Moreover, SEM presents three advantages compared with bioluminescence. It allows counting all bacteria irrespective of their metabolic activity level, and data showed lower standard deviations. This is partly due to the high sensitivity of bioluminescence, as well as to the fact that values are near the limit of detection with bioluminescence. Last, SEM observations allow three-dimensional details to be visualized on a sample surface,
43 with the direct observation of adhering bacteria and biofilm morphologic characteristics. Thus, even if differences between PMMA, heparinized PMMA, and silicone were not statistically significant, SEM observations demonstrated that the behavior of attached bacteria depended on the IOL surface by showing two different types of adhesion. For hydrogel, hydrophilic acrylic and heparinized PMMA IOLs, bacteria seemed to be isolated
(Figs. 1 2 5) . In contrast, clusters of bacteria were seen on PMMA and silicone IOLs
(Figs. 3 4) . These bacteria were accumulated on the lens surface as microcolonies, meaning that the bacteria probably secreted a layer of adhesive material (slime) and were literally embedded.
18 44 45 Biofilm formation is commonly thought to be a two-step process,
18 44 45 46 47 involving various specific factors: adhesion of cells to a solid substrate followed by cell-to-cell adhesion. The first phase is mediated by nonspecific physicochemical forces and/or the capsular polysaccharide/adhesin (noted classically PS/A) and/or several surface proteins. PS/A was described as mediating primary attachment to silastic catheter surfaces.
48 The second phase is mediated by the bacterial production of a polysaccharide glycocalyx (slime) on the IOL surface,
5 47 49 including a polysaccharide intercellular adhesin (noted classically PIA). In 1992, Mack et al.
50 demonstrated the implication of PIA for the accumulation of multilayered biofilm. Our observations with cluster formation are totally in keeping with the observations of Mack et al. in 1994.
51 Two wild strains were able to produce clusters, whereas mutants unable to produce PIA were impaired in the accumulative phase of biofilm production, and isolated bacteria were observed. A recent study has shown that the intercellular adhesion (
ica) locus of
S. epidermidis encodes production of both PS/A and PIA.
44 It so happens that the strain under study carried the
ica locus. Thus, the formation of clusters of bacteria on IOLs, as opposed to isolated bacteria, could present an additional step in forming a biofilm, and consequently in colonizing a biomaterial. Therefore, PMMA and silicone materials could facilitate bacterial colonization and biofilm production, more so than hydrogel, hydrophilic acrylic, and heparinized PMMA. Another hypothesis is that bacterial colonization is only delayed with hydrophilic IOLs. However, host defenses and/or antibiotics are more efficient on bacteria simply attached than on bacteria embedded within a layer of slime, because they have trouble penetrating the biofilm and killing the embedded bacteria.
52 Therefore, even if colonization of these IOLs is only delayed, host defenses are given more time to react. There is little information in the literature concerning the relationship between IOL materials and endophthalmitis. In a retrospective study, Montan et al.
53 showed that implanting a heparinized PMMA IOL significantly reduces the risk of endophthalmitis compared to using PMMA or silicone IOLs. Even if these results are in complete keeping with our present hypothesis, only further epidemiologic studies will help to confirm and complete this relationship.