Relations between PM10 composition and cell toxicity: a multivariate and graphical approach

Previous studies have used particle mass and size as metrics to link airborne particles with deleterious health effects. Recent evidence suggests that particle composition can play an important role in PM-toxicity; however, little is known about the specific participation of components (individually or acting in groups) present in such a complex mixture that accounts for toxicity. This work explores relationships among PM(10) components in order to identify their covariant structure and how they vary in three sites in Mexico City. Relationships between PM(10) with cell toxicity and geographical location were also explored. PM(10) was analyzed for elemental composition, organic and elemental carbon, endotoxins and the induction of inhibition of cell proliferation, IL-6, TNFalpha and p53. PM(10) variables were evaluated with principal component analysis and one-way ANOVA. The inhibition of cell proliferation, IL-6 and TNFalpha were evaluated with factorial ANOVA and p53 with the Welch test. The results indicate that there is heterogeneity in particle mass, composition and toxicity in samples collected at different sites. Multivariate analysis identified three major groups: (1) S/K/Ca/Ti/Mn/Fe/Zn/Pb; (2) Cl/Cr/Ni/Cu; and (3) endotoxins, organic and elemental carbon. Groups 1 and 3 showed significant differences among sites. Factorial ANOVA modeling indicated that cell proliferation was affected by PM concentration; TNFalpha and IL-6 by the interaction of concentration and site, and p53 was different by site. Radial plots suggest the existence of complex interactions between components, resulting in characteristic patterns of toxicity by site. We conclude that interactions of PM(10) components determine specific cellular outcomes.