Effects of DNA extraction, DNA integrity, and laboratory on the precision of qPCR-based telomere length measurement - a multi-lab impartial study

Abstract

Measuring telomere length (TL) with high precision is challenging. Currently there is insufficient understanding of the causes of variation in measurement precision, particularly for qPCR-based measurement. To better understand how DNA extraction protocols and laboratory-specific analytical factors influence qPCR-based TL measurement precision, we conducted a multi-laboratory study involving four labs and six DNA extraction protocols assaying the same blinded human whole blood samples. DNA extraction protocols differed in underlying principles (magnetic beads, salting out, silica membrane) and commercial kits. A fifth lab performed Telomere Restriction Fragment (TRF) analysis using Southern Blot technique with one DNA extraction protocol. All labs performed TL measurement using their standard procedures on two sets of fifty double blinded samples. Data was sent to a central point for unblinding and statistical analyses. Precision was quantified using the Intraclass Correlation Coefficient (ICC). Correlations with TRF measurements were also calculated. Repeated qPCR-based measurements of the same DNA extraction yielded ICC values ranging from 0.24 to 0.94. ICC values calculated over measurements of repeated DNA extractions were on average 0.23 lower and ranged from 0.02 to 0.83. The latter ICC estimates more strongly predicted the association between qPCR- and Southern blot-based measurements across the protocol / lab combinations (R2=0.56 vs. R2=0.93). We conclude that ICC calculated over measurements on repeated DNA extractions reliably indicates measurement precision, while ICC calculated over multiple measurements of the same DNA extraction overestimates measurement precision. Variation in ICC was driven by variation between laboratories, with few consistent DNA extraction protocol effects. Values of DNA integrity and purity generally characterized as reflecting high sample quality, (e.g. OD 260/280 of 1.8 and OD 260/230 of 2.0) were associated with qPCR-based measurement precision, but did not always predict higher ICCs.