Skip to content

KRAS G12C Treatment Landscape in 7 Commands

How an AI agent uses BioMCP to assemble a working treatment picture for a KRAS G12C NSCLC patient — drugs, safety, trials, and resistance — in under two minutes.

TL;DR: A clinician has a patient with KRAS G12C non-small cell lung cancer. An AI agent chains 7 BioMCP commands across 10+ data sources to surface drugs targeting KRAS, review FDA regulatory and safety data, find recruiting trials, and retrieve the latest resistance research. Counts shown below are example outputs from a live run and will change as upstream databases update.

KRAS G12C Treatment Landscape

The Problem

Your patient has KRAS G12C-mutated non-small cell lung cancer. You need answers:

  • What drugs target this mutation?
  • Which are FDA-approved? What's the safety profile?
  • What does the post-market adverse event reporting look like?
  • What clinical trials are recruiting?
  • What does the latest research say about resistance?

Each answer lives in a different database: CIViC for curated variant evidence, Drugs@FDA for regulatory status, OpenFDA FAERS for post-market adverse event reporting, ClinicalTrials.gov for recruiting studies, PubMed for recent publications. Assembling the picture manually means a dozen browser tabs and an afternoon of copy-paste.

An AI agent with access to BioMCP does it in seven calls.

How to Think About It

A clinician building a treatment landscape works through a natural cascade:

  1. Orient. What data exists? How big is the landscape?
  2. Check variant annotations. What do knowledgebases say about this specific variant?
  3. Map the drugs. What targets KRAS in the structured drug-target databases?
  4. Review the lead drug. Regulatory status, mechanism, approval history.
  5. Check safety reporting. What does the FDA adverse event reporting system show?
  6. Find trials. What options exist beyond approved therapies?
  7. Track resistance. What's the cutting-edge research on why these drugs stop working?

Each step maps to a BioMCP command.

Step 1: Orient Across All Entity Types

Start broad. BioMCP's search all gives a counts-first overview across every entity type — genes, variants, drugs, trials, articles, pathways, and more — in a single call.

$ biomcp search all --gene KRAS --disease "non-small cell lung cancer" --counts-only

# Search All: gene=KRAS disease=non-small cell lung cancer

## Genes (1)
## Variants (1741)
## Diseases (1)
## Drugs (1)
## Trials (84)
## Articles (42)
## Pathways (0)
## PGx (0)
## GWAS (0)

1,741 known variants in KRAS. 84 trials mentioning both KRAS and NSCLC. 42 articles at the intersection. The landscape is large — the agent now knows to focus on the G12C variant specifically rather than KRAS broadly.

Step 2: Pull Variant Annotations and Evidence Records

Now the agent pulls variant annotations and evidence records from CIViC (Clinical Interpretation of Variants in Cancer) and ClinVar. These are curated knowledgebase entries — some historical, some disease-specific — not a direct therapeutic recommendation.

$ biomcp get variant "KRAS G12C" civic clinvar

# KRAS p.G12C - civic, clinvar
## ClinVar
Top disease (ClinVar): Endometrial carcinoma (1 report)
Conditions (8 reports):
- Endometrial carcinoma (1 report)
- Gallbladder cancer (1 report)
- Lung adenocarcinoma (1 report)
- Non-small cell lung carcinoma (NSCLC) (1 report)
- RASopathy (1 report)
...
## CIViC
### Cached Evidence (MyVariant)
| Evidence | Type | Level | Disease | Therapies |
|---|---|---|---|---|
| EID227 | DIAGNOSTIC | B | Lung Cancer | - |
| EID1217 | PROGNOSTIC | B | Lung Non-small Cell Carcinoma | - |
| EID1216 | PREDICTIVE | B | Colorectal Cancer | Gefitinib, Erlotinib |
...

CIViC provides diagnostic, prognostic, and predictive evidence records across multiple cancer types. Note that the therapies shown here (gefitinib, erlotinib) are from CIViC's historical evidence in colorectal cancer — not the current G12C-targeted therapy story. The current FDA-approved G12C therapies (sotorasib, adagrasib) appear in the next step. ClinVar shows 8 condition associations spanning NSCLC, lung adenocarcinoma, and other cancer types. The variant is well-characterized across multiple knowledgebases.

Step 3: Map Drugs with Structured Target Annotations

What drugs have KRAS listed as a target in the drug-target interaction databases (DGIdb, OpenTargets)?

$ biomcp gene drugs KRAS

# Drugs: target=KRAS

Found 5 drugs

| Name | Mechanism | Target |
|---|---|---|
| garsorasib | small-molecule | KRAS |
| sotorasib | Inhibitor of GTPase KRas | KRAS |
| fulzerasib | small-molecule | KRAS |
| adagrasib | Inhibitor of GTPase KRas | KRAS |
| lonafarnib | Inhibitor of Protein farnesyltransferase | KRAS |

Five drugs from the sources queried. Sotorasib (Lumakras, Amgen) and adagrasib (Krazati, Mirati) are the two U.S.-approved G12C-specific covalent inhibitors. Garsorasib and fulzerasib are investigational. Lonafarnib is a farnesyltransferase inhibitor with a different mechanism.

Note: the broader KRAS G12C pipeline includes additional agents in advanced development (divarasib, olomorasib, glecirasib) that don't appear in this gene-to-drug pivot. Structured target annotations vary across upstream drug-target sources and can lag clinical development. BioMCP can look these agents up individually (biomcp get drug divarasib), and the trial search in Step 6 surfaces them through their clinical programs.

The agent picks sotorasib as the lead approved drug for a deeper look.

Step 4: Review the Lead Drug

What's sotorasib's regulatory history, and what safety data can BioMCP pull?

$ biomcp get drug sotorasib regulatory safety

# sotorasib - regulatory, safety
## Regulatory (US - Drugs@FDA)

### NDA214665

- Sponsor: AMGEN INC
- Brands: LUMAKRAS
- Generic Names: SOTORASIB
| Submission Type | Number | Status | Date |
|---|---|---|---|
| SUPPL | 9 | AP | 2025-01-16 |
| SUPPL | 12 | AP | 2024-09-30 |
| ORIG | 1 | AP | 2021-05-28 |

## Safety (US - OpenFDA)
...

FDA-approved since May 2021 under NDA214665. Multiple supplemental approvals through January 2025 indicate an expanding label. The agent can now check what the real-world safety data looks like.

Step 5: Check Real-World Adverse Events

BioMCP queries the FDA Adverse Event Reporting System (FAERS) directly.

$ biomcp search adverse-event --drug sotorasib --serious

# Adverse Events: drug=sotorasib, serious=true
Found 10 reports

## Summary
- Total reports (OpenFDA): 2465
- Returned reports: 10
| Reaction | Count | Percent |
|---|---|---|
| Dehydration | 3 | 30.0% |
| Oedema peripheral | 2 | 20.0% |
| Acute kidney injury | 1 | 10.0% |
| Acute respiratory distress syndrome | 1 | 10.0% |
| Drug-induced liver injury | 1 | 10.0% |
| Hepatic failure | 1 | 10.0% |
...

OpenFDA FAERS shows 2,465 serious reports mentioning sotorasib in this query. FAERS is a spontaneous reporting system — these counts do not establish incidence or causality, and the percentages reflect only the 10 sampled reports, not the full corpus. These reports are hypothesis-generating: useful for surfacing events worth reviewing alongside the prescribing information, which lists diarrhea, musculoskeletal pain, nausea, fatigue, hepatotoxicity, and cough as the most common adverse reactions for single-agent NSCLC use.

Step 6: Find Recruiting Trials

What trials are recruiting for KRAS G12C NSCLC right now?

$ biomcp search trial -c "NSCLC" --mutation "KRAS G12C" -s recruiting --limit 5

# Trial Search Results

Results: 5 of 62

Query: condition=NSCLC, status=recruiting, mutation=KRAS G12C

| NCT ID | Title | Status | Phase | Conditions |
|---|---|---|---|---|
| NCT06582771 | A Study of Sotorasib in P... | RECRUITING | 2 | Non Small Lung Cancer |
| NCT07012031 | Sotorasib in Combination... | RECRUITING | 1/2 | Locally Advanced Lung Non-Small Cell Carcinoma... |
| NCT06119581 | A Study of First-Line Olo... | RECRUITING | 3 | Carcinoma, Non-Small-Cell Lung, Neoplasm Metastasis |
| NCT06172478 | A Study of HER3-DXd in Su... | RECRUITING | 2 | Advanced Solid Tumor, Melanoma, Head and Neck... |
| NCT04302025 | A Study of Multiple Thera... | RECRUITING | 2 | Non-small Cell Lung Cancer |

62 recruiting trials for KRAS G12C NSCLC. The landscape includes sotorasib combinations, new G12C inhibitors, and ADC (antibody-drug conjugate) approaches. A clinician could use biomcp get trial NCT06582771 eligibility to check specific patient eligibility.

Step 7: Latest Research on Resistance

Acquired resistance is the clinical challenge driving the trial landscape. What does the latest research say?

$ biomcp search article -g KRAS --keyword "G12C resistance" --since 2024 --limit 5

# Articles: gene=KRAS, keyword=G12C resistance

Found 5 articles

| PMID | Title | Source(s) | Date |
|---|---|---|---|
...

Showing 1-5 of N results.

The latest publications cover resistance mechanisms (secondary KRAS mutations, bypass pathway activation), combination strategies designed to overcome resistance, and emerging approaches beyond covalent inhibitors.

What Just Happened

An AI agent assembled a working treatment landscape in 7 commands:

Step Command What It Surfaced
1 biomcp search all Orientation: variants, trials, articles across entity types
2 biomcp get variant CIViC + ClinVar variant annotations and evidence records
3 biomcp gene drugs 5 drugs with structured KRAS target annotations
4 biomcp get drug Sotorasib FDA approval history and regulatory status
5 biomcp search adverse-event FAERS spontaneous reports (hypothesis-generating)
6 biomcp search trial Recruiting trials for KRAS G12C NSCLC
7 biomcp search article Latest resistance research

Seven entity types queried. Ten-plus data sources: CIViC, ClinVar, OpenTargets, DGIdb, Drugs@FDA, OpenFDA FAERS, ClinicalTrials.gov, PubTator3, Europe PMC, and more. Each command took seconds. Not a substitute for clinical judgment — but a fast way to assemble the evidence landscape that informs it.

Try It

uv tool install biomcp-cli

Then start with your own mutation:

biomcp search all --gene BRAF --disease melanoma --counts-only

Documentation: biomcp.org | Code: github.com/genomoncology/biomcp