The differences are quite small, less than 10 kJ/mol (that's less than 0.4%), which may be less than the error of the measurement. Supposing that the numbers are meaningful, the molecules have different structures and nuclear connectivity, which can easily lead to small differences in the combustion energy. As an example, fructose forms a 5 member ring in its closed chain form, whereas glucose forms a 6 member ring. These would have some differences in strain energy associated with them. Also differences in condensed phase intermolecular interactions (crystallinity, packing density, hydrogen bonding, and so forth) can influence the measured combustion energy.