Can the Feynman-Hellmann theorem be used to separate the connected- and disconnected-diagram contributions to the nucleon sigma term ?

In recent lattice QCD studies, the Feynman-Hellmann theorem is often used to estimate separate contributions of the connected and disconnected diagrams to the nucleon sigma term. We demonstrate through a simple analysis within an effective model of QCD why this could be dangerous although the theorem is naturally expected to hold for the sum of the two contributions, i.e. the net nucleon sigma term.