Can big data estimate prevalence of defective DNA repair variants?
Acta Eruditorum
Dr. Schwarzenberger is the former physician editor of DermWorld. She interviews the author of a recent study each month.
By Kathryn Schwarzenberger, MD, March 1, 2019
In this month’s Acta Eruditorum column, Physician Editor Kathryn Schwarzenberger, MD, talks with Kenneth Kraemer, MD, and Jennifer Pugh, BS, about their recent JAMA Dermatology article, “Use of big data to estimate prevalence of defective DNA repair variants in the U.S. population.”
Dr. Schwarzenberger: Can you briefly describe your study?
Dr. Kraemer and Pugh: With the increasing use of exome sequencing, we decided to see how closely “big data” corresponded with our clinical observations of xeroderma pigmentosum (XP). XP is a rare, recessively inherited, cancer-prone disease that we have been studying at the National Cancer Institute for more than four decades. Patients with XP have defective DNA repair and more than a 10,000-fold increased risk of sunlight-induced skin cancer. Based on the literature and our clinical observations, we estimate that there are about 300 XP patients in the U.S., corresponding to a frequency of about one per million population. However, the results of our study suggest very different frequencies. We looked at three large databases that included more than 200,000 alleles. We identified a total of 156 XP associated mutations 65 of those mutations — were listed in gnomAD which provides information such as genetic variations, allele count, allele frequency, and number of homozygotes.
Dr. Schwarzenberger: Were you surprised by your findings?
Dr. Kraemer and Pugh: We were very surprised. We knew that these 156 mutations were associated with XP by the patients in our cohort at the NIH and the listings in the Human Genome Mutation database. However, the numbers were so surprising that we took a closer look at these variants. We found that two variants alone had frequencies estimating that there should be more than 8,000 people with XP in the U.S. with these mutations. Yet only four individuals have been clinically identified.
Dr. Schwarzenberger: Your findings would suggest that we cannot necessarily equate genotype with phenotype in XP.
Dr. Kraemer and Pugh: Yes, there is a large discordance in the expected number of XP patients based on the frequency of mutations and the actual frequency of clinically diagnosed patients. Based on our results, it appears that disease-associated genotypes might not always yield an expected phenotype. The high frequency of XP mutations, compared to the low prevalence of clinical XP, might suggest that there are other manifestations of these mutations outside of the known XP phenotype. Mutations in known DNA repair genes may contribute to the high frequency of skin cancers in the general population and/or unexplained late onset neurodegeneration. Alternatively, our understanding of the severity of the features associated with these mutations may not be correct.
Dr. Schwarzenberger: Do you think your findings are likely to be true for other rare diseases? What are the next steps?
Dr. Kraemer and Pugh: Others might discover the same discordance when studying rare diseases. There are a few other studies that have reported similar observations with dominantly inherited disorders, including Li-Fraumeni syndrome and DICER1 cancer predisposition syndrome. Previous studies have started by identifying patients with known clinical features and then performing DNA sequencing to determine which genes are defective. We used the reverse approach and began with the data from DNA sequencing and tried to determine the frequency of the predicted phenotype. The results emphasize the importance of following up with patients who might possibly have genomic abnormalities to determine whether they will develop any clinical manifestations. With our study, we are limited in the ability to follow-up on our observations because the DNA was obtained from donors who were promised anonymity. However, a new NIH precision medicine initiative study (All of Us at https://allofus.nih.gov) has a goal of obtaining clinical information as well as DNA sequences from one million volunteers of diverse genetic backgrounds. This will help us better understand the complex relationship of genotype-phenotype correlation as we enter into this developing era of genomics.
Dr. Schwarzenberger: Your study seems to be a cautionary tale about the reliability of big data. As we amass information about dermatologic diseases through DataDerm™ the Academy’s clinical data registry — do you have thoughts about similar potential pitfalls?
Dr. Kraemer and Pugh: This is a new era. The data available have great potential but there is still much work to be done. This study suggests that gathering clinical information, along with genetic information, can provide a more in-depth understanding of different clinical variations. Until we gain a better understanding of these large data sets, clinicians should approach large genomic databases with caution when trying to correlate genetic variants with prevalence of disease risk. This caution is especially relevant with prenatal testing and identifying disease-associated mutations that may never be expressed.
Kenneth Kraemer, MD, is a dermatologist and chief of the DNA Repair Section in the Laboratory of Cancer Biology and Genetics, Center for Cancer Research, National Cancer Institute, in Bethesda, Maryland. At the NIH, he has studied clinical and laboratory abnormalities in patients with xeroderma pigmentosum since 1971. Jennifer Pugh, BS, was an NIH Academy Enrichment Program Scholar, Office of Intramural Training & Education, Office of the Director, NIH. She performed this study while training in Dr. Kraemer’s laboratory. She is now a first-year medical student at Wright State University, Boonshoft School of Medicine, in Dayton, Ohio, and is interested in dermatology. Their article appeared in JAMA Dermatology. doi:10.1001/jamadermatol.2018.4473.
Funding and support: This research was supported by the Intramural Research Program of the National Institutes of Health, National Cancer Institute, Center for Cancer Research, and Division of Cancer Epidemiology and Genetics.
