High-utility itemset mining under periodicity constraints has recently attracted considerable interest because it can identify patterns that are both highly profitable and temporally regular. Nevertheless, the mining process becomes significantly more difficult when itemsets contain both positive and negative utilities. In this paper, we present efficient approaches for the Top-K periodic high-utility itemset mining problem and conduct an evaluation of two representative algorithms, namely TKPHUMN and TKPUFP. The effectiveness of these algorithms is examined using benchmark datasets with different levels of density, specifically mushroom and retail. Experimental results show that dataset characteristics have a strong impact on algorithm performance. On dense datasets, TKPUFP exploits incremental computation and sophisticated pruning strategies to achieve substantial efficiency improvements, operating several orders of magnitude faster than TKPHUMN. In contrast, for sparse datasets, TKPHUMN exhibits superior performance due to its lower inherent computational complexity, whereas the overhead of TKPUFP becomes more evident. These results highlight the respective advantages and limitations of the two approaches and provide practical guidance for selecting appropriate algorithms based on data properties.