I would load the trial information into an array / list (in a similar way to the way I use in independent randomisation) and then have the main trials loop read from the list rather than the original spreadsheet. I use this method for a Hypertext fiction experiment where the next trial is dependent on the choice just made.