Design: Retrospective single-laboratory comparison
Population: 455 cytology specimens and 2,759 FFPE controls on a hybrid-capture NGS panel
Primary outcome: Percentage of targeted bases at ≥100× coverage
Effect: Cell block ≥98%; Papanicolaou smear 74.7%; combined preparations 89.2% (medians)

A single laboratory reviewed 10,900 specimens run on a custom hybrid-capture NGS panel, focusing on 455 consecutive cytology specimens against 2,759 FFPE surgical controls.

Cell-block sections performed like FFPE tissue, with median target coverage at or above 98%. Stained smears produced enough DNA but less uniform coverage — a median of 74.7% of targets reaching 100× for Papanicolaou-stained smears and 89.2% for combined preparations — with more GC bias. The differences came from the smear preparation and staining, not from DNA input or tumour purity.

When a smear is the only material, it can still be sequenced, but it should be validated as its own specimen type rather than assumed to behave like a block. Where there is a choice, a cell block is the better substrate.

Prepare a cell block at the time of the procedure whenever molecular testing is likely.
Prefer the cell block to stained smears for hybrid-capture NGS when both are available.
If sending a smear, flag its type and stain so the laboratory applies the right quality thresholds.
Laboratories should validate smears as a separate specimen type, not under FFPE criteria.

Why it matters

Applying FFPE quality thresholds to smears may misjudge failed or low-coverage results.

The statistics, in plain English

Target coverage is the share of the panel's bases read at least 100 times. A median of 74.7% for Papanicolaou smears means about a quarter of targets fell short in a typical case, which can hide a variant. This is one laboratory's panel, so exact numbers will differ elsewhere.