This methods paper by Geisenberger and colleagues (LMU Munich; Biology Methods and Protocols, published 8 December 2025) presents ht-HiTE, a high-throughput adaptation of the Tris-mediated FFPE DNA extraction protocol published by Oba et al. in 2022. The workflow uses only off-the-shelf reagents (Tris-HCl, SDS, Proteinase K, mineral oil), fits 96-well plates, can run manually or on a Beckman Biomek i5 liquid handler, and offers both SPRI-bead and silica-column purification .
Key quantitative results: (1) Manual 24-h HiTE yielded 15.8 ng/µl, statistically indistinguishable from the Maxwell CSC FFPE kit reference (P>.65, ANOVA); 24-h incubation significantly outperformed 1 h (P=2×10⁻⁵). (2) Automated plate processing of 92 FFPE samples plus 4 empty controls succeeded; bead purification yielded 97.0 vs 68.6 ng/µl for columns (P=0.023, Welch's t-test). (3) Methylation array beta-value correlations for technical replicates reached 0.96–0.98, exceeding between-sample correlations (0.89–0.94). (4) Exome mutational signatures did not differ by extraction method (P=.996, chi-square). (5) The lowest cost configuration (beads, no RNase, diluted binding buffer, subset Qubit correction) reached 1.69 €/sample .
A practically valuable secondary contribution is the Nanodrop correction method: UV spectrometry overestimated DNA concentration ~2-fold versus Qubit, and a plate-specific geometric-mean correction using only 20 concentration-spaced Qubit measurements kept corrected values within 0.5–2× of truth for 94.5% (beads) and 90.4% (columns) of samples — explicitly framed as unsuitable for IVDR-compliant diagnostics .
Strengths: transparent methods, deposited raw data (figshare) and analysis code (GitHub), negative-control wells, electrophoretic corroboration, and honest acknowledgment of the 8 evaporated samples and platform confound. Compared to other cost-reduction efforts — e.g., factorial reduction of reagent volumes in the DNAdvance kit (75% cost savings on deer muscle) or older low-cost FFPE protocols validated only by PCR — this paper's downstream NGS and methylation validation is more rigorous.
Limitations the paper does not fully resolve: downstream validation rests on very small cohorts — only 3 lung cancer samples for methylation and 2 for exome sequencing; reference and ht-HiTE methylation samples were run on different array platforms (EPIC v1 vs v2), a confound the authors acknowledge but cannot fully remove; the yield comparison for the automated workflow confounds purification method with plate identity (one plate each); one colorectal sample dominated manual validation; cost figures use May 2024 German list prices and exclude labor, consumable overheads, and liquid-handler depreciation; and there is no prospective or clinical diagnostic validation — C-to-T FFPE deamination artifacts were present regardless of method, and no UDG/FFPE-repair comparison was attempted . The claim of "equivalent quality" is therefore well-supported for equivalency of gross quality metrics, but powered only to detect large differences — a真 null result here cannot exclude modest batch effects that could matter for machine-learning methylation classifiers.
What would change the conclusion: a blinded multi-lab head-to-head comparison on matched, same-platform methylation arrays with larger n; demonstration that ht-HiTE DNA does not degrade methylation-classifier confidence scores; or evidence of cross-contamination/oil carryover at scale.
Reproducibility: excellent — the protocol is published at protocols.io (dx.doi.org/10.17504/protocols.io.6qpvr3jr3vmk/v2), raw methylation data at figshare, and code at GitHub (cgeisenberger/ht-hite).
Know what changed, what holds up, and what remains uncertain. Every Friday. No ads.