The extremely high sensitivity that has been suggested for magnetic particle imaging has its roots in the unique signal produced by the nanoparticles at the frequencies of the harmonics of the drive field. That sensitivity should be translatable to other methods that utilize magnetic nanoparticle probes, specifically towards magnetic nanoparticle spectroscopy that is used to measure molecular biomarker concentrations for an "in vivo ELISA" assay approach. In this paper, we translate the predicted sensitivity of magnetic particle imaging into a projected sensitivity limit for in vivo ELISA. The simplifying assumptions adopted are: 1) the limiting noise in the detection system is equivalent to the minimum detectable mass of nanoparticles; 2) the nanoparticle's signal arising from Brownian relaxation is completely eliminated by the molecular binding event, which can be accomplished by binding the nanoparticle to something so massive that it can no longer physically rotate and is large enough that Neel relaxation is minimal. Given these assumptions, the equation for the minimum concentration of molecular biomarker we should be able to detect is obtained and the in vivo sensitivity is estimated to be in the attomolar to zeptomolar range. Spectrometer design and nonspecific binding are the technical limitations that need to be overcome to achieve the theoretical limit presented.