In optical fibres we reflect light by the use of total internal reflection (TIR). I know there will be some energy loss in optical fibres if we use mirrors but why there isn't significant loss in TIR? (some light must be absorbed)
The basic answer is: good luck coating the interior of a sub-micron tube that is kilometers long with a material that is close to 100% reflective at grazing incidence angles.
Now, there is a minor amount of loss due to failure of the fiber material to maintain TIR. Causes include roughness of the boundary or excess bending of the fiber as installed.
But in general, aside from the injection loss ("acceptance angle") at the start of the fiber, loss per km is incredibly low -- and even a theoretical mirrored-tube would be unlikely to achieve equivalent throughput.
Side note: many/most optical fibers are designed such that energy is carried in the 'evanescent wave' within the first cladding layer, so the mode structures allowed/desired are different from what a mirrored tube would allow. Not to mention the "holey fibers" which feature empty (air) cores with interesting shapes to the bounding material.