Theoretically long gamma-ray bursts (GRBs) are expected to happen in low-metallicity environments, because in a single massive star scenario, low iron abundance prevents loss of angular momentum through stellar wind, resulting in ultra-relativistic jets and the burst. In this sense, not just a simple metallicity measurement but also low iron abundance ([Fe/H]<-1.0) is essentially important. Observationally, however, oxygen abundance has been measured more often due to stronger emission. In terms of oxygen abundance, some GRBs have been reported to be hosted by high-metallicity star-forming galaxies, in tension with theoretical predictions. Here we compare iron and oxygen abundances for the first time for GRB host galaxies (GRB 980425 and 080517) based on the emission-line diagnostics. The estimated total iron abundances, including iron in both gas and dust, are well below the solar value. The total iron abundances can be explained by the typical value of theoretical predictions ([Fe/H]<-1.0), despite high oxygen abundance in one of them. According to our iron abundance measurements, the single massive star scenario still survives even if the oxygen abundance of the host is very high, such as the solar value. Relying only on oxygen abundance could mislead us on the origin of the GRBs. The measured oxygen-to-iron ratios, [O/Fe], can be comparable to the highest values among the iron-measured galaxies in the Sloan Digital Sky Survey. Possible theoretical explanations of such high [O/Fe] include the young age of the hosts, top-heavy initial mass function, and fallback mechanism of the iron element in supernova explosions.