In general, the AR method for estimating XI skewing is considered to accurately reflect skewing in the tissue tested. A generally accepted caveat is that we know it does not necessarily reflect skewing in other tissues (relevant data are largely from mice, with concordant but more limited data in humans). However, the AR marker is the most common methodology used clinically to assay XI skewing in patient populations, mostly because it's minimally invasive. Any other methodology from peripheral blood is likely to have the same problem: it may not reflect skewing in the tissue of interest. The AR test (or really, any other proven method) in biopsied tissue is best in that instance.

What about the threshold of skewing (ratios) and the need for controls? I have read that a 3:1 ratio is the accepted cut-off for actual skewing, although it's not clear to me what visualization and software method this is based upon. I have also read that MIC2 is a worthwhile control because it escapes X-inactivation and therefore can be used to assess the completion of isoschizomer digestion. Do you have insight on this threshold or appropriate controls, Kristy?

Yes, you do have to control for digestion. In our lab, I've just finished validating a new protocol for assaying XI by AR, with a backup locus for those samples homozygous at the AR locus. It's very similar to what's been reported before. I use a third locus, one that has an HpaII restriction recognition site but is not methylated on either X, as a control for complete digestion. Each experiment includes a known random and a known highly skewed sample as controls.

For interpretation, most clinical labs are pretty conservative: anything less skewed than 80:20 is "random", 80:20-89:11 is "moderately skewed," and 90:10 or greater is "highly skewed." These cutoffs are somewhat arbitrary, but there are data showing that about 3% of unselected women are highly skewed, while 30% of known carriers of X-linked intellectual disability are. These cutoffs are what most reference labs use. Not sure whether you're interested in humans, but in animal experiments the binning is much finer. Let me know if you'd like our protocol. I'd be glad to share it.

Hello Bert, I have done some work with two different methods for assessing XI. I have found that methylation at the HpaII AR locus does not necessarily reflect XI based on transcriptional studies. I would like to direct you to my 2008 (Blood, 112(8): 3186-3193) and 2009 (Blood, 114(11): 2357-2358) publications. Please contact me if you require further information on the methodologies or if you would like additional references from studies performed by other authors on this topic.