Dear Kristina, welcome:
1. There will always be some background in a sensitive assays like ELISpot/FluoroSpot. Unstimulated T-cells and NK-cells can release IFNg even if they are only incubated in cell culture media. In general, this background tends to be very low but will be increased due to a myraid of reasons. For example, bad batches of FBS are known to cause higher backgrounds in ELISpot/FluoroSpot. Likewise, if cells have been incorrectly handled during isolation, freezing or thawing this can cause increased levels of background. If the viability is poor and the there are many apopototic cells in the sample, release of DAMPs during the incubation can induce background production. One should also consider the possibility that increased levels of background reflect an increased in vivo T-cell activation from the donors and is thus not an artefact in that sense.
2. With properly isolated PBMC from healthy donors I would expect below 10 spots per well in around 90% of donors. For remaining 10% this background can be anywhere from 10-50 spots/well, however, compared to antigen stimulated wells there is still a clear stimulation index. Running patients samples is very different compared to totally healthy controls. In dieases such as cancer and widespread inflammation, background can certainly be much higher. In your attached PDF It looks like elevated levels of background compared to normal, but nothing outragous. In most of the cases tough, there seems to be very little induction of spots in the antigen wells. This could be an indication of poor viability of the cells but might also reflect that you simply don't have an antigen response present. The low number of spots in your aCD3 wells do indicate it is a problem with viability.
3. Yes, the uneven spot distribution with antiCD3 is common. Stimuli should always be added first, and then cells in a secondary step. However, antiCD3 tend to always have a bit scewed distribution despite doing everything correctly. Although it may look "strange" our experience is that counted spot-numbers are very reliable despite the poor distribution.