The H-zone paradox
Textbook Mohs triage says the H-zone is the high-risk territory — nose, ear, eye, lip. You'd expect it to need the most sections. In this cohort of 408 procedures, the opposite happens. L-zone tumours require almost twice as many sections on average. Why?
Three zones, three stories
nose · ear · eye · lip · temple
Mean sections
9.0
Mean area (cm²)
2.09
% ≥ 13 sections
41%
Sections / stage
3.0
Anatomically high-risk for recurrence. Cosmetically sensitive; surgeons tend to take smaller, more conservative stages.
forehead · cheek · scalp
Mean sections
12.3
Mean area (cm²)
5.57
% ≥ 13 sections
59%
Sections / stage
4.6
Intermediate anatomical risk.
neck · hand · trunk · pretibial
Mean sections
15.7
Mean area (cm²)
12.55
% ≥ 13 sections
89%
Sections / stage
8.6
Anatomically low-risk — but L-zone tumours in this cohort are ~6× larger than H-zone tumours and require many more sections.
See the flip
Compare the three zones side by side. Switch the metric to spot the lurking variable.
Zones compared by…
Toggle between these two metrics to see the paradox unfold.
Counterintuitive reading: L-zone (lowest anatomical risk) requires the most sections. H-zone (highest risk) is lowest.
Key finding
Size explains section count, not anatomical risk label.
L-zone tumours in this cohort averaged 12.55 cm² versus 2.09 cm² for H-zone — a six-fold difference. Patients only end up with a Mohs referral on L-zone sites (neck, trunk, hand) when the tumour has grown large enough to warrant it.
Smaller H-zone tumours reach Mohs earlier because of aesthetic concerns, producing a lopsided size distribution between zones.
Surgical behaviour clue
H-zone surgeons cut in smaller slices.
Sections per stage:
2.98
H-zone
8.58
L-zone
Same surgeon, same scalpel — but on the nose they take smaller, more conservative cuts to spare cosmetically sensitive tissue. Fewer sections per stage, more stages, same total caution.
Stages-per-case didn't differ significantly between zones (p = 0.11 in the manuscript) — the real variable is slice size.
Every case — area vs sections
Each dot is one of 408 procedures. Toggle zones to isolate. The 13-section horizontal line is the ≥13 outcome cut-off.
Notice how the L-zone cluster (coral) lives in the upper-right — large tumours, many sections. H-zone (teal) is spread left and low.
Caveat from the paper: the L-zone cohort is small (n = 18), so these conclusions need external validation. Still, the pattern is consistent with the size-driven interpretation of the ML model's predictions.