Reply to Veitia

To the Editor:

Establishment of common methods is an advantage for comparability of studies. We have read the insightful Correspondence in EJHG in which Veitia addresses the performance of two recently published formulas for transforming intensity signals from SNP-array data of mosaic loss of chromosome Y (LOY) into the percentage of cells with LOY in a sample. Acquired loss of chromosome Y is the most common somatic mutation in human blood cells [1,2,3,4,5,6,7,8] and affected men have an increased risk for all-cause mortality [1, 5]. Furthermore, LOY in blood has been found to be associated with risk for various human disease such as hematological and non-hematological cancers, autoimmune conditions, Alzheimer’s disease, cardiovascular diseases, schizophrenia, diabetes, and age-related macular degeneration [1,2,3, 5, 8] (and references therein). In addition to age, replicated risk factors associated with LOY in blood include smoking [4, 5, 9] and genetic susceptibility [4, 6].

LOY mosaicism has been quantified using technologies such as karyotyping, qPCR, SNP-arrays, droplet digital PCR (ddPCR), and next generation sequencing. A commonly used method to estimate the level of LOY is calculation of the mLRRY [1] from data generated by SNP arrays as the median Log R Ratio of probes located in the male specific part of chromosome Y. Since the mLRRY is non-linear and on an inversed scale, different formulas has been suggested recently for the transformation of individual mLRRY values into the proportion of cells with LOY in a sample, to increase the comparability of readouts from SNP-arrays with other methods. Thus, we recently published in EJHG [8] the empirically derived formula LOY(%) = 100(1 − 2(2*mLRRY)) describing the relationship between mLRRY from Illumina arrays and LOY(%) in blood samples. This formula (referred to as Danielsson’s) was established by comparing LOY-estimations in matched DNA samples that were analyzed using SNP arrays, whole-genome sequencing and ddPCR. A strength of this formula is that it returns the LOY(%) in the range of zero to 100% of affected cells with a considerable concordance with LOY-readouts using independent methods (for example see Fig. 3 in [8]). We also compared the performance of Danielsson’s formula with a previously suggested formula published by Veitia and colleges [10]. Our results showed that Veitia’s formula generated unrealistic estimates in samples with high levels of LOY (see Fig. 4 in [8]).

In the correspondence, the accuracy and usefulness of these two formulas are discussed. As recognized also by Veitia, the two formulas generate similar estimates of LOY(%) in samples with low-level mosaicism. Nevertheless, a comparative analysis indicated that Veitia’s formula would perform marginally better in such samples. It should be noted, however, that this analysis was performed with an incomplete dataset that was extracted from Fig. 2a and b in [8] using the WebPlotDigitizer tool. Thus, even though the full dataset was available in Supplementary Table 3 in [8], the analyzed data omitted one data point in Fig. 2a and included only 71 of the 121 data points in Fig. 2b. It is unclear if and how this limitation in Veitia’s analysis biased the results and conclusions regarding the relative accuracy of the two formulas. Yet, it is suggested in the Correspondence that Veitia’s formula should be preferred for analysis of samples with low-level LOY mosaicism and that Danielsson’s should be used for samples with higher level of mosaicism.

This discussion is appreciated and we agree that it is important to establish guidelines within this field. We however disagree with the suggestion to implement different formulas for samples with low and high levels of LOY for several reasons. It should first be emphasized again that the two formulas generate highly comparable estimates of LOY(%) in samples with low-level mosaicism. Regardless of the statistically significant result in Veitia’s comparative analysis of the two formulas, it should be noted that the biological relevance of this result is rather insignificant. Importantly, as illustrated in Fig. 3 in [8] comparing results from three independent technologies, it is clear that Danielsson’s formula accurately transform mLRRY values to LOY(%) over the entire spectra of mosaicism, from zero to 100% of cells affected with LOY. Linear regression analyses displayed R2 values of 0.965 and 0.959 for comparisons between SNP-array/WGS and SNP-array/ddPCR, respectively. These results show that Danielsson’s formula is sufficient for transformation of mLRRY into LOY(%) also in samples with low level of LOY mosaicism. In contrast, as mentioned above, Veitia’s formula yields unrealistic estimates of LOY(%) in samples with higher levels of LOY mosaicism, even beyond the theoretical maximum of 100%. In this context, Veitia reasons that samples with high levels of LOY mosaicism “concerns a few data points” that would represent an “exception rather than the rule”. However, recent large population studies of normally aging men suggest the opposite; a considerable proportion of aging men are indeed affected with LOY in at least 50% of peripheral blood cells (for example see Fig. 1a in [4]).

In conclusion, we argue that the suggestion to use different formulas for samples with low and high levels of LOY mosaicism adds unnecessary complexity without providing substantial additional value. Furthermore, the comparability of studies in the field would benefit from adopting the same formula for estimating LOY(%) from SNP-array data. As a final point, it is important to consider that Danielsson’s formula was optimized for mLRRY data generated from Illumina’s SNP-arrays. Further studies are needed to evaluate similar formulas for LOY(%) transformations from intensity data generated by DNA-arrays from other manufacturers.

References

  1. 1.

    Forsberg LA, Rasi C, Malmqvist N, Davies H, Pasupulati S, Pakalapati G, et al. Mosaic loss of chromosome Y in peripheral blood is associated with shorter survival and higher risk of cancer. Nat Genet. 2014;46:624–8.

    CAS  Article  Google Scholar 

  2. 2.

    Forsberg LA, Gisselsson D, Dumanski JP. Mosaicism in health and disease - clones picking up speed. Nat Rev Genet. 2017;18:128–42.

    CAS  Article  Google Scholar 

  3. 3.

    Forsberg LA. Loss of chromosome Y (LOY) in blood cells is associated with increased risk for disease and mortality in aging men. Hum Genet. 2017;136:657–63.

    CAS  Article  Google Scholar 

  4. 4.

    Wright DJ, Day FR, Kerrison ND, Zink F, Cardona A, Sulem P, et al. Genetic variants associated with mosaic Y chromosome loss highlight cell cycle genes and overlap with cancer susceptibility. Nat Genet. 2017;49:674–9.

    CAS  Article  Google Scholar 

  5. 5.

    Loftfield E, Zhou W, Graubard BI, Yeager M, Chanock SJ, Freedman ND, et al. Predictors of mosaic chromosome Y loss and associations with mortality in the UK Biobank. Sci Rep. 2018;8:12316.

    Article  Google Scholar 

  6. 6.

    Thompson DJ, Genovese G, Halvardson J, Ulirsch JC, Wright DJ, Terao C, et al. Genetic predisposition to mosaic Y chromosome loss in blood. Nature 2019;575:652–7.

    CAS  Article  Google Scholar 

  7. 7.

    Forsberg LA, Halvardson J, Rychlicka-Buniowska E, Danielsson M, Moghadam BT, Mattisson J, et al. Mosaic loss of chromosome Y in leukocytes matters. Nat Genet. 2019;51:4–7.

    CAS  Article  Google Scholar 

  8. 8.

    Danielsson M, Halvardson J, Davies H, Torabi Moghadam B, Mattisson J, Rychlicka-Buniowska E, et al. Longitudinal changes in the frequency of mosaic chromosome Y loss in peripheral blood cells of aging men varies profoundly between individuals. Eur J Hum Genet. 2020;28:349–57.

    CAS  Article  Google Scholar 

  9. 9.

    Dumanski JP, Rasi C, Lonn M, Davies H, Ingelsson M, Giedraitis V, et al. Smoking is associated with mosaic loss of chromosome Y. Science. 2015;347:81–3.

    CAS  Article  Google Scholar 

  10. 10.

    Grassmann F, International AMDGC, Weber BHF, Veitia RA. Insights into the loss of the Y chromosome with age in control individuals and in patients with age-related macular degeneration using genotyping microarray data. Hum Genet. 2019;139:401–7.

    Article  Google Scholar 

Download references

Acknowledgements

Our work is supported by the European Research Council ERC (#679744), the Swedish Research Council (#2017-03762) and Kjell och Märta Beijers Stiftelse to LAF.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Lars A. Forsberg.

Ethics declarations

Conflict of interest

LAF is cofounder and shareholder in Cray Innovation AB.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Danielsson, M., Halvardson, J., Mattisson, J. et al. Reply to Veitia. Eur J Hum Genet (2021). https://doi.org/10.1038/s41431-020-00802-9

Download citation

Search

Quick links