Extensive analysis of the experimental data of JET H-mode plasma using interferometry and Thomson scattering measurements shows that the density peaking factor n0/<n> strongly depends on the effective collisionality veff 10-14 RZeffne/Te2 . Scalings derived from these experiments predicts a peaked density profile in ITER H-mode plasmas, which should result in higher performance (higher bootstrap current and fusion reaction rate). One of the main difficulties of an extrapolation of JET data towards reactor conditions is that the majority of H-mode shots were obtained with dominant NBI heating, while the few H-modes in the database with only ICRH have low bN 1 (due to lack of available power), significantly below the ITER targets (bN 2). However theory predicts a strong TEM destabilization in case of dominant electron heating, which may result in density profile flattening by the appearance of an outward convective particle flux. In contrast to this expectation, purely electron heated H-modes with bN = 1 and Te/Ti 2 have recently been obtained on TCV, using 3rd harmonic ECRH, showing that significantly peaked density profiles can persist in electron heated plasmas at reactor relevant values of bN, lending support to the predictions.