The current study extended a previous investigation on the obesity-HPV association from a specific age group (aged 35–60)  to adult women in general (aged 20–59). Overall, we found a null or weak association between obesity or central obesity and HPV infection in this nationally-representative adult female population. However, subgroup analysis showed that obese or centrally-obese women had a significantly reduced HR-HPV infection among those who reported an early sex debut (<16 years) or those included in the fasting subpopulation. We proposed three potential mechanisms for these observed negative correlations.
First, despite the early sex debut, adult women with excessive adiposity might have a lower HPV burden than women with normal BMI or WC. Early studies on adolescents have consistently linked early sex debut to an increased risk for sexually transmitted infections (STI), including HPV infection [30, 31]. The heightened STI risks were reportedly mediated by concomitant risky behaviors such as frequency of sex, irregular use of condom, alcohol use and a smoking habit . In our analysis, additional adjustment for behavioral factors, such as condom use, drinking history in the last 12 months and cumulative smoking exposure did not change our results. It was possible that obese or centrally obese women who initiated sex activities early may have had a less risky behavioral profile, either in adolescence or early adulthood, resulting in fewer new acquisitions or re-infections overall.
Secondly, we hypothesized that the altered host immune responses in obese or centrally-obese women may have facilitated viral clearance (in adolescence) or suppress viral reactivation from latent, persistent HPV infections (in mid- or late adulthood). Higher concentrations of several cytokines, including tumor necrosis factor (TNF)-α, interleukin (IL)-6 and others, were found in immature cervical epithelium than those in mature, squamous cervical epithelium . Whether such elevation in innate immune responses translates into effective defenses against viral pathogens in adolescents remains unknown. Women who had been obese (or centrally-obese) since adolescence could have outperformed their peers in controlling early HPV infections due to their pro-inflammation milieu. Besides, for adult women with excessive adiposity, their hyper-activated adaptive immunity  might have held latent viruses in check, reducing chances of ageing-associated viral reactivation as compared to women of normal BMI or WC .
Lastly, obese individuals may have paid more visits to healthcare providers because of their cardio-metabolic co-morbidities, such as hypertension or diabetes . The observed negative associations between excessive adiposity and HR-HPV infection were probable results of vigilant screening or diagnosis. Furthermore, a cohort effect could have also explained for an overall, low lifetime exposure risk for HPV in older women;  this potential risk-modifying effect by age at sex debut was particularly evident in the younger (aged 20–34) than the older cohorts (aged 35–50 or 50–59).
To our knowledge, the current study was also among the few investigations in examining metabolic health effects on a specific infection. The finding that obese or MUO (metabolically unhealthy obese) women showed a significantly low probability for HR-HPV but not any-type HPV infection has raised our concerns about a biased, fasting subpopulation. Statistically, we have accounted for the relative low participation of diabetic participants in the fasting session by use of appropriate sampling weights as suggested . Fasting women also appeared to have a similar degree of adiposity to those in the primary analysis in terms of BMI, waist circumference or WHtR (Additional file 1: Table S1). The only apparent difference between the primary and the fasting population was the highest degree of educational attainment by study participants; 29.7 % of fasting women reported a college education or higher whereas 26.2 % of the overall analytic sample did so (P = 0.041). We believe that selection bias was unlikely to fully explain for the strong reverse associations found between HR-HPV and metabolically unhealthy obesity or obesity alone. Large longitudinal studies are needed to prospectively assess influences of excessive adiposity and its associated metabolic derangements on the natural history of HPV infection.
There are several limitations in interpreting our study results. First, the lack of temporality was inherent in the cross-sectional data. Cervical HPV infection is a well-defined local infection, and the potential for causing systemic ectopic disposition of fatty tissue is thus unlikely. Nevertheless, the conjecture of infection-associated adipogenesis needs to be prospectively evaluated as in other obesity-inducing viral infections . Secondly, the current study could not fully examine whether or how the HPV immunization could have affected the adiposity-HPV relationship. The documentation of HPV vaccination among survey participants began since 2006 and only a minority of women had received the vaccine: 3.1 % in 2007–2008 and 5.2 % in 2009–2010. When we repeated the analysis on women enrolled in these two survey cycles (2007–2010, N = 2568) and adjusted for the HPV vaccination history, the strength of association between central obesity and any-HPV was similar (aPR = 0.92) but was not statistically significant (P = 0.171). Moreover, as we did not include women of BMI < 18.5 Kg/m2 in the current analysis, whether low adiposity correlated with a disproportionately high likelihood for prevalent HPV (supporting a monotonically deceasing, dose–response association); or whether women of an abnormally low BMI they similarly exhibited low rates of HPV infection (suggesting an inverse U-shaped association) remains unexplored.
The current study also failed to examine for the contribution of inflammatory proteins or cytokines to the observed reverse associations because conventional inflammatory biomarkers, including ferritin or C-reactive protein, were not available for women of all age groups . Among well-studied adipokines secreted by white adipose tissues, leptin and adiponectin have been implicated in changed immune responses by exerting their pro- and anti-inflammatory activity in both innate and adaptive immunity, separately [37, 38]. To elucidate the mechanism involved in the polarization between pro- and anti-inflammatory pathways during the development and maintenance of visceral adiposity, longitudinally collected immune correlates at the cervix may provide anchoring information to disentangle the causal from collateral events.
Lastly, according to Sumner and colleagues, the currently-adopted, lipid-based definition for MetS were likely to misclassify NHB women as metabolically healthy when in fact they might already exhibit significant insulin resistance . Yet, ad-hoc sensitivity analysis that excluded NHB women from the fasting population revealed unaltered results (data not shown). As such, we believed that findings of the current study were at little risk for ethnicity-associated misclassification bias.