Papers in Press. Published November 2, 2011 as doi:10.1373/clinchem.2011.175281 The latest version is at http://www.clinchem.org/cgi/doi/10.1373/clinchem.2011.175281 Clinical Chemistry 58:2 000 – 000 (2012)
Letters to the Editor
Do Platform-Specific Factors Explain MicroRNA Profiling Disparities? To the Editor: Because of the paucity of information on microRNA (miRNA)1 populations in many body fluids, we read with great interest the report by Weber et al. on the miRNA profiles of 12 body fluids (1 ). This important work suggests that fluids—some obtained noninvasively—are rich sources of miRNA biomarkers. Using quantitative PCR (Qiagen), the authors determined that all of the fluids examined have a complement of several hundred miRNAs, from approximately 200 in urine to ⬎450 in saliva. Some miRNAs, they found, are specific to a single fluid, whereas others, such as miR335*, miR-377*, miR-518e, and miR-616*, are among those with the highest abundance, as estimated by the threshold cycle of amplification (Cq), and occur in multiple types of fluids. We recently reported that a change in the production of 6 plasma miRNAs predicts the development of lentivirus-associated central nervous system disease (2 ), and we compared the plasma findings of Weber et al. with data from our ongoing work. Like Weber et al., we used quantitative PCR (Life Technologies) and global normalization. We were surprised to detect only a third of the miRNAs listed by Weber et al. as the 100 most abundant. Some were not detected in all pools. None of the top 100 miRNAs of Weber et al. were among our 20 earliest-amplifying targets. Conversely, more than half of the 100 miRNAs that had the highest abundance in our study were not detected by Weber et al.
Nonstandard abbreviations: miRNA, microRNA; Cq, threshold cycle (in real-time PCR).
Fig. 1. miRNAs in each data set [Arroyo et al. (3 ), Weber et al. (1 ), and Witwer et al. (unpublished)] were ranked by threshold cycle. Linear regression was performed with the ranks of miRNAs detected by Arroyo et al. and the corresponding ranks in the Weber et al. data and in our data set. Dashed lines indicate 95% CIs. The slope of the Arroyo–Weber line was not significantly different from zero (P ⫽ 0.86), whereas the data sets of Witwer et al. and Arroyo et al. were correlated significantly (P ⬍ 0.0001). Regression of the data sets of Witwer et al. and Weber et al. (not shown) also revealed a slightly negative but nonsignificant slope. For closely related miRNAs (e.g., miR-103/miR107) that were measured together in the Qiagen assay used by Weber et al., the rank was set for the Arroyo et al. and Witwer et al. data at the lowest Cq for the group.
This lack of overlap prompted us to consult a data set, from Arroyo et al. (3 ), obtained with quantitative PCR (Exiqon). All but one of the top 20 miRNAs from our study and that of Arroyo et al. were detected in both studies; our top 20 fell within a median of 11.5 in rank position of the corresponding Arroyo et al. miRNAs. Only 3 of the top 20 miRNAs from the study of Weber et al. were detected by both Arroyo et al. and our group, and all 3 were outside the top 50. Linear regression with the Arroyo data set and ours (Fig. 1) revealed a slope that was significant different from zero (P ⬍ 0.0001), a finding absent in the comparison with the data of Weber et al. (P ⫽ 0.86). Although they are too numerous to cite here, various published reports have indicated that well-studied miRNAs that were undetected by Weber et al. (e.g., miR-16, miR-17, miR19b, miR-20a, miR-223) are abundant in plasma. In short, the agree-
ment between the data of Weber et al. and other published data is limited. That Weber et al. also report miRNAs putatively abundant in plasma to be also abundant in other fluids suggests that these discrepancies are not restricted to plasma. Slight differences in abundance rank can be traced to numerous factors. Because the various primers and probes differ in their hybridization kinetics, Cq provides only an approximate measure of abundance without a calibration curve for each target. This problem is compounded by the heterogeneity within miRNA species, which authors of the Weber et al. report have described elsewhere (4 ). Additional sources of variation include differences in sample processing, the specificities of mature vs precursor miRNAs, data filters, and normalization strategies. The difficulty of obtaining consistent results with low-abundance RNAs 1
Copyright (C) 2011 by The American Association for Clinical Chemistry
Letters to the Editor from biological fluids has been noted (5 ). However, the striking discrepancies reviewed here—with the top miRNAs in one study going undetected in another—are not easily explained and give one serious pause. It is important to address such disparate results because Weber et al. have provided the only published profiles to date for several fluid types. Because Weber et al. evaluated no biological or technical replicates but instead made measurements of a single pool of 5 samples for each fluid, samples from a single individual with an unreported pregnancy or undiagnosed disease might have skewed results by containing miRNAs that would otherwise have been absent. Yet, pooling 5 samples would have diluted such effects and thereby worked to keep aberrantly produced miRNAs out of the top tier of abundance. Unfortunately, Weber et al. did not present the results of individual verification experiments to resolve these questions. Array template issues could explain disparities. We recall a recent occasion when puzzling profiling results led us to investigate a manufacturer’s array layout template. We discovered an error that, if uncorrected, would have effectively rendered any data useless, and we convinced the manufacturer to correct the template. The problem was not publicized, however, and previous customers were not alerted. One wonders how
Clinical Chemistry 58:2 (2012)
many promising projects were abandoned because profiling data from this platform could not be validated. Indeed, the chilling effects of inconsistent data on scientific exploration prompted us to write this letter. We encourage Weber et al. to revisit their data set and determine whether template or other platformspecific issues could explain the differences we have noted. We also emphasize the importance of biological replicates and reporting the results of validation experiments. miRNA profiling of body fluids has opened up a new realm of biomarker discovery, but it can have no meaningful clinical application until findings are consistent and method independent.
trum in 12 body fluids. Clin Chem 2010;56: 1733– 41. Witwer KW, Sarbanes SL, Liu J, Clements JE. A plasma microRNA signature of acute lentiviral infection: biomarkers of central nervous system disease. AIDS 2011;25:2057– 67. Arroyo JD, Chevillet JR, Kroh EM, Ruf IK, Pritchard CC, Gibson DF, et al. Argonaute2 complexes carry a population of circulating microRNAs independent of vesicles in human plasma. Proc Natl Acad Sci U S A 2011;108: 5003– 8. Lee LW, Zhang S, Etheridge A, Ma L, Martin D, Galas D, et al. Complexity of the microRNA repertoire revealed by next-generation sequencing. RNA 2010;16:2170 – 80. Zubakov D, Boersma AW, Choi Y, van Kuijk PF, Wiemer EA, Kayser M. MicroRNA markers for forensic body fluid identification obtained from microarray screening and quantitative RT-PCR confirmation. Int J Legal Med 2010;124:217–26.
Andria K. Watson2 Kenneth W. Witwer2* 2
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article. Authors’ Disclosures or Potential Conflicts of Interest: No authors declared any potential conflicts of interest. Acknowledgments: The authors thank Joel N. Blankson for providing plasma samples and are grateful to Janice E. Clements for support and helpful comments.
References 1. Weber JA, Baxter DH, Zhang S, Huang DY, Huang KH, Lee MJ, et al. The microRNA spec-
Department of Molecular and Comparative Pathobiology Johns Hopkins University School of Medicine Baltimore, MD * Address correspondence to this author at: Department of Molecular and Comparative Pathobiology Johns Hopkins University School of Medicine 733 North Broadway, Suite 831 Baltimore, MD 21205 Fax 410-955-9823 E-mail: [email protected]
Previously published online at DOI: 10.1373/clinchem.2011.175281