GAN-based I2I translation techniques for unpaired data are employed for the synthesis of biometric finger and hand vein presentation attack instrument (PAI) samples corresponding to three public presentation attack datasets. Compared to earlier work in this area, we extend the set of available synthetic PAI samples in their number and in subjects considered. By doing so, we are also able to extend the used training set by 20%. In the conducted vulnerability assessment using these synthetic samples in presentation attacks, we find the resulting Impostor Attack Presentation Match Rate (IAPMR) to be increased as compared to (i) using real PAI data and to (ii) using the synthetic data generated earlier with a smaller training set, respectively. However, this is only true for high quality synthetic data as generated with CycleGAN, while for low quality synthetic data (e.g. generated by DRIT), the impact on IAPMR is not predictable and depends on the actual recognition scheme used in the vulnerability assessment. Thus, when employing high quality synthesis methods, the generation of artificial PAI samples can be used to create datasets with higher attack capability as compared to their real counterparts.