Alzheimer's "Genes". Another Faux Replication

I will make the same point for this GWAS for Alzheimer's Disease. Again, we have an example of meta-analysis, with new data that has never been studied independently folded into previous studies with a claim of more loci "discovered" and, as one person noted, "the gold standard in field," for replication.  It's amazing how they have watered down replication and now try to call this a gold standard.
So what they did was take an old study on Alzheimers with an N = 63,000 and added a new cohort of N = 18,000.  The original study had 23 loci reaching significance and this addition gave them 25 loci reaching significance.  However, this was 5 new loci, with 3 of the previous failing to reach significance.  Once again, no independent analysis of the new 18,000.  So, there is no attempt to see which loci were significant for the new data and compare it to the previous study.  This would be  the actual "gold standard" for replication.  So we have no idea if any of the loci were actually replicated, independently. This, again, is the shell game of almost all new GWAS studies for cognitive traits.  If we have a study with 63,000 people and it finds 23 loci to reach significance, then what should happen when we add another 18,000?  Shouldn't this only server to reinforce the original loci?
For example, if we flipped a coin 63,000 times and the heads vs. tails ratio was close to 50%, what would be the chances that adding another 18,000 flips would move it farther away from 50% if we are assuming an equal chance of heads or tails?  To what extent then, does losing 3 significant loci indicate that the original "significant" values were false positives.  Perhaps a statistician could actually work this out, but I suspect it is a good indication not only that the original 3 were false positives, but many of those that managed to retain significance only did so because they were bolstered by the original data.  The way this study is constructed, we cannot assess whether any of the previous loci have been replicated and I'm beginning to believe that that is by design.