Monoallelic Methylation of the Human Androgen Receptor Gene as Detected by the HUMARA Clonality Assay Does Not Consistently Reflect X-Chromosome Inactivation and Is Therefore Unreliable for Assessment of Clonal Hematopoiesis.

Abstract 4566

Even in the absence of a disease specific chromosomal marker, clonality can be assessed in somatic tissues of female origin by using assays based on the pattern of X-chromosome inactivation. The most widely used technique for quantifying X-chromosome inactivation is the HUMARA method that is based on the putative role of DNA methylation at CpG sites close to trinucleotide repeats in exon 1 in silencing of the AR locus. However, using the HUMARA method, we and others have observed that approximately 30% of healthy elderly volunteers appear to have clonal hematopoiesis. This observation is at odds with the concept that normal hematopoiesis is a polyclonal process, suggesting that the finding of monoclonality or oligoclonality in a high percentage of healthy volunteers by using the HUMARA method, is a technical artifact. To address this issue, we developed a clonality assay based on gene expression of five X-linked polymorphic genes and found no evidence of clonal hematopoiesis in healthy elderly volunteers, although we confirmed extreme skewing of X inactivation (consistent with monoallelic methylation of AR) in 30% of the study subject when analyzed by HUMARA (Swierczek et al., Blood 2008). In the present studies, we have validated the accuracy and reproducibility of our quantitative transcription-based clonality assay (qTCA) using two different methods for quantifying gene expression and compared the results with those obtained using the HUMARA method. DNA and RNA were extracted from peripheral blood samples from 31 healthy female volunteers (age in years as follows: range, 22-55; mean, 35; median, 34). RNA was reverse transcribed (RT) and analyzed by using our qTCA in which expression of three polymorphic genes (MPP1, IDS and FHL1), that are subject to X inactivation, is quantified by allele-specific, real-time RT-PCR. Based on DNA analysis, 25 of the 31 (80%) volunteers were polymorphic for at least one of the test genes. Results are reported as the percentage of each of the two single nucleotide polymorphisms (SNPs) that is present in the sample (e. g., 60% A; 40% G). PCR primers are designed to provide maximum discrimination between SNPs with >13 PCR cycles (i. e., 13 log2) separating true-positive from false-positive amplification. Aliquots of the isolated RNA from the test samples were sent to an independent investigator (JJ), at a separate institution, who was blinded to the results of our qTCA, and the allele ratio was determined by using a different technique (quantitative pyrosequencing). Comparison of the results, confirmed the accuracy and reproducibility of the two methods with coefficients of correlation for each gene as follows: (MPP1, r=0.9385; IDS, r=0.8565; FHL1, r=0.8657). One of the 25 informative females (4%) showed extreme skewing (SNP ratio >75%:25%) of X inactivation by both methods. Based on allelic differences in the number of CAG repeats, 29/31 participants were informative in the HUMARA. Among most of the samples, a good correlation was observed between the pattern of X chromosome inactivation as determined by HUMARA and that determined by both qTCA and quantitative pyrosequencing, however, 8/29 (27%) samples analyzed by the HUMARA showed extreme skewing of allele methylation (ratio >75%:25%). Of the 8 subjects with extreme skewing, 3 were homozygous (i. e., non-informative) for all of the X-chromosome polymorphic genes used in the qTCA. Samples from the 5 informative participants were analyzed by using the qTCA, and, in contrast to the HUMARA results, only one subject showed extreme skewing of the SNP ratio (the same subject as identified in the original qTCA). We also quantified HUMARA gene expression using the difference in the number of exon 1 CAG repeats between the two AR alleles as the polymorphic marker. These experiments showed that, of the 8 volunteers with skewing of X inactivation based on HUMARA, 5 had skewing of AR allele expression and 3 had expression of both AR alleles, indicating that the correlation between DNA methylation at the AR locus and AR mRNA transcription is inconsistent. In conclusion, we found a good correlation between the HUMARA and qTCA in some females; however, this was not the case in many healthy females both elderly and young. These experiments demonstrate the accuracy and reproducibility of the qTCA and confirmed that this technique is not subject to the artifact of aberrant skewing of X-inactivation due to monoallelic methylation of AR that limits the applicability and value of the HUMARA.