Nine extrasolar planets with masses between 110 and 430M are known to transit their star. The knowledge of their masses and radii allows an estimate of their composition, but uncertainties on equations of state, opacities and possible missing energy sources imply that only inaccurate constraints can be derived when considering each planet separately. Aims: We seek to better understand the composition of transiting extrasolar planets by considering them as an ensemble, and by comparing the obtained planetary properties to that of the parent stars. Methods: We use evolution models and constraints on the stellar ages to derive the mass of heavy elements present in the planets. Possible additional energy sources like tidal dissipation due to an inclined orbit or to downward kinetic energy transport are considered. Results: We show that the nine transiting planets discovered so far belong to a quite homogeneous ensemble that is characterized by a mass of heavy elements that is a relatively steep function of the stellar metallicity, from less than 20 earth masses of heavy elements around solar composition stars, to up to 100M for three times the solar metallicity (the precise values being model-dependant). The correlation is still to be ascertained however. Statistical tests imply a worst-case 1/3 probability of a false positive. Conclusions: Together with the observed lack of giant planets in close orbits around metal-poor stars, these results appear to imply that heavy elements play a key role in the formation of close-in giant planets. The large masses of heavy elements inferred for planets orbiting metal rich stars was not anticipated by planet formation models and shows the need for alternative theories including migration and subsequent collection of planetesimals.
