Life-course origins of social inequalities in adult immune cell markers of inflammation in a developing southern Chinese population: the Guangzhou Biobank Cohort Study

Socioeconomic position (SEP) throughout life is usually inversely associated with morbidity and mortality from cardiovascular disease, although the underlying biological pathway is not entirely clear [1, 2]. Cardiovascular disease has been associated with higher levels of inflammatory molecules, perhaps as a consequence of exposure to pathogenic organisms [3], although it is unclear whether pathogen burden mediates SEP differences in cardiovascular risk [3, 4]. Poor early life conditions are usually associated with higher levels of inflammatory markers [5‐11] and poorer adult immune function [12, 13]. These associations are less clear, however, amongst men from middle income countries [10]. Furthermore, little is known about the association of SEP across the life course with immune function. The duration or number of exposures across the life course may be most important (the accumulation hypothesis) [14]. Alternatively, the timing of exposure to poor socioeconomic conditions may be crucial as a number of sensitive periods or simply as a single critical period (the critical period hypothesis). It is also possible that either inter- or intra-generational social mobility plays a part.

Developmental trade-offs between growth, maintenance, and reproduction may occur when there are competing demands for energy resources between biological systems [13, 15, 16], potentially at the expense of immune function in resource-poor environments. Alternatively, intergenerationally and environmentally driven up-regulation of the gonadotropic axis with economic development may obscure some of the normally protective effects of social advantage in the first few generations of men to experience better living conditions [17, 18], thus generating epidemiologically stage specific associations between SEP and immune-related functions, such as pro-inflammatory states, among men [18, 19].

Rapidly developing mega-cities of China may provide a sentinel for the changes in non-communicable diseases expected with economic development and inform effective interventions to reduce the disease burden. In a large sample of older residents from one of the most developed mega-cities in China, Guangzhou in southern China, we assessed the association of SEP at four life stages with proxies of inflammation (total white blood cell, granulocyte, and lymphocyte counts) and compared models representing the accumulation, sensitive periods and critical period hypotheses. Additionally, we hypothesise that 1) higher life course SEP is protective for adult inflammation, 2) the normal protective effect of social advantage is obscured in men experiencing rapid socioeconomic development.

Of the 10,088 phase 3 participants examined, 1.1% (n = 107) had missing data for total white blood cell, granulocyte or lymphocyte counts. Analysis was based on the remaining 9,981 participants. There were more women (n = 7,445) than men (n = 2,536) and the women were younger [mean age 59.2 years (S.D. 7.6)] than the men [mean age 63.1 years (S.D. 7.6)]. Overall the mean white blood cell and granulocyte counts were lower in women than men (Table 1).

The associations of SEP with white blood cell, granulocyte or lymphocyte counts did not vary with age (data not shown). However, associations of SEP with lymphocyte count varied with sex, so only sex stratified results have been presented for this outcome. For white blood cell count and granulocyte count, the sensitive periods model performed better than the fully saturated model, accumulation or critical period models (Table 2). The sensitive periods model shows that some life stages had stronger negative associations than others with white blood cell count and granulocyte count; the early adult stage had the strongest association for both outcomes.

The pattern for lymphocyte cell count was somewhat different. Associations varied by sex. Table 3 shows that for both sexes the accumulation and sensitive periods models did not perform as well as critical period models. The early adult life stage was a critical period for women, with a negative association between SEP and lymphocyte cell count. By contrast, for men, all estimates of association between SEP and lymphocyte cell count were positive, although all confidence intervals included zero.

Additional adjustment for lifestyle factors (smoking, alcohol use, and physical exercise) attenuated estimates slightly (see Appendix) but the pattern of associations generally remained the same. Smoking among men is associated with both low SEP and higher lymphocyte count, hence adjustment for smoking strengthened the positive association of SEP with lymphocyte count (Appendix). Further adjustment for BMI (Appendix) produced very similar results; estimates of association were little changed. All results were similar in a complete case analysis (Appendix).

Consistent with other studies in developed and developing settings examining the association between SEP and inflammation [5‐11], we found that SEP was negatively associated with adult immune cell numbers, particularly among women. Consistent with the only other study from a developing country setting, the advantage of higher SEP for adult inflammation was less marked among men [10]. In general, considering SEP at all four life stages was better than considering individual life stages (critical periods) except for lymphocyte cell counts.

This study has a number of strengths. To our knowledge, it is the first study to investigate the role of life course SEP in later adulthood inflammation in a non-western, developing setting. Moreover, we explicitly determined the most parsimonious representation of life course SEP. The large sample size allowed sex-specific analysis. Nevertheless, there are limitations. First, it is a cross-sectional study with recalled SEP, which may be imprecise, although most likely non-differential. Second, in a cross-sectional design reverse causality must be considered although it is unlikely that inflammation has a causal effect on life course SEP. Third, there may have been gender bias in the allocation of resources within families, most likely favouring boys and men, which may have mitigated the disadvantages of low SEP. However it is unclear why this should have mitigated the effect of SEP for lymphocytes but not for white blood cells and granulocytes. Fourth, our cohort may not be fully population representative. However, prevalence of certain morbidities, such as diabetes, were similar to those in a representative sample of urban Chinese [31]. Fifth, survivor bias is possible, which may have limited participants' socioeconomic and health diversity, biasing results towards the null. If survivorship were an issue we would have expected differences in associations by age, of which there was no evidence. Sixth, we did not explicitly consider the life course effects of social mobility since these are particularly hard to define and test clearly. Inter- and intra-generational mobility, upward and downward mobility are all potential risk factors.

Seventh, a single measurement of white blood cells and differential cell counts may not accurately reflect long-term immune function or inflammation. However, white blood cell count is used as a marker of immune status in clinical settings and is a well-established and routinely-used marker of systemic inflammation [32]. White blood cell count is associated with disease risk and predicts disease outcome [33, 34]. Eighth, although we report associations between SEP and differential white blood cell counts, clinical significance remains to be determined. Within the normal range, elevated white blood cell counts are associated with risk factors for chronic diseases, such as cardiovascular disease [32, 35]. White blood cell counts can be conceptualised as a mixed marker of exposure and response, even a relatively small shift towards a healthier inflammation-immunological profile might have significant public health benefits at the population level [33, 34]. Ninth, acute infection, trauma and underlying chronic disease or medication could be mediators. There is no evidence to suggest that participants were experiencing infection during the assessment process, nor significant trauma. Although only those with life-threatening illness were specifically excluded, those experiencing significant acute infection or trauma were less likely to attend this study, which should have minimized any bias from this source. We also performed descriptive analysis of the data to detect and exclude outliers, which may have resulted from unknown underlying disease, medication, or recording error.

One possible explanation for the association of low SEP with inflammation is via current health behaviour linked to inflammation [5, 36‐38]. Although we did not perform formal tests of mediation, we did adjust for smoking, alcohol consumption, physical activity and BMI in separate models (Appendix), which had little effect among women, but among men, this attenuated the negative association of early adult SEP with white blood cell and granulocyte counts and strengthened the positive association of early adult SEP with lymphocyte counts. This suggests that any associations are unlikely to be driven by adult health behaviour in women, though these may obscure negative associations of early adult SEP with inflammatory markers in men.

Low SEP may increase exposure to pro-inflammatory agents, such as microbial pathogens, pollutants or adverse work conditions. Mechanisms for increased exposure or vulnerability to pathogens in low SEP groups include earlier and/or greater lifetime exposure due to adverse living conditions, such as overcrowding, and increased susceptibility to primary infection through nutritional deficiencies, or stress-related immune dysfunction [3]. A gender bias may have protected low SEP men from such exposures and adverse work conditions, although it is not clear why the effects should be most obvious for lymphocytes. Lower birth weight amongst those with low childhood SEP is another possible explanation, but birth weight is not available for our participants. Birth weight is inversely associated with inflammatory markers [6, 39]. However, birth weight appears to be less relevant in developing country settings such as ours [40], and there is no reason why birth weight should have sex-specific effects on some white cell sub-types.

An alternative explanation is that better early life conditions would be expected to promote development of the adaptive immune system, particularly of the thymus,[41] whose development takes place in early life [41] and which is sensitive to malnutrition, micro-nutrient deficiencies and infections during growth and development [42‐44]. Moreover, the same exposure would also allow upregulation of the gonadotropic axis resulting in sex-specific effects on some immune cell sub-populations [45‐47], particularly those relating to adaptive immunity. Consistent with this mechanism we have previously observed similar sex-specific associations, in the Guangzhou Biobank Cohort Study, of childhood stress with white cell count [48] and of childhood diet with lymphocytes but not granulocytes [19]. However, we do not have measurements that would allow proof of this mechanism.

Socioeconomic position was inversely associated with white blood cell differential counts, as a marker of inflammation, with a clearer and more consistent association among women than men. Environmentally and inter-generationally driven changes to the gonadotropic axis may obscure the normally protective effect of social advantage in the first few generations of men, but not women, to experience better living conditions. Given the links between the immune system, inflammation and chronic disease, this provides a biological mechanism between SEP and the pathophysiological genesis of chronic disease. Understanding such mechanisms for populations experiencing the epidemiological transition is of public health significance.

Associations of socioeconomic position with markers of inflammation, additionally adjusted for lifestyles factors (smoking, alcohol use and physical exercise), are shown below: associations with white cell counts and granulocyte counts are shown in Table 4 and sex-specific associations with lymphocyte counts are shown in Table 5. Associations further adjusted for body mass index are shown in Table 6 (for white cell count and granulocyte count) and Table 7 (lymphocyte count). Results from the complete case analysis are shown in below in Table 8 (white cell count and granulocyte count) and Table 9 (lymphocyte count).