FULCRUM CDS
Melanoma — cutaneous
DEMOJames MorrisonAisha PatelRobert ChenMaria Santos|Enter Real Patient
AP

Aisha Patel

54 y/o FemaleMRN: 0931582Melanoma — cutaneousBRAF V600E positivePD-L1 60%
Stage IV (M1c)Window closing — act now

Treatment Window

S3 Withdraw
The immune system is withdrawing from the fight. This is the critical decision point. Checkpoint blockade can re-engage the immune response, but the window is closing. Delay significantly reduces response probability.
Critical question: Does the immune system have enough firepower?
If Immune Advantage is above 1.0×, checkpoint blockade can re-engage the retreating immune system. If below 1.0×, ICI alone is unlikely to be sufficient.
▲ HIGH URGENCY — treatment window closing

Immune Advantagei

Likely Responder
4.9×
Immune advantage (adjusted: 4.8×)
Immune system has a clear advantage over tumour proliferation. Checkpoint blockade is structurally indicated.
Structural analysis available
If scRNA-seq data with cell type annotations is available, enter the fractions below for structural drain/signal analysis.
Primary drain
Innate signal
Secondary drain

Step Coverage

2/4 Steps
Detection
Fianlimab
Priming
Open
Encounter
Cemiplimab
Maintenance
Open
AgentStepMechanism
FianlimabAnti-LAG-3 — restores T cell differentiation
CemiplimabAnti-PD-1 — releases T cell brake
Open Escape Routes
Priming, Maintenance steps uncovered. Adding a TIGIT inhibitor would NOT help — same Encounter step.

Structural Recommendations

?

Consider adding CTLA-4 (ipilimumab/tremelimumab)

CTLA-4 operates at the Priming step — a different mechanism from PD-1. Extends coverage.

FULCRUM CDS is a structural analysis tool. It does not replace clinical judgement. Not FDA approved.

Trial Evidence — Melanoma — cutaneous

Observed ORR
RegimenTrialORRnYear
dab+tramCOMBI-d69.0%2112015
bini+encoCOLUMBUS63.0%1922018
ipi+nivoCheckMate 06758.0%3142015
nivoCheckMate 06640.0%2102015
pembro_mKEYNOTE-00633.7%2792015
ipiMDX010-2010.9%1372010
Source: published Phase II/III trial results. Observed outcomes, not model predictions.