Low and high serum IgG associates with respiratory infections in a young and working age population

Summary

Background

We investigated health consequences and genetic properties associated with serum IgG concentration in a young and working age general population.

Methods

Northern Finland Birth Cohort 1966 (NFBC1966, n = 12,231) health data have been collected from birth to 52 years of age. Relationships between life-long health events, medications, chronic conditions, lifestyle, and serum IgG concentration measured at age 46 years (n = 5430) were analysed. Regulatory mechanisms of serum IgG concentration were considered.

Findings

Smoking and genetic variation (FCGR2B and TNFRSF13B) were the most important determinants of serum IgG concentration. Laboratory findings suggestive of common variable immunodeficiency (CVID) were 10-fold higher compared to previous reports (73.7 per 100,000 vs 0.6–6.9 per 100,000). Low IgG was associated with antibiotic use (relative risk 1.285, 95% CI 1.001–1.648; p = 0.049) and sinus surgery (relative risk 2.257, 95% CI 1.163–4.379; p = 0.016). High serum IgG was associated with at least one pneumonia episode (relative risk 1.737, 95% CI 1.032–2.922; p = 0.038) and with total number of pneumonia episodes (relative risk 2.167, 95% CI 1.443–3.254; p < 0.001).

Interpretation

CVID-like laboratory findings are surprisingly common in our unselected study population. Any deviation of serum IgG from normal values can be harmful; both low and high serum IgG may indicate immunological insufficiency. Critical evaluation of clinical presentation must accompany immunological laboratory parameters.

Funding

Oulu University Hospital VTR, CSL Behring, Foundation for Pediatric Research.

Research in context.

Evidence before this study

The role of defective adaptive immunity and low serum immunoglobulin (IgG) concentration is well established in common variable immunodeficiency (CVID) patients suffering from recurrent pneumonia and respiratory tract complications. It has also been shown that patients with milder forms of hypogammaglobulinemia may suffer from respiratory infections. Although these patients benefit from IgG replacement therapy, they may remain fully asymptomatic for years. Genetic causes of abnormal B cell maturation or low IgG in CVID are only partially understood. Elevated serum IgG in older individuals may also indicate risk of pneumonia-related mortality and recurrent pneumonia although mechanisms are incompletely described.

Currently it is thought that a delay in recognition of CVID and hypogammaglobulinemia can cause significant morbidity and mortality; early diagnosis and consideration of IgG replacement therapy is believed to be beneficial. Screening protocols to support early identification of those suffering from B cell deficiency have been suggested. However, understanding of the significance of low serum IgG concentration found in asymptomatic individuals is incomplete.

Respiratory infection burden associated with serum immunoglobulin concentration in the general population is poorly understood. CVID is thought to be rare (0.6–6.9 cases per 100.000) although population level prevalence is not well described. Current supplies for IgG replacement products obtained from blood donations are limited; there is an obvious shortage of immunoglobulins and therefore a necessity of wise use of replacement therapy.

Added value of this study

We demonstrated that at population level CVID-like laboratory findings were approximately 10-fold higher when compared to previous reports based on clinical diagnoses. Not only low but even high serum IgG concentrations were associated with respiratory infection burden. Genetic properties and smoking are involved in regulation of serum IgG concentration.

Implications of all the available evidence

Availability of IgG replacement products is currently limited, thus treatment of fully asymptomatic individuals and individuals with normal vaccine responses does not appear prudent. Such high frequencies of CVID-like laboratory findings should be considered when screening protocols of CVID and hypogammaglobulinemia are created. Subtly low IgG levels among smokers strongly suggest cessation of smoking as the first line intervention.

Introduction

Prevalence, long-term persistence, and prognosis of low serum immunoglobulin levels in the general population are not known. Common variable immunodeficiency (CVID) patients with low serum immunoglobulin G (IgG), and A (IgA) and/or M (IgM) concentrations commonly suffer from recurrent pneumonia and benefit from IgG replacement therapy.1, 2, 3, 4, 5 However, such patients may for years remain fully asymptomatic after discovery.⁶ Patients with milder forms of hypogammaglobulinemia may also suffer from respiratory infections.^7,8 Among older individuals, even high serum IgG has been associated with risk of pneumonia-related mortality and recurrent pneumonia.⁹ At an unselected population level, however, the significance of serum IgG concentrations on respiratory infection burden among the young and working age population is incompletely understood.

Prospective and lifelong follow up of Northern Finland Birth Cohort 1966 (NFBC1966) has provided high quality population level health information.¹⁰ We recently investigated the secondary risk factors and conditions related to pneumonia in the NFBC 1966 birth cohort with 52 years of follow-up.¹¹ In the present study, we compare the clinical and behavioural parameters, occurrence of pneumonia, and complicated upper respiratory infection burden with serum immunoglobulin concentrations among the cohort participants. We also aim to explore the relationships between genome-wide polymorphisms in the regulation of serum IgG concentration in a genome-wide association study (GWAS). Finally, we aim to define the role of serum IgG on pneumonia and complicated upper respiratory infection burden in the young and working age population.

Methods

Northern Finland birth cohort 1966

We used lifelong data from the NFBC1966 to evaluate the clinical presentations. The original cohort size and the follow-up have previously been explained in detail.¹⁰ The NFBC1966 covers the entire population born during one year in the Northern provinces of Finland. It covers all individuals born with the expected date (12,055 mothers) during the year 1966 and comprises 12,231 individuals (12,058 life born) (96.3% of all births during 1966 in the area) (www.oulu.fi/nfbc).

Ethics

The study was originally approved by the ethical committee of the Northern Ostrobothnia hospital district (94/2011, 12/2003). Permission to use nationwide register data was sought from the institutions administrating the registers. Written informed consent was obtained from all participants to use the collected cohort data and their registry data for scientific purposes. The home page of NFBC1966 program includes a full description of the study (https://www.oulu.fi/en/university/faculties-and-units/faculty-medicine/northern-finland-birth-cohorts-and-arctic-biobank/research-program-health-and-well-being).

Data collection

Fig. 1 summarises the follow-up and the data collection of the NFBC study.¹⁰ Our study population consists of those whose serum immunoglobulins were measured (n = 5430) with available clinical or questionnaire data in the 46-year follow-up together with consents to use their data in combination with the national register data. Additional data for chronic disease diagnoses, medications and selected pneumonia risk factors were obtained from national Finnish registry databases as listed previously.¹¹ The study participants were analysed for numerous details related to their health and behaviour. Causes of death among the deceased study participants has been reported.¹¹ Data collection, analysis and manuscript preparation agree with the STROBE checklist.

Fig. 1.

The flow of the Northern Finland Birth Cohort (NFBC) 1966 data collection. The study includes all individuals born with an expected date of birth in the year 1966. All study participants were subjects to careful analysis of their lifestyle and health properties at the age of 46 years. Health data from 1971 to 2018 were obtained from national registers. Information on numbers of study participants is shown. Females, employed cohort members, those with high social class, married, those who have children and higher education participated more actively to the follow-up.¹⁰ Causes of death among NFBC 1966 participants have been reported.¹¹

Airway infection definitions

Pneumonia episodes that required hospitalisation, visits, and procedures at an otorhinolaryngologist's clinic, and diagnoses of associated complicated upper airway infections were obtained from the Care Register for Health Care (CRHC), previously named Finnish Hospital Discharge Register (FHDR), maintained by the Finnish Institute for Health and Welfare as previously described in detail.¹¹ In case of multiple diagnoses of pneumonia, only episodes at least 90 days apart were counted. Microbiological data were not available.

Questionnaire at age 46 years provided self-reported data on life-time burden of pneumonia episodes treated at hospital or at home (Supplementary material 1). In addition, data on complicated upper airway infections, such as recurrent otitis and sinus operations, as well as chronic respiratory symptoms (prolonged productive cough, chronic bronchitis, and allergic rhinitis) were retrieved from the questionnaire. Supplementary material 1 includes the full description of questions regarding respiratory infections and their symptoms.

Autoimmune conditions, asthma, and chronic respiratory symptoms

Diagnosis of diseases with potential immunological origins was obtained from national registers based on the ICD 8, 9 and 10 codes. Conditions with at least ten cases were considered and included rheumatoid arthritis, psoriasis, coeliac disease, sarcoidosis, vasculitis, purpura, and multiple sclerosis. The diabetes variable was created by combining the data of hospital registers from CRHC, medicine purchases and reimbursement documentation from the Social Insurance Institution.¹¹ Those suffering from asthma were identified based on reimbursement of medical expenses from the Social Insurance Institution. Detailed criteria for asthma diagnosis and a physician's medical certificate are required for reimbursement eligibility in Finland. Chronic respiratory symptoms were analysed based on the questionnaire at age 46 years (Supplementary material 1).

Serum immunoglobulin concentrations

Serum samples were collected at age 46 years. Serum immunoglobulin G (IgG), immunoglobulin A (IgA), immunoglobulin M (IgM) and immunoglobulin E (IgE) concentrations were measured with accredited immunoturbidimetric methods (Vita laboratories, Helsinki, Finland). Immunoglobulin subclass (IgG1, IgG2, IgG3, IgG4) concentrations were measured from those with serum IgG below 5.0 g/L or above 20.0 g/L.

Antibiotic consumption

Since 1993, antibiotic consumption data have been collected from medicine purchases by the Social Insurance Institution. A physician's prescription is required for all antibiotic purchases. Information on antibiotic classes included in this study are listed in Supplemental Table S2.

Genome-wide and phenome-wide association studies

A genome-wide association study (GWAS) was conducted to detect genetic variation associated with serum IgG concentration in 3591 of the studied 5430 individuals. Genotyping was performed with Illumina Infinium 370cnvDuo array. Close relatives were excluded (pi-hat<0.2) based on identity by descent calculations performed with PLINK.¹² Principal components (PCs) were calculated with PLINK¹² to allow accounting for population substructure. Imputation of the genotype data was done with the HRC imputation pipeline. Before analyses, the IgG concentrations were inverse rank normalised, then adjusted for sex and the first ten PCs to account for population substructure, and inverse rank-based normalisation was used to transform the resulting residuals to a normal distribution. Single-nucleotide polymorphisms (SNPs) with imputation info score <0.8 or minor allele frequency (MAF) <0.01 were excluded, as well as those violating the Hardy–Weinberg equilibrium (p < 0.00001). GWAS was performed under the additive model with SNPtest v. 2.5.4.¹³ Regional association plots were drawn with LocusZoom.¹⁴ A phenome-wide association study (PheWAS) was performed using PhenoScanner V2¹⁵; in these SNP look-ups, p-value<1 × 10⁻⁵ was considered as evidence of association.

Other health data

Several health and behavioural parameters including body mass index (BMI), waist circumference, alcohol, and tobacco consumption, were collected as described.¹¹ Self-reported data of daily alcohol doses exceeding 20 g in women and 30 g in men were considered excessive.¹⁶ Physical activity was collected from the questionnaire data, and it was calculated as the metabolic equivalent of task scores in hours per week.¹⁷

Statistical analysis

In this study, the role of previously identified pneumonia risk factors among the NFBC 1966 study population was analysed with serum IgG concentrations at 46 years in 5430 subjects.¹¹ In addition to these previously identified risk factors, the role of physical activity was considered.¹⁷

To test the association between pneumonia risk factors (Table 1, Supplemental Table S1) as well as infection burden (Supplemental Table S3), the patients were divided into categories by serum IgG concentration (low, IgG ≤6.8 g/L; normal, IgG 6.9–15.5 g/L; high, IgG ≥15.6 g/L) and by sex. Cross-tabulation with the Pearson's chi-square test or Fisher's exact test were used as indicated in Table footnotes. We used Kruskal–Wallis test to compare physical activity measures between IgG categories (Table 1). These IgG concentration categories based on ±2 standard deviations (±2SD) were chosen according to the distribution of measured IgG levels in the study cohort. Risk factors with a count of less than ten were not evaluated.

Table 1.

Education, smoking, alcohol consumption, obesity, physical activity (metabolic equivalent of task scores in hours per week) and chronic conditions by sex at age 46 years divided according to serum IgG concentration categories.

Serum IgG (g/L)	Male				Female
	0–6.8	6.9–15.5	>15.6	p	0–6.8	6.9–15.5	>15.6	p
Education				0.510a				0.084a
Basic	5 (16.1)	186 (8.3)	8 (10.0)		2 (5.4)	166 (5.6)	5 (6.1)
Secondary	20 (64.5)	1528 (67.9)	56 (70.0)		31 (83.8)	1877 (63.8)	51 (62.2)
Tertiary	6 (19.4)	535 (23.8)	16 (20.0)		4 (10.8)	898 (30.5)	26 (31.7)
Smokingd				<0.001a				<0.001a
Never	9 (30.0)	1066 (47.9)	40 (51.3)		10 (27.0)	1704 (58.6)	53 (64.6)
Former	4 (13.3)	688 (30.9)	26 (33.3)		4 (10.8)	720 (24.7)	18 (22.0)
Current	17 (56.7)	471 (21.2)	12 (15.4)		23 (62.2)	486 (16.7)	11 (13.4)
Excessive alcohole				0.774a				0.669a
No	27 (87.1)	1909 (84.8)	65 (82.3)		33 (89.2)	2716 (92.4)	77 (93.9)
Yes	4 (12.9)	342 (15.2)	14 (17.7)		4 (10.8)	224 (7.6)	5 (6.1)
BMI >30				0.261a				0.215a
No	28 (90.3)	1775 (78.8)	65 (81.3)		32 (86.5)	2337 (79.4)	60 (73.2)
Yes	3 (9.7)	477 (21.2)	15 (18.8)		5 (13.5)	605 (20.6)	22 (26.8)
Asthma				0.365a				0.126a
No	27 (87.1)	2106 (93.4)	74 (92.5)		33 (89.2)	2693 (91.4)	80 (97.6)
Yes	4 (12.9)	149 (6.6)	6 (7.5)		4 (10.8)	252 (8.6)	2 (2.4)
Autoimmune disease				0.506b				0.029a
No	29 (93.5)	2075 (95.4)	73 (93.6)		35 (94.6)	2722 (93.0)	70 (85.4)
Yes	2 (6.5)	101 (4.6)	5 (6.4)		2 (5.4)	206 (7.0)	12 (14.6)
Sarcoidosis				0.475b				0.197b
No	31 (100)	2151 (98.9)	76 (97.4)		37 (100)	2909 (99.4)	80 (97.6)
Yes	0 (0)	25 (1.1)	2 (2.6)		0 (0)	19 (0.6)	2 (2.4)
Heart disease				0.655a				0.395a
No	26 (83.9)	2002 (88.8)	70 (87.5)		33 (89.2)	2701 (91.7)	72 (87.8)
Yes	5 (16.1)	253 (11.2)	10 (12.5)		4 (10.8)	244 (8.3)	10 (12.2)
Chronic kidney disease				1.00b				0.709b
No	31 (100)	2236 (99.2)	80 (100)		37 (100)	2915 (99.0)	81 (98.8)
Yes	0 (0)	19 (0.8)	0 (0)		0 (0)	30 (1.0)	1 (1.2)
Chronic liver disease				0.071b				0.381b
No	29 (93.5)	2230 (98.9)	80 (100)		36 (97.3)	2912 (98.9)	82 (100)
Yes	2 (6.5)	25 (1.1)	0 (0)		1 (2.7)	33 (1.1)	0 (0)
Diabetes				0.822b				0.525b
No	29 (93.5)	2103 (93.4)	75 (93.8)		34 (91.9)	2829 (96.1)	80 (97.6)
Type 1	0 (0)	17 (0.8)	1 (1.3)		0 (0)	13 (0.4)	0 (0)
Type 2	2 (6.5)	132 (5.9)	4 (5.0)		3 (8.1)	103 (3.5)	2 (2.4)
Hypothyroidism				0.138b				0.522b
No	31 (100)	2243 (99.5)	78 (97.5)		36 (97.3)	2899 (98.4)	81 (98.8)
Yes	0 (0)	12 (0.5)	2 (2.5)		1 (2.7)	46 (1.6)	1 (1.2)
Solid cancer				0.587b				1.00b
No	31 (100)	2159 (99.2)	77 (98.7)		37 (100)	2913 (99.5)	82 (100)
Yes	0 (0)	17 (0.8)	1 (1.3)		0 (0)	15 (0.5)	0 (0)
Haematological cancer				0.223b				0.077b
No	30 (96.8)	2164 (99.4)	78 (100)		37 (100)	2918 (99.7)	80 (97.6)
Yes	1 (3.2)	12 (0.6)	0 (0.0)		0 (0)	10 (0.3)	2 (2.4)
Physical activity				0.920c				0.390c
Median	13.1	12.5	13.1		12.6	14.3	13.1
Range	0–56	0–84	0–63		0–63	0–84	0–70

^aWith Pearson's Chi-square test.

^bWith Fisher's Exact test.

^cWith Kruskall–Wallis test.

^dSelf-reported data on smoking divided into “never”, “former” and “current” categories.

^eExcessive alcohol consumption is defined as self-reported daily consumption of ≥30 g for males and ≥20 g for females. Data are presented as n (%), unless otherwise stated.

To understand the overall respiratory disease burden associated with serum IgG, we composed the combination variable of upper and lower respiratory tract infections with at least one pneumonia episode, chronic sinus infection or sinus surgery. Pearson's chi-square test was used to analyse this combination variable with the IgG categories by sex. The association between antibiotic consumption and serum IgG concentration (Supplemental Tables S2 and S3) was analysed with cross-tabulation and Pearson's chi-square significance test. The means of serum IgG concentrations were analysed between three smoking categories with variance analysis (ANOVA) and with independent t-test between current smokers with or without history of pneumonia as well as between non-smokers with or without history of pneumonia (Fig. 4).

Fig. 4.

Distribution of serum IgG concentrations among participants divided based on history of pneumonia (A) (at least one episode or no episodes) and smoking status (current). Serum IgG was low among those who smoke and a history with at least one pneumonia episode (mean 10.5 ± 2.4 g/L) when compared to smokers who had no pneumonia (mean 10.9 ± 2.2 g/L, p = 0.02). Non-smokers with (11.8 ± 2.7 g/L) or without (11.5 ± 2.0 g/L, p = 0.137) history of pneumonia had a similar serum IgG concentration. Cases with serum immunoglobulin concentrations (IgG, IgM, IgA, g/L) suggestive of common variable immunodeficiency (CVID) according to European Society for Immunodeficiencies (ESID) criteria (B). Those with low serum IgG (−2SD, <6.7 g/L) and low serum IgM (<0.4 g/L) or low serum IgA (<0.8 g/L) were included. Lifelong history of hospital treatment for pneumonia is indicated. Self-reported data on respiratory tract symptoms, other symptoms and information on smoking is included. Current smokers (cases 5 and 6) were excluded from calculation of CVID prevalence.

We used the Poisson regression models to evaluate the unadjusted and adjusted relative risk (RR, 95% CI) of future pneumonia and number of sinus surgeries. Because of the over-dispersion assumptions for the Poisson models the unadjusted and adjusted relative risk of antibiotic consumption was analysed with negative binomial models. Numbers of pneumonia episodes, number of antibiotic consumption and number of sinus surgeries were used as dependent variables. To compute the unadjusted and adjusted Odds Ratios (OR, 95% CI) in the binary logistic regression models, at least one pneumonia was a dependent variable. In all models, the categorised IgG concentration was an independent variable where the associations with outcomes by low and high concentrations were compared with those within the normal range. The criteria to select the adjustment variables i.e. potential confounders (sex, smoking, education, asthma, autoimmune disease, chronic liver disease, haematological cancer, physical activity) to the models were based on the significance level of the test in cross tabulations, the previously identified association (p < 0.05) with pneumonia¹¹ and the sufficiency of the number of samples per category (Fig. 3).

Fig. 3.

Number of participants per 10,000 with at least one pneumonia hospitalisation episode caused by pneumonia (n = 367) (A) in relation with serum IgG concentration (g/L). Number of participants per 10,000 with at least two self-reported pneumonia episodes (n = 531) (B) in relation with serum IgG concentration (g/L). Number of participants who have experienced sinus surgery (n = 231) (C) per 10,000 in relation with serum IgG concentration. Antibiotic consumption in relation with serum IgG concentration (D). Multivariate analysis (E) of pneumonia (at least 1 hospitalisation), lifetime number of pneumonia episodes, antibiotic consumption, and sinus surgery in unadjusted and adjusted conditions.

The incidence of upper and lower respiratory infections as well as sinus operations per 10,000 were calculated using a formula ((10,000 × n)/N), where N is the number of subjects in each specific IgG category and n is the number of subjects with the first infection episode. For antibiotic consumption, the number of antibiotic courses (n) was divided with the number of subjects in a specific IgG category (N) using formula (n/N). Serum IgG subgroups were formed by assigning separate subgroups for each increment of 1 g/L in IgG concentrations in every 1 g/L between 6 and 18 g/L. Those with serum IgG concentration under 6 g/L or concentration of over 18 g/L formed the lowest and the highest IgG subgroups (Fig. 3).

The p-values of <0.05 were considered statistically significant. The statistical analyses were performed using IBM SPSS Statistics for Windows, Version 28 (IBM Corp., Armonk, NY, USA).

Role of funders

This study was partly supported by Oulu University Hospital VTR, CSL-Behring and Foundation for Pediatric Research.

Results

Serum immunoglobulin concentrations

Mean serum IgG concentration of the whole study population (n = 5430) at age 46 years was 11.20 g/L (SD 2.2 g/L) (Fig. 2A). Two SD below and above the mean were 6.84 g/L (−2SD) and 15.56 g/L (+2SD), respectively. 57 participants had serum IgG 6.8 g/L or lower. In 162 cases, serum IgG level was above 15.6 g/L. IgG subclasses were measured for those with serum IgG lower than 5.0 g/L or higher than 20.0 g/L. In all cases, subclass findings were consistent with an even distribution.

Fig. 2.

Distribution of serum immunoglobulin G (IgG) concentrations among all study participants (n = 5430) (A). The mean serum IgG concentration is 11.20 g/L. The two standard deviations below (−2SD) the mean was 6.84 g/L. The +2SD serum IgG was 15.56 g/L. Serum IgG was below 2SD in 57 cohort participants and above 2SD in 162 participants. Distribution of serum IgG among non-smokers (n = 2882) (B), former smokers (n = 1460) (C), and current smokers (n = 1020) (D) are shown. Mean serum IgG was 10.3 g/L (−2SD 5.9 g/L; +2SD 14.6 g/L) and 11.5 g/L (−2SD 7.3 g/L; +2SD 15.7 g/L) among smokers and non-smokers, respectively (p < 0.001). Genome-wide association study of serum immunoglobulin G levels (E–G). A Manhattan plot summarising the results of GWAS of IgG levels (n = 3591) is shown (E). Regional association plots at the two associated loci are shown in panels F and G. Chromosomal positions are shown on the x axis and -log 10 (p-values) on the y axis, and each dot is a single SNP. The red dashed line indicates the level of genome-wide significance (p < 5 × 10⁻⁸). Genomic positions refer to human genome build hg19. Linkage disequilibrium values refer to the 1000 Genomes European population.

Parameters associated with low or high serum IgG concentration

The participants’ characteristics associated with low, normal, or high serum IgG were considered. Table 1 shows the associations of IgG classes with social background, health, and lifestyle variables as well as with chronic conditions. Diagnosis of any autoimmune condition among females was associated with serum IgG concentration (p = 0.029). Supplemental Table S1 shows that serum IgG level was also associated with rheumatoid arthritis although the number of cases was low. Current smoking was associated with low serum IgG both in males (p < 0.001) and females (p < 0.001) (Table 1). Serum IgG concentrations among current smokers and non-smokers were 10.3 g/L (−2SD 5.9 g/L; +2SD 14.6 g/L) and 11.5 g/L (−2SD 7.3 g/L; +2SD 15.7 g/L), respectively (p < 0.001) (Fig. 2). Mean serum IgG concentration in former smokers (11.2 g/L; −2SD 6.9 g/L; +2SD 15.6 g/L) did not differ from non-smokers. Other studied parameters including obesity, alcohol consumption, cardiovascular disease, diabetes, or malignancy were not associated with serum IgG concentrations.

Additional potential aetiologies for secondary hypogammaglobulinemia among those with low serum IgG concentrations were considered according to previously published criteria.^3,18 In this population-based cohort without subjects with advanced age, low serum IgG was not associated with malignancies, or use of corticosteroids or other immunosuppressants. In summary, despite smoking, secondary causes to explain hypogammaglobulinemia were not found suggesting additional hereditary factors.

Laboratory findings suggestive of common variable immunodeficiency (CVID)

We analysed the immunoglobulin profiles of study subjects for consistency with CVID criteria according to the European Society for Immunodeficiency (ESID).¹⁹ In summary, laboratory values in a total of seven cases with both reduced serum IgG (≤6.7 g/L) and serum IgA (<0.8 g/L) or serum IgM (<0.4 g/L) concentrations were found (Fig. 4B). None of them had received a diagnosis of immunodeficiency. Only two of them had suffered a single pneumonia episode (cases 2 and 3 in Fig. 4B). Two cases (5 and 6) were current smokers which potentially aggravated their immunoglobulin findings. However, the immunoglobulin profiles of cases 1–4 in Fig. 4B suggest a potentially CVID-like condition, with a very high population prevalence (73.7 cases/100,000).

Serum IgG concentrations and pneumonia

Those with low, but also with high serum IgG concentrations had experienced pneumonia episode(s) more frequently than their peers (Fig. 3), resulting in over 1000/10,000 pneumonia episodes among those with low or high serum IgG concentration. Multivariate analysis confirmed that also participants with high serum IgG (≥15.6 g/L) had, based on hospital discharge register data, frequently been hospitalised for pneumonia and experienced higher numbers of pneumonia episodes compared to those with normal serum concentration (6.9–15.5 g/L) (Fig. 3E).

Smoking, serum IgG and pneumonia

Serum IgG was lower (mean 10.5±2.4 g/L) among smokers with a history of at least one pneumonia episode when compared to smokers who had not experienced pneumonia (mean 10.9±2.2 g/L, p = 0.02) (Fig. 4A). Non-smokers with (11.8±2.7 g/L) or without (11.5±2.0 g/L, p = 0.137) a history of pneumonia episodes had a similar serum IgG concentration. Associations with continuous serum IgG concentration and smoking status are illustrated with a line diagram in Fig. 4A.

Upper airway infections in patients with hypogammaglobulinemia

Fig. 3 also demonstrates the associations between serum IgG concentrations and sinus surgery due to chronic sinusitis. Multivariate analysis confirmed an association between low serum IgG concentrations and sinus surgery (Fig. 3E). Results also unsurprisingly demonstrated that those with low serum IgG concentration had higher antibiotic use compared with those with normal serum IgG levels (Fig. 3E).

Combined data of upper and lower respiratory infections

A Combination parameter of at least one pneumonia episode, chronic sinus infection or sinus surgical operation was also analysed. We found that participants with any of these lower or upper respiratory infections had evidence of deviating serum IgG. A high proportion of those with low (20.6%) or high (16.0%) serum IgG concentration had suffered from any respiratory infection complication when compared to normal serum IgG group (10.9%, p < 0.007).

Genome-wide association study

In GWAS conducted to detect genetic variations associated with serum IgG concentration, two associated loci were detected (Fig. 2E–G): FCGR2B and TNFRSF13B. The most significant SNPs (lead SNPs) in these loci were rs7554873 (upstream of FCGR2B; MAF = 0.07, beta = −0.48, p = 2.3 × 10⁻²³) and rs4273077 (intronic within TNFRSF13B; MAF = 0.09, beta = 0.25, p = 1.6 × 10⁻⁹). Both loci include biologically relevant genes that are involved in immunity: FCGR2B encodes low affinity immunoglobulin gamma Fc region receptor II-b (aka CD32b)²⁰ and TNFRSF13B encodes tumour necrosis factor receptor superfamily member 13B (aka TACI).²¹ The FCGR2B lead SNP rs7554873 has previously been associated with haematocrit, red blood cell count and haemoglobin concentration,²² and with some HDL-related cholesterol measures.²³ Interestingly, FCFR2B-rs7554873 is also associated with mRNA levels of, e.g., FCGR2B, FCGR2C and HSPA7,²⁴ and with blood protein levels, including those of FCGR2B, FCGR2A and X isoform of amelogenin.^25,26 We detected that the TNFRSF13B lead SNP rs4273077 is associated, e.g., with serum total protein,²⁷ risk of multiple myeloma,²⁸ and monogalactosylation of IgG,²⁹ as well as blood levels of certain proteins, such as tumour necrosis factor receptor superfamily 17 and HLA class II histocompatibility antigen, DM alpha chain.²⁵ Overall, the IgG-associated SNPs are characterised by having associations across multiple immunity-related phenotypes.

Discussion

The role of severe hypogammaglobulinemia in monogenic CVID is well established³⁰ and milder hypogammaglobulinemia is a known risk factor for infections.^7,8 In addition, there is some evidence suggestive of unfavourable properties associated with high serum IgG concentration among older individuals.⁹ Our lifelong follow-up study on the health of a young and working age birth cohort population and their serum IgG concentrations suggests novel insights into the significance of adaptive immunity.^10,11 Highly intriguingly, not only the low IgG but also high serum IgG concentration in our cohort are associated with unfavourable outcomes. It seems possible that any deviation of serum IgG concentration from normal values may be an indication of an unfavourable immunological property. Importantly, a significant role for serum IgG concentration in our study was found with several self-reported and register based parameters including pneumonia burden, upper respiratory tract health and antibiotic use.

Ameratunga et al. followed cohorts of hypogammaglobulinemia patients and concluded that patients with significant hypogammaglobulinemia should receive IgG substitution regardless of their symptoms.⁶ In our study, the high prevalence of hypogammaglobulinemia cases with laboratory findings suggestive of CVID-like condition (73.7 cases per 100,000) questions the feasibility of such an approach when compared to previous reports of already exceptionally high CVID prevalence from Finland (6.9 per 100,000) or elsewhere (0.6–3.8 per 100,000).³¹ In our study population, none of the CVID-like cases had received immunodeficiency diagnosis or immunoglobulin treatment. This high frequency must be considered especially when selecting criteria for screening of B cell deficiencies.32, 33, 34 Our findings highlight the importance of careful consideration of clinical history and disease spectrum when evaluating the immunological laboratory parameters in clinical settings.^3,35

We considered the known associated secondary factors for IgG concentrations at population level^10,18; low serum IgG concentration was most obviously associated with smoking in our study cohort. Mean IgG concentration was lower among current smokers compared to non-smokers consistent with previous reports.^36,37 Smokers who had experienced at least one pneumonia episode had lower serum IgG concentration compared to non-smokers or smokers without a history of pneumonia. Our findings support the view that increased pneumonia risk among smokers is at least partly explained by immunological mechanisms. It seems possible that low serum IgG may also at least partially explain the previously observed high antibiotic prescription rate among tobacco users.³⁸ Serum IgG concentration among former smokers, however, was comparable to non-smokers suggesting that the adverse effects of tobacco components on serum IgG are reversible. Although autoimmunity was associated with high serum IgG concentration among females, infection burden in our female cohort was low. We did not find statistically significant associations with infections, autoimmunity, and serum IgG.

B cell maturation and IgG production are genetically regulated.³⁹ In our birth cohort, GWAS analysis found a role for FCGR2B and TNFRSF13B in the regulation of serum IgG concentration. Similar TNFRSF13B associations have previously been found in Chinese⁴⁰ and Japanese⁴¹ populations. In addition, both TNFRSF13B and FCGR2B were associated with serum IgG in an Icelandic study.⁴² Our study confirms that FCGR2B and TNFRSF13B are associated with serum IgG even in the genetically distinct and isolated population of Northern Finland. Interestingly, variants in TNFRSF13B are common among CVID patients. These may increase the risk of CVID in combination as polygenic risk factors, through dominant negative effects or haploinsufficiency.^21,43, 44, 45 Interestingly, theTNFRSF13B lead SNP rs4273077 is in our study associated with monogalactosylation of IgG; these biological events play an important role in the interaction of IgG with FcγRs.^29,46 Although the role of FCGR2B in CVID is not obvious, this only inhibitory Fcγ receptor may have a role in the regulation of IgG production and development of autoimmunity.^47,48 Since smoking in our study was the only secondary factor associated with low serum IgG, it is tempting to speculate that FCGR2B and TNFRSF13B may also be implicated in population level respiratory infection burden. The role of these genes is further supported by our PheWAS analysis in which multiple immunological associations were observed.

Limitations of our study include the possibility that those with poor immunity and high number of infections may have been lost from the follow up. Since the number of deaths caused by infections or immunological causes in the NFBC 1966 cohort is low, our study cohort may well represent the attributes of the general population.¹¹ At age 46 years, especially females, employed cohort members, and those with high social class participated actively in our NFBC 1966 study.¹⁰ Active participants were also more likely to be married, to have children and higher education. Selected social parameters (education, alcohol consumption), however, did not associate with serum IgG concentration. Importantly, it must be recognised that the potential selection bias should not interfere with our results of associations between the IgG concentration and infection susceptibility. It is also important to note that for the GWAS analyses we had a restricted sample size.

In conclusion, our study reveals that a significant population level infection burden is associated with low or high serum IgG concentrations. While previously documented genetic associations with serum IgG concentrations were confirmed, smoking clearly has an impact on serum IgG concentrations. This and the high frequency of CVID-like laboratory findings must be recognized when low serum IgG concentration is encountered. While our current report provides a historical view of 52 years of infections in the young and working age population, our study still lacks prognostic data on serum IgG and long-term survival at population level. Further follow up of future infectious events among the NFBC 1966 will provide invaluable understanding on population level significance of serum IgG and adaptive immunity.

Contributors

PH, PP, JA, TH, MRJ: study design, data collection, data analysis, manuscript preparation.

PH, PP, MRJ, TH: verification of the underlying data.

MKK: GWAS study and manuscript preparation.

JK, SV, MRJS, ES, AH: data analysis and manuscript preparation.

All authors have contributed to the study design and manuscript preparation, and they have approved the manuscript.

Data sharing statement

All data are available upon reasonable request. Instructions for material request portal can be found at Northern Finland Birth Cohort study home page (Faculty of Medicine | University of Oulu).

Declaration of interests

PH: received scientific conference sponsorship from Octapharma and Takeda.

TH: received scientific conference sponsorship from CSL Behring.