At relatively high concentrations, the scattering intensities, from the samples you want to measure, can be exceedingly high. This runs the risk that you are measuring in the limit of "multiple scattering" which, in turn, renders the data analysis of sizes incorrect (it intrinsically assumes that each photon only gets scattered once by a particle before reaching the detector). To address this, you should re-measure all your samples and make sure that the results you get at those high concentrations match with numbers obtained at sufficiently low concentrations for which the standard data analysis is suitable. This is the only way to make sure those numbers are trustworthy.

At high concentration it may leads to aggregates due to the charge of particle and/or also settle down rapidly so it change the hydrodynamic diameter. As these hydrodynamic diameter is mostly depends on the principle of scattering phenomenon. so you can dilute your sample 10 times or 100 times as require and measure the particle size it gives you better result with good Polydispersity Index below 0.5.

Beside the already mentioned reasons (multiply scattering, aggregation of particles) also a conribution of hydrodynamic interactions can be present at higher concentrations. I.e. also you are still in the limit of single scattering, you will NOT see pure Brownian diffusion but more complex dynamics. You can check this by measuring you DLS signal for several scattering angle. By this, you will obtain the decay parameter of the time correlation function, say Gamma(q), for several wave vector transfer q. By plotting q^2 vs Gamma (q) you will obtain a straight line intercepting zero for BROWNIAN dynamics. In case of hydrodynamic interactions, you will obtain a more complex curve. (See for instace: http://hera.physik.uni-konstanz.de/igk/news/workshops/Naegele.pdf)