It would be much easier to retain a digital record than physical material, so why destroy it?
Furthermore, as much of the work was done by UK forensic scientists, I'm confident that UK police would keep a copy with their records.
Jassi, my reply to Vixte concerned the Hazel case (2004). It's possible that the DNA in her case didn't match his, and was ruled out at some point and I missed an episode.
My comment was more general. Tabloids tend to make a hooha about destroying DNA evidence, when in fact it may be simply the biological sample that is, unless a request is made to keep it. Not sure why, possibly due to health and safety. There are also laws in several countries as to how long DNA results can be kept on databases... I'd have to check back as I'm not sure PT even had one back in 2004. However, the lab would presumably have retained the results in their files and the results could have been checked more recently.
In the American lady's case, I doubt that his conviction could have been made on the basis of an mtDNA result from a hair, therefore it would seem more likely to me that they had a full nuclear DNA result. In 5A, there were numerous hairs which could only be tested for mtDNA (as the hairs didn't have roots), and that didn't correspond to any of the mtDNA profiles the PT lab was requested to check them against.
If his mtDNA haplotype (which could also be obtained from that hair) did correspond to one of the unknown ones in 5A, it wouldn't positively identify him, but would include him as a possibility, pending exclusion for other reasons.
The press appears to be mute as to whether his DNA is connected to any of the child molestation cases that occurred prior to Madeleine's disappearance (cf smelly man, etc).
It's possible that, despite his numerous convictions, he was simply unlucky at being in the wrong place at the time, and that someone else was responsible for some of the suspicions against him, but then he needs to be ruled out.