The electron-positron pairs observed in heavy-ion collisions at Gesellschaft für Schwerionenforschung Darmstadt mbH have been interpreted as the decay products of yet unknown particles with masses around 1.8 MeV. The negative results of resonant Bhabha scattering experiments, however, do not Support such an interpretation. Therefore we focus on a more complex decay scenario, where the e +e -lines result from a two-collision process. We discuss the induced decay of a metastable 1 + + state into e + e -pairs. For most realizations of a 1 + + state such a decay in leading order can only take place in the Coulomb field of a target atom. This fact has the attractive consequence that for such a state the Bhabha bounds are no longer valid. However, the absolute value of the e ' e production Cross section tums out to be unacceptably small. PACS numberk): 34.90. + q, 12.20.D~ For a number of years the Gesellschaft für Schwerionenforschung Darmstadt mbH (GSI) e +e -lines have posed a vexing problem. Hardly anybody expects really new physics, such as new particles or novel types of resonances, in the energy range of a few MeV. However, a satisfactory conventional explanation of the anomalous e +e -coincidences observed in heavv-ion collisions at GSI is still lacking in spite of numerous experimental and theoretical efforts to explain them, e.g., as due to conversion processes. Until now a considerable arnount of experimental information on these line structures in single positron and correlated electron-positron spectra has been collected by the EPOS and ORANGE groups at GSI with different experimental setups and the existence of the observed structures seems to be a well established fact [1,2]. Presently the phenomenon is also studied by an independent experimental group at the APEX facility at Argonne National Laboratory [3].In this contribution we discuss a particle decay scenario which gives interesting relations between particle decay in high-Z and low-Z media. A particle decaying into an e +e -pair seemed to be a natural explanation of the observed coincidences. However, it soon became clear that the decaying particle could not be an elementary one. This was ruled out, e.g., by the nonobservation of e + e -decays in high-energy beam dump experiments 141. All efforts to save the particle hypothesis in terms of composite particle models failed due to the negative results of low-energy Bhabha scattering. See, however, [5]. One possible exception is the proposal by Spence and Vary [6], who argued that the decay width might strongly depend on the surrounding Coulomb field. It will turn out that the scenario we propose actually realizes this possibility.It is natural to assume that a particle decaying in free space into e + e -should show up as a resonance in the time-reversed process, i.e., Bhabha scattering [7]. Assuming that the dominant decay channel is the e + echannel, the most sensitive Grenoble measurements rule out all resonances with lifetimes