand the potential effects of these environmental exposures on the solar activity and immune response relationship. This was determined by first running a mixed linear model separately for PM_2.5, BC, PN, or Log(β) estimating effects on immune cell counts using up to 28-day moving averages without including solar radiation variables. This was done to determine the most significant moving day average for each air pollution variable, that would be adjusted for in the final models. The moving day average with the lowest p-value for each air pollution variable was selected for each individual immune outcome. The air pollutant moving day averages, effect estimates, and associated p-values are shown in table S1.

The generic structure of the post-hoc model can be represented as:

$$\begin{gathered} {\mathit{\text{Immune Outcome (WBC, MONO, LYMPH, or EOSIN)}}}_{ij}\text{ = }{B}_{\mathit{0}}\text{ + }{b}_{\mathit{0}i}\text{ + }{B}_{\mathit{1}}{\mathit{\text{ solar variable}}}_{ij}\text{ + }{B}_{\mathit{2}}\mathit{\text{ air pollution}} \\ {\mathit{\text{variable}}}_{ij}\text{ + }{B}_{\mathit{3}}{\mathit{\text{ temperature}}}_{ij}\text{ + }{B}_{\mathit{4}}{\mathit{\text{ humidity}}}_{ij}\text{+ }{B}_{\mathit{5}}{\mathit{\text{ DOW}}}_{ij}\text{ + }{B}_{\mathit{6}}\text{ (}sin\text{(2*}pi\text{*}DO{Y}_{ij}\text{/365)) +} \\ {B}_{\mathit{7}}\text{ (}cos\text{(}\mathit{2}\text{*}pi\text{*}DO{Y}_{ij}\text{/365)) + }{B}_{\mathit{8}to\mathit{17}}{\mathit{\text{ Individual predictors}}}_{ij}\text{ + }{e}_{ij} \end{gathered}$$

Where for each individual i and visit j; B₀ is the model intercept, b_0i is the random intercept for each subject, B₁ is the coefficient for each solar activity parameter included one at a time (lagged up to 28-days moving average), B₂ is a the coefficient for air pollution and radioactivity variables included one at a time (each with a selected moving average lag), B₃ and B₄ are coefficients for temperature and relative humidity, B₅ to B₇ are coefficients for seasonal variables, B_{8 to 17} are vectors of coefficients for 9 individual-level factors, and e_ij is the residual error.

The results are presented as cell count per interquartile range (IQR) increase in solar activity variables. The residuals of models were examined for potential deviations from linearity. We chose compound-symmetry variance-covariance structure to handle potential of unbalanced longitudinal data. All analyses were performed using R software version 3.6.3.

3. Results

3.1. Descriptive Results

The descriptive statistics of the NAS Cohort study subjects for all visits and for the first visit are shown in Table 1. There were 728 individual subjects, with 2,048 visits between 2000 and 2013. Of the 728 subjects, 728 presented for one visit, 537 presented for 2 visits, 413 presented for 3 visits, 286 presented for 4 visits, 87 presented for 5 visits, and 2 presented for 6 visits. At baseline, the mean age was 75.8 (±6.9) years and 98% of participants were white. The study population had a mean body mass index of 27.9 (±4.1) kg/m³. Of the 728 individuals 222 (30.5%) never smoked, and 590 (81.0%) had less than two drinks per day. Over all visits the study population had a mean total WBC count of 6,347 (±1,854) thousands/mm³, a mean absolute neutrophil count of 3,986 (±1,312) thousands/mm³, a mean absolute monocyte count of 542 (±169) thousands/mm³, a mean absolute lymphocyte count of 1,575 (±1,042) thousands/mm³, a mean absolute eosinophil count of 214 (±154) thousands/mm³, and a mean absolute basophil count of 38.0 (±35.2) thousands/mm³. These values for absolute peripheral leukocytes counts are within the normal range and do not suggest a baseline hyper or hypo-inflammatory state.

Table 1:

Characteristics of the study population in the Normative Aging Study and weather (2000–2013)

	All Visits	Visit 1
	N	N
Total Unique Observations	2,048	728
Total Unique IDs	728	728
	Mean ± SD	Mean ± SD
Age (y)	75.8 ± 6.9	73.0 ± 6.9
BMI (kg/m3)	27.9 ± 4.1	28.2 ± 4.0
Cigarette Pack years (y)	20.3 ± 24.0	21.6 ± 25.6
Estimated Glomerular Filtration Rate (mL/min/1.73m2)	68.1 ± 16.1	71.3 ± 15.6
Relative Humidity (%)	68.0. ± 17.4	69.6 ± 17.4
Ambient Temperature (°C)	12.8 ± 8.8	12.8 ± 8.4
White Blood Cells (cells/mm3)	6,347 ± 1,854	6,313 ± 2,109
Neutrophils (cells/mm3)	3,986 ± 1,312	3,934 ± 1,294
Monocytes (cells/mm3)	542 ± 169	535 ± 163
Lymphocytes (cells/mm3)	1,575 ± 1,042	1,604 ± 1,423
Eosinophils (cells/mm3)	214 ± 154	209 ± 141
Basophils (cells/mm3)	38.0 ± 35.2	33.9 ± 32.5
	n(%)	n(%)
Race
White	1986 (98.0)	707 (98.1)
Nonwhite	42 (2.0)	14 (1.9)
Coronary Heart Disease	717 (35.0)	218 (29.9)
Diabetes	223 (10.9)	65 (8.9)
<Two Drinks/Day	1,674 (81.7)	590 (81.0)
Smoking Status
Current	87 (4.3)	35 (4.8)
Former	1,318 (64.4)	471 (64.7)
Never	643 (31.4)	222 (30.5)
Lung Function
Diagnosis of chronic bronchitis, emphysema or asthma	370 (18.1)	133 (18.3)
No diagnosis of chronic bronchitis, emphysema or asthma	1,678 (81.9)	595 (81.7)

The descriptive statistics of the solar activity measures (IMF and SSN) and geomagnetic disturbances (K_p) are shown in Table 2. During the dates of analysis, the mean IMF was 5.9 (±3.0) nT, the mean K_p Index was 19.1 (±12.2), and the mean SSN were 85.88 (±74.6). Over the course of the study period solar activity and geomagnetic disturbances vary due to both the 11-year solar cycle and the location of Earth. During the years 2000 to 2013 maximum IMF of 29.2 nT and the KP maximum of 80.9 occurred on October 10, 2003. The maximum SSN of 351.5 occurred on July 19, 2000. Variations in solar activity and geomagnetic disturbances are shown in Figure 1.

Table 2:

Distribution of solar activity measures, particulate air pollution variables, and β activity (2000–2013)

	Mean ± SD	Min	Max	IQR
IMF (nT)	5.9 ± 3.0	1.8	29.2	3.0
Kp Index*	19.1 ± 12.2	0.0	80.9	16.2
SSN	85.8 ± 74.6	0.0	351.5	117
PM2.5 (μg/m3)	9.9 ± 6.6	1.1	58.4	6.6
Black Carbon (μg/m3)	0.8 ± 0.4	0.2	2.5	0.5
Particle Number (#/cm3)	21,978 ± 12,163	3,447	92,400	16137.9
Log β Activity*	−5.9 ± 0.3	−5.9	−4.1	0.4

^*Dimensionless

Figure 1:

Variations in solar variables over the 13 year study period

3.2. Associations of Solar and Geomagnetic Activity Variables and Immune Outcomes

There was a significant inverse association between solar and geomagnetic activity variables (IMF, SSN, and K_p index) and absolute total WBC, NEUT, and BASO counts. With greater measures of solar and geomagnetic activity there were lower concentrations of total WBC, NEUT, and BASO counts. The largest magnitude of effect was observed in models with IMF as the exposure. While still significant, SSN and K_p index had smaller effect estimates per IQR of exposure for total WBC, absolute NEUT, and absolute BASO counts. Trends showed a greater magnitude of effect over time with greater effects for longer moving day averages. All observed trends and associations remained consistent after adjusting for ambient air pollution, based on the most significant air pollution moving day average. A full listing of leukocyte results is shown in Tables 3, 4, and 5. The sensitivity analysis excluding CRP >10 and no CRP, as well as visits taking place outside of Massachusetts did not change the magnitude or direction of association.

Table 3:

Changes in WBC (cells/mm³) associated with IQR increases in exposure variables (95% CI) for 2048 clinical visits (n=728)

Exposure	Moving Average	No Air Pollution Adjustment	PM2.5	BC	PN	Log β
IMF(nt)	0	−72.472(−198.576 , 53.633)	−63.263(−189.528 , 63.862)	−77.461(−233.921 , 48.999)	−29.656(−177.776 , 118.464)	−63.468(−183.502 , 56.566)
	1	−83.801(−234.613 , 66.415)	−75.383(−225.742 , 74.976)	−82.494(−233.432 , 68.343)	−45.085(−211.924 , 119.854)	−79.537(−222.461 , 63.388)
	7	−200.066(−439.661 , 39.528)	−188.557(−428.955 , 51.798)	−177.945(−418.479 , 62.578)	−102.601(−368.196 , 162.994)	−131.081(−362.323 , 151.161)
	14	−165.750(−446.462 , 114.962)	−154.732(−437.422 , 127.958)	−124.189(−487.649 , 159.229)	−33.307(−357.486 , 283.874)	−114.357(−387.669 , 158.894)
	21	−201.607(−542.916 , 99.691)	−184.116(−487.824 , 119.592)	−145.379(−449.976 , 159.219)	−86.020(−427.971 , 255.932)	−180.231(−473.972 , 113.568)
	28	−206.012(−515.592 , 193.067)	−186.701(−497.896 , 124.495)	−141.134(−454.376 , 172.147)	−105.385(−459.485 , 248.715)	−181.142(−483.638 , 121.355)
Kp	0	−1.569(−4.642 , 5.952)	−1.367(−3.854 , 1.121)	−1.523(−3.991 , 5.944)	−0.976(−3.591 , 1.639)	−1.316(−3.665 , 1.233)
	1	−1.769(−4.528 , 4.989)	−1.619(−4.393 , 1.156)	−1.657(−4.457 , 1.892)	−1.144(−3.998 , 1.712)	−1.391(−4.117 , 1.228)
	7	−1.479(−5.699 , 2.742)	−1.330(−5.568 , 2.948)	−1.058(−5.261 , 3.144)	0.107(−4.286 , 4.501)	−0.766(−4.798 , 3.265)
	14	−0.744(−5.688 , 4.199)	−0.594(−5.555 , 4.366)	−0.338(−5.263 , 4.587)	0.048(−5.163 , 5.258)	−0.420(−5.181 , 4.345)
	21	−2.784(−8.576 , 2.536)	−2.741(−8.984 , 2.649)	−2.130(−7.418 , 3.149)	−1.939(−7.465 , 3.587)	−2.524(−7.648 , 2.599)
	28	−2.811(−8.326 , 2.753)	−2.779(−8.363 , 2.854)	−2.076(−7.582 , 3.431)	−1.710(−7.462 , 4.242)	−2.167(−7.512 , 3.179)
SSN	0	−0.439(−7.855 , −0.682)	−0.421(−4.783 , −6.568)	−0.342(−8.715 , 1.739)	−0.570(−8.986 , −0.156)	−0.447(−5.791 , −9.194)
	1	−0.432(−1.794 , −1.577)	−0.412(−1.776 , −7.349)	−0.334(−7.779 , 7.446)	−0.554(−9.971 , −9.139)	−0.440(−5.786 , −5.395)
	7	−0.414(−3.793 , −1.736)	−0.397(−5.777 , −5.118)	−0.310(−6.792 , 9.988)	−0.498(−1.932 , −8.464)	−0.433(−3.795 , −5.472)
	14	−0.405(−0.798 , −8.812)	−0.382(−9.777 , 1.211)	−0.298(−0.736 , 8.118)	−0.483(−2.938 , −5.528)	−0.416(−4.793 , −7.447)
	21	−0.445(−6.851 , −1.546)	−0.412(−1.819 , −3.255)	−0.325(−3.747 , 0.097)	−0.544(−1.319 , −1.374)	−0.455(−2.844 , −3.566)
	28	−0.428(−6.841 , −7.116)	−0.390(−0.854 , 4.824)	−0.296(−1.726 , 5.134)	−0.525(−1.009 , −0.841)	−0.435(−0.831 , −7.545)

^*All models were adjusted for seasonality based on day of year (DOY), DOW, ambient temperature, relative humidity, age, race, eGFR, lung diseases, diabetes mellitus, history of any chronic heart disease, smoking history, two or more alcoholic drinks per day, pack years of smoking cigarettes, ambient air pollutant exposures (no air pollution, PM2.5, BC, PN, or Log β) and solar variable (IMF(nt), K_p or SSN)

Table 4:

Changes in NEUT (cells/mm³) associated with IQR increases in exposure variables (95% CI) for 2048 clinical visits (n=728)

Exposure	Moving Average	No Air Pollution Adjustment	PM2.5	BC	PN	Log β
IMF(nt)	0	−90.568(−188.841 , 7.773)	−83.192(−181.644 , 15.261)	−90.870(−189.559 , 7.819)	−57.382(−158.789 , 44.125)	−69.220(−168.361 , 29.925)
	1	−100.429(−217.335 , 16.477)	−92.680(−269.773 , 24.411)	−95.471(−212.948 , 22.316)	−71.436(−192.986 , 54.113)	−74.015(−191.956 , 43.926)
	7	−194.350(−378.577 , −18.125)	−178.754(−364.116 , 6.521)	−168.813(−354.749 , 17.124)	−129.618(−336.863 , 71.626)	−136.604(−325.571 , 51.792)
	14	−207.210(−421.576 , 7.584)	−184.863(−431.518 , 31.791)	−163.683(−381.964 , 54.598)	−117.792(−367.877 , 125.292)	−126.808(−347.938 , 94.321)
	21	−246.131(−475.341 , −16.921)	−220.306(−452.398 , 11.786)	−191.817(−425.836 , 42.222)	−169.096(−433.136 , 94.945)	−158.682(−395.437 , 78.273)
	28	−262.875(−497.562 , −28.199)	−233.778(−471.292 , 3.734)	−201.956(−442.488 , 38.576)	−196.781(−469.627 , 76.745)	−171.720(−415.664 , 71.623)
Kp	0	−1.560(−3.499 , 2.378)	−1.453(−3.464 , 4.497)	−1.483(−3.424 , 2.458)	−1.197(−3.192 , 8.796)	−1.175(−3.129 , 2.779)
	1	−1.635(−3.798 , 4.527)	−1.558(−3.733 , 0.617)	−1.482(−3.644 , 5.679)	−1.396(−3.638 , 1.837)	−1.282(−3.461 , 8.896)
	7	−2.261(−5.544 , 1.022)	−2.087(−5.388 , 1.214)	−1.786(−5.669 , 1.496)	−1.272(−4.713 , 2.159)	−1.626(−4.954 , 1.712)
	14	−2.823(−6.659 , 1.911)	−2.517(−6.376 , 1.349)	−2.251(−6.393 , 1.596)	−2.012(−6.044 , 2.228)	−1.724(−5.644 , 2.195)
	21	−4.431(−8.525 , −1.343)	−4.201(−8.338 , −3.564)	−3.693(−7.798 , 5.411)	−3.922(−8.243 , 8.398)	−3.225(−7.429 , 9.977)
	28	−4.454(−8.732 , −7.207)	−4.261(−8.568 , 5.745)	−3.653(−7.922 , 4.615)	−3.891(−8.413 , 8.621)	−3.124(−7.492 , 1.243)
SSN	0	−0.328(−7.655 , −6.252)	−0.296(−8.576 , −4.018)	−0.244(−9.534 , 0.244)	−0.330(−5.665 , −4.001)	−0.260(−4.548 , 3.919)
	1	−0.316(−8.595 , −1.738)	−0.283(−8.564 , −8.923)	−0.231(−2.522 , 9.962)	−0.318(−2.649 , 5.013)	−0.247(−9.529 , 4.333)
	7	−0.302(−4.593 , −8.512)	−0.268(−4.561 , 5.224)	−0.210(−1.514 , 9.193)	−0.276(−9.629 , 7.375)	−0.225(−8.525 , 6.768)
	14	−0.291(−1.593 , 7.411)	−0.252(−6.557 , 8.352)	−0.193(−5.511 , 3.123)	−0.256(−7.629 , 2.115)	−0.204(−3.511 , 0.142)
	21	−0.308(−6.629 , 1.373)	−0.263(−5.577 , 2.051)	−0.198(−9.527 , 6.129)	−0.295(−7.685 , 1.693)	−0.220(−4.537 , 3.696)
	28	−0.301(−5.618 , 3.915)	−0.252(−5.572 , 4.267)	−0.182(−1.516 , 1.152)	−0.292(−5.698 , 6.145)	−0.211(−9.533 , 3.179)

^*All models were adjusted for seasonality based on day of year (DOY), DOW, ambient temperature, relative humidity, age, race, eGFR, lung diseases, diabetes mellitus, history of any chronic heart disease, smoking history, two or more alcoholic drinks per day, pack years of smoking cigarettes, ambient air pollutant exposures (no air pollution, PM2.5, BC, PN, or Log β) and solar variable (IMF(nt), K_p or SSN)

Table 5:

Changes in BASO (cells/mm³) associated with IQR increases in exposure variables (95% CI) for 2048 clinical visits (n=728)

Exposure	Moving Average	No Air Pollution Adjustment	PM2.5	BC	PN	Log β
IMF(nt)	0	−5.459(−8.729 , −2.189)	−5.235(−8.699 , −1.771)	−5.293(−8.594 , −1.997)	−4.075(−7.478 , −8.674)	−5.782(−9.699 , −2.466)
	1	−8.348(−12.225 , −4.471)	−8.193(−12.157 , −4.228)	−8.151(−12.963 , −4.241)	−6.477(−13.594 , −2.361)	−8.830(−12.767 , −4.894)
	7	−13.504(−19.524 , −7.483)	−11.756(−17.913 , −5.642)	−12.190(−18.336 , −6.551)	−9.311(−16.137 , −2.485)	−14.593(−23.813 , −8.384)
	14	−16.010(−22.947 , −9.173)	−13.474(−22.656 , −6.293)	−14.251(−21.369 , −7.133)	−10.840(−19.051 , −2.629)	−17.675(−24.997 , −14.445)
	21	−17.605(−24.994 , −11.216)	−14.875(−22.586 , −7.163)	−15.949(−23.536 , −8.361)	−11.729(−23.657 , −2.862)	−19.653(−27.367 , −11.946)
	28	−17.503(−25.761 , −9.947)	−14.622(−22.521 , −6.724)	−15.741(−23.494 , −7.987)	−11.085(−21.295 , −1.885)	−19.823(−27.748 , −11.899)
Kp	0	−0.132(−8.198 , −8.668)	−0.127(−8.195 , −3.861)	−0.127(−0.192 , −8.962)	−0.115(−7.183 , −3.249)	−0.135(−0.201 , −3.969)
	1	−0.154(−5.227 , −7.282)	−0.147(−1.221 , −6.173)	−0.150(−2.223 , −4.878)	−0.128(−6.244 , −1.653)	−0.157(−7.231 , −5.585)
	7	−0.235(−7.344 , −4.127)	−0.197(−7.311 , −1.785)	−0.219(−0.329 , −3.139)	−0.179(−0.294 , −4.164)	−0.240(−5.351 , −5.131)
	14	−0.263(−1.389 , −0.137)	−0.217(−0.349 , −1.485)	−0.243(−3.371 , −7.116)	−0.194(−5.329 , −3.767)	−0.279(−0.439 , −6.149)
	21	−0.291(−0.425 , −5.157)	−0.245(−5.387 , −0.134)	−0.269(−4.435 , −3.134)	−0.207(−1.351 , −1.163)	−0.318(−4.458 , −4.179)
	28	−0.304(−9.444 , −1.166)	−0.259(−1.446 , −6.112)	−0.282(−7.423 , −3.142)	−0.209(−6.368 , −1.564)	−0.332(−5.477 , −1.188)
SSN	0	−0.016(−2.125 , −0.807)	−0.015(−0.024 , −4.506)	−0.013(−9.123 , −7.475)	−0.007(−0.618 , 6.234)	−0.017(−1.126 , −2.398)
	1	−0.016(−2.425 , −3.357)	−0.015(−0.024 , −8.956)	−0.013(−8.323 , −0.614)	−0.006(−9.018 , 6.604)	−0.017(−1.526 , −3.098)
	7	−0.015(−9.325 , −5.556)	−0.014(−7.124 , −1.765)	−0.013(−2.323 , −6.353)	−0.005(−3.517 , 7.387)	−0.017(−0.927 , −3.817)
	14	−0.017(−2.427 , −5.307)	−0.015(−7.926 , −6.396)	−0.014(−4.225 , −8.984)	−0.005(−9.519 , 8.157)	−0.018(−5.529 , −1.818)
	21	−0.018(−7.329 , −5.909)	−0.016(−9.927 , −1.127)	−0.015(−8.326 , −2.285)	−0.007(−3.521 , 6.156)	−0.020(−1.731 , −7.617)
	28	−0.019(−1.129 , −0.019)	−0.017(−3.228 , −0.697)	−0.016(−0.927 , −7.955)	−0.007(−5.821 , 4.736)	−0.020(−6.131 , −8.510)

^*All models were adjusted for seasonality based on day of year (DOY), DOW, ambient temperature, relative humidity, age, race, eGFR, lung diseases, diabetes mellitus, history of any chronic heart disease, smoking history, two or more alcoholic drinks per day, pack years of smoking cigarettes, ambient air pollutant exposures (no air pollution, PM2.5, BC, PN, or Log β) and solar variable (IMF(nt), K_p or SSN)

Mixed relationships were observed for other non-significant blood cell outcomes (MONO, LYMPH, and EOSIN). Inverse associations were observed between IMF, K_p and MONO counts, while a positive associated was seen between SSN and MONO. Positive associations were observed between IMF and K_p and LYMPH counts, while an inverse association was seen between SSN and LYMPH. A positive association was also seen between IMF and EOSIN, as well as with K_p and EOSIN, while an inverse association was seen between SSN and EOSIN. Full results for MONO, LYMPH, and EOSIN are shown in the supplementary materials tables S2, S3, and S4.

The results of the sensitivity analyses of the categorical solar variables are similar in direction and significance to those of the main linear mixed-effects model analysis, but offer a stronger magnitude of effect estimates. The full results of the categorical sensitivity analysis are shown in Tables S5 to.

3.3. White Blood Cells

The associations for WBC followed a downward trend across the 0–28 day moving averages with the largest magnitude of effects coinciding with 28^th moving day average. WBC counts decreased by 0.429 cells/mm³ (95% CI: −0.841, −0.016) (p=0.041) for every IQR increase in SSN on the 28^th moving day average (Figure 2). While no significant associations were observed between IMF and WBC or KP and WBC similar trends were seen with the largest magnitude of WBC count decreases occurring on the 28^th moving day average. IMF and KP both induced a larger magnitude of decrease in total number of WBCs; however, these changes only resulted in significant changes of neutrophils and basophils and not significant changes in the overall total WBC count.

Figure 2:

The change in White Blood Cells (WBC) per IQR increases in SSN over moving day average exposure windows of 0, 1, 7, 14, 21 and 28 days

3.4. Granulocytes

The associations for NEUT and BASO followed similar downward trend across the 0–28 day moving averages with the largest magnitude of effects coinciding with 28^th moving day average NEUT counts decreased by 262.88 cells/mm³ (95% CI: −497.56, −16.92) (p=0.028) for every IQR increase in IMF on the 28^th moving day average. Significant associations were also observed between KP and NEUT concentrations with a 4.45 cells/mm³ (95% CI: −8.70, −0.21) (p=0.039) decrease in NEUT counts per IQR increase in KP on the 28^th moving day average (Figure 3).

Figure 3:

The change in Neutrophils (NEUT) per IQR increases in Solar Variables over moving day average exposure windows of 0, 1, 7, 14, 21 and 28 days (y-axis scales vary based on solar variable)

BASO counts decreased by 17.50 cells/mm³ (95% CI: −25.06, −9.95) (p<0.001) for every IQR increase in IMF on the 28^th moving day average. Significant associations were also observed between SSN and BASO concentrations with a 0.019 cells/mm³ (95% CI:−0.029, −0.009) (p<0.001) decrease in BASO counts per IQR increase in SSN on the 28^th moving day average. For every IQR increase in K_p Index there was a 0.30 cells/mm³ (95% CI: −0.44, −0.17) (p<0.001) change in BASO counts on the 28^th moving day average (Figure 4).

Figure 4:

The change in Basophils (BASO) per IQR increases in Solar Variables over moving day average exposure windows of 0, 1, 7, 14, 21 and 28 days (y-axis scales vary based on solar variable)

4. Discussion

This study demonstrates an association between greater exposure to solar and geomagnetic activity and lower peripheral leukocytes counts, with effects on granulocyte cells types (NEUT, BASO) The largest magnitudes of effect (and lowest standard errors) were observed with greater cumulative exposures as suggested by greater effects seen with longer moving day averages. The findings are consistent with suppression of bone marrow granulocyte precursor cells that ultimately differentiates into neutrophils, basophils, and eosinophils. The decrease in these cell types are consistent with bone marrow effects caused by other exogenous exposures, such as some types of medications (i.e chemotherapy) or radiation exposures (Banfi et al., 2001; Kamiya et al., 2015). Bone marrow effects are also observed in several hematologic conditions, such as myelodysplastic syndrome or multiple myeloma, a diseases more commonly observed in older individuals (Friberg & Karlsson, 2003). Effects of myelosuppression include an increased risk of infection and bleeding complications (Carey, 2003). The association with neutrophils and other granulocytes, which are short-lived cells in comparison to lymphocytes are consistent with the week to month-s long exposure windows we are examining as compared to longer exposure periods. Biological systems and associated immune system responses are known to be influenced by solar and geophysical cycles through the mechanism of melatonin production, circadian rhythm disruption, and autonomic nervous system signaling.

One possible explanation for the effects of solar activity on peripheral absolute leukocyte counts could be due circadian rhythm disruption and ANS dysregulation (Wyse et al., 2020). The suprachiasmatic nucleus (SCN), part of the ANS situated at the base of the hypothalamus, is the principal circadian pacemaker in the human brain. As such, it regulates circadian rhythms during rest and activity, core body temperature, neuroendocrine function, autonomic function, memory, psychomotor performance, and a host of other behavioral and physiological processes. It is thus conceivable that SCN signaling may also have an impact on immune system regulation (Swaab, 2003).

The ANS and immune system communicate through an integrative interface, both the hypothalamic-pituitary-adrenal (HPA) axis and the sympathetic nervous system (SNS). (Elenkov et al., 2000). Activation of the ANS triggers pro-inflammatory responses, while increases in cytokines also results in increased vagus nerve activity (Ek et al., 1998; Saper et al., 2012). Multiple neural axes bridge the relationship between cytokine production and ANS activation, which in turn modulates the peripheral immune response. Neurotransmitters released by ANS nerve endings bind to receptors on immune cells, initiating an immune response (Kenney & Ganta, 2014). The SNS directly innervates lymphatic organs in signaling an immune response (Friedman & Irwin, 1997). This direct link between the nervous system and immune organs modulates immune response through the binding of norepinephrine, epinephrine, and dopamine to lymphocyte receptors, released by the SNS (Irwin, 1993). Environmental factors modulating changes in either the immune system or the nervous system, can have implications for long-term health including stress, depression, and overall immune function (Haskó & Szabó, 1998). Immune system effects also contribute to overall health via the ANS and SNS alterations resulting in hypertension, heart failure and fever (Ferrario & Strawn, 2006; Francis et al., 2004; Kang et al., 2009; Shi et al., 2010).

The pineal gland, responsible for melatonin secretion, is modulated by, and connected to, the SCN (Borjigin et al., 2012; Swaab, 2003). Melatonin is a biochemical factor modulating control of human circadian rhythms. The primary role of melatonin is synchronizing eukaryotic systems with associated photo periods (Currier et al., 2000). Dysregulation or removal of the pineal gland has a different effect on immune function, leading to an increase in total WBC count (Ana B. Rodriguez & Lea, 1994). Circadian fluctuations imposed by the SCN trigger fluctuations in melatonin secretion, often having direct implications for immune signaling or recruitment of inflammatory markers (A. B. Rodriguez et al., 1999; Scheiermann et al., 2013).

Variations of the geomagnetic field have been documented to induce changes in the biological response of animals and humans (Krylov et al., 2014; Touitou & Selmaoui, 2012). In addition, solar activity and geomagnetic field disturbances may reduce melatonin levels and impair the immune inflammatory responses to oxidative stress (Burch et al., 1999). Previous studies suggest an association between changes in geomagnetic activity and electromagnetic fields, decreased melatonin secretion and adverse physiological responses (Griefahn et al., 2002; Halgamuge, 2013; Weydahl et al., 2001).

Melatonin is a direct actor in immune system response through the stimulation of T-Lymphocytes, interleukin-2 and -6, and natural killer cells (Cherry, 2002; Peña et al., 2007; Szczepanik, 2007). The role of melatonin levels on immune system outcomes is thought to be modulated by cytokines. Melatonin receptors in T-Cells are a part of the melatonin-cytokine cascade, which induces the release of hematopoietic growth factors in bone marrow stimulated by interleukin production (Currier et al., 2000; García-Mauriño et al., 1999; Maestroni, 1995). Melatonin secretion, occurring primarily at night, has been linked both to inflammation and phagocytosis in animal studies (Hriscu, 2005; Terrón et al., 2004). Factors including alcohol, consumption, BMI, and gender differences have also been shown to alter nocturnal melatonin secretion (Cocco et al., 2005). Evidence also suggests that melatonin may delay leukocyte apoptosis and counteract the effects of age related oxidative stress in cells (Espino et al., 2010, 2011). Multiple links between melatonin and innate immune system responses, specifically responses in neutrophils and eosinophils have been investigated, thereby impacting total WBC counts (Calvo et al., 2013).

Neutrophils are a key component of the innate immune system, making up the largest percentage of the total white blood cell count (Kobayashi et al., 2005; Nauseef, 2007). In individuals taking melatonin supplements, an increased neutrophil response was observed along with increased intracellular chemokine expression; however, this response was not observed in monocytes or lymphocytes (Peña et al., 2007). In vitro studies have suggested a potential pathway by which melatonin contributes to the dysregulation of neutrophils, by inhibiting enzymes needed by L-selectin cleavage (Recchioni et al., 1998). L-selectin is a glycoprotein and adhesion molecule expressed on leukocytes, contributing to adhesion, migration, and signal transduction. Interestingly, L-selectin is expressed by all leukocytes except by activated, memory lymphocytes and thus may by an explanation for this observation (Ivetic et al., 2019).

Multiple additional pathways may influence the relationship between solar activity and peripheral leukocyte counts. Electromagnetic fields (EMF) have been shown to alter the action of cell membrane gates and flow through cation channels (Balcavage et al., 1996). Low frequency EMF has been shown to influence the interactions between leukocytes and the vascular endothelium (Ushiyama & Ohkubo, 2004). In mice exposed to low frequency EMF significant decreases in total leukocyte counts were observed (Bonhomme-Faivre et al., 1998). Alterations in membrane gate action coupled with direct EMF influences on leukocytes may also influence the relationship between solar activity and leukocytes leading to the observed effects.

Individual health susceptibility to solar radiation may vary temporally (daily or seasonal) and be influenced by solar activity periods. Furthermore, solar activity also appears to react synergistically with other environmental risk factors, such as ambient air pollution (Grandi & D’Ovidio, 2019). Solar activity is thought to play a major role in the atmospheric photochemical reactions, which affects the formation and concentration of aerosols. This reaction could intensify existing air pollutant concentration, specifically, quantitative measures of ultrafine particles (Ma & Birmili, 2015). Increased solar activity has been associated with increases in PN, with demonstrated seasonal fluxes (Vieira et al., 2020). Ambient air pollution also induces effects on peripheral leukocyte count (Steenhof et al., 2014). However, our results showed consistent pattern despite controlling for air pollutants.

Strengths & Limitations:

Solar activity measures do not account for indoor, medical, or anthropogenic sources of radiation exposure, which may contribute to the noted effects in immune response. This study was also limited to the analysis of one solar cycle, whereas a more robust exposure-outcome pathway would consider the variability in radiation measures between different solar cycles and control for this potential source of bias using data from two solar cycles, over a 22-year time window. In using the established NAS cohort, the results are limited to those observed in white men with an average age of 75.8 years and may not be generalizable to the entire population; however, the use of this limited population also prevents the introduction of additional confounding variables. This paper is among the first to investigate the epidemiological associations between solar activity, geomagnetic activity and peripheral leukocytes in an aging cohort. This research provides a basis for future research on human health effects of continuous environmental exposures.

5. Conclusions

Overall, we found that different parameters representing solar and geomagnetic activity were associated with a significant decrease in absolute white blood cell counts, absolute neutrophil counts, and absolute basophil counts thus alluding to the idea that increased solar and geomagnetic activity may have a generalized immune-suppressive effect. The results of this study represent a novel approach to understanding the implication of solar activity on the immune system. Multiple exposure metrics of solar activity were used to evaluate how variations in activity may impact immune outcomes after controlling for health and lifestyle confounders. Further research into the mechanisms of immune system disruption from solar activity by expanding the analysis to include additional measures of immune response, such as antibodies and interferon. The analysis should also be expanded across a larger population to generalize results.. The immune effects presented here are critical in understanding the totality of effects and potential interactions of solar activity and geomagnetic field fluctuations on human health

List of Abbreviations

BMI

Body mass index

eGFR

Estimated Glomular Filtration Rate

DOW

Day of Week

DOY

Day of Year

PM2.5

Particulate matter with aerodynamic diameter less than or equal to 2.5 μm

BC

Black carbon

PN

Particle Number

lbeta

Log β activity

IMF

Interplanetary Magnetic Field

K_p

K_p Index

SSN

Sunspot Number

EMF

Electromagnetic Field

WBC

White blood cells

MONO

Monocytes

LYMPH

Lymphocytes

EOSIN

Eosinophils

BASO

Basophils

ANS

Autonomic Nervous System

SNS

Sympathetic Nervous System

SCN

Suprachiasmatic Nucleus

HPA

Hypothalamic-Pituitary-Adrena