Keiji AI LogoKeiji AI

Competitive intelligence guide

How to build a KRAS G12C clinical trial landscape quickly

A step-by-step method for a KRAS G12C competitive trial landscape from ClinicalTrials.gov, EU registries and abstracts, with a dated table of 24 active trials.

To build a KRAS G12C competitive landscape quickly, query the ClinicalTrials.gov API for "KRAS G12C" with active statuses, normalize drug code names to generic names, de-duplicate, and tag each trial by drug, sponsor, phase, line of therapy and tumor. Then check EU registries (CTIS and the EU Clinical Trials Register) and recent conference abstracts for trials and data the US registry misses. On 1 November 2026 that query returned 144 recruiting, active or not-yet-recruiting trials, 24 of them phase 3; the table below shows the most informative.

What a KRAS G12C landscape needs to show

A portfolio committee usually asks four things. Which drugs are in development? Who sponsors them? How far along is each? Which settings are crowded?

So each row needs a drug, a sponsor, a phase, a setting or line of therapy, a tumor type, a status and a link to the record. Without the link, nobody can check the row.

For KRAS G12C the settings matter as much as the drugs. Most late-stage work is in non-small cell lung cancer (NSCLC), split between previously treated, first-line and resected disease. Colorectal cancer is the second cluster.

Step 1: Pull the trials from ClinicalTrials.gov

The ClinicalTrials.gov API v2 returns structured study records in JSON. We used this query:

https://clinicaltrials.gov/api/v2/studies?query.term=KRAS%20G12C&filter.overallStatus=RECRUITING|ACTIVE_NOT_RECRUITING|NOT_YET_RECRUITING&pageSize=200&countTotal=true

  • query.term searches the whole record, so it catches trials that name the mutation only in eligibility criteria.
  • filter.overallStatus keeps trials that are recruiting, active but not recruiting, or not yet recruiting. Without it, the same search returned 224 trials, including completed and terminated ones.
  • countTotal=true reports the total, so you know whether to page with pageToken.

Run a second pass with query.intr for each drug name and code. This finds trials that name the drug but describe the mutation differently, such as "KRAS p.G12C".

KRAS G12C trial landscape, as of 1 November 2026

The query returned 144 trials: 24 phase 3, 2 phase 2/3, and most of the rest phase 1 or 2. The table keeps the phase 3 trials of G12C-selective drugs and a few earlier trials of newer agents. Settings come from each record's official title.

Drug and regimenSponsor (registry)PhaseSetting or lineTumorStatusRecord
Sotorasib + platinum doubletAmgen3First line, PD-L1 negativeNSCLCRecruitingNCT05920356 (CodeBreaK 202)
Sotorasib + panitumumab + FOLFIRIAmgen3First lineColorectalRecruitingNCT06252649 (CodeBreaK 301)
Adagrasib vs docetaxelMirati Therapeutics3Previously treatedNSCLCActive, not recruitingNCT04685135 (KRYSTAL-12)
Adagrasib + cetuximabMirati Therapeutics3After first-line therapyColorectalActive, not recruitingNCT04793958 (KRYSTAL-10)
Adagrasib + pembrolizumab + chemoMirati Therapeutics3First line, non-squamousNSCLCRecruitingNCT06875310 (KRYSTAL-4)
Divarasib vs sotorasib or adagrasibHoffmann-La Roche3Previously treatedNSCLCActive, not recruitingNCT06497556 (Krascendo 1)
Divarasib + pembrolizumabHoffmann-La Roche3First line, non-squamousNSCLCRecruitingNCT06793215 (Krascendo 2)
Divarasib vs immunotherapy or observationHoffmann-La Roche3Resected stage II–III (adjuvant)NSCLCRecruitingNCT07541170 (KRASCENDO 3)
Olomorasib + pembrolizumab ± chemoEli Lilly3First lineNSCLCRecruitingNCT06119581 (SUNRAY-01)
Olomorasib + immunotherapyEli Lilly3Resected or unresectableNSCLCRecruitingNCT06890598 (SUNRAY-02)
Calderasib (MK-1084) + pembrolizumabMerck Sharp & Dohme3First line, PD-L1 ≥50%NSCLCRecruitingNCT06345729 (KANDLELIT-004)
Calderasib + cetuximab + mFOLFOX6Merck Sharp & Dohme3First lineColorectalRecruitingNCT06997497 (KANDLELIT-012)
Calderasib + pembrolizumab (adjuvant)Merck Sharp & Dohme3Resected stage IIA–IIIBNSCLCRecruitingNCT07431827 (KANDLELIT-013)
Glecirasib (JAB-21822) + JAB-3312Allist Pharmaceuticals3First line, non-squamousNSCLCRecruitingNCT06416410
Garsorasib (D-1553) vs docetaxelChia Tai Tianqing3After standard therapyNSCLCRecruitingNCT06300177
Garsorasib (D-1553) + IN10018InxMed3First line, non-squamousNSCLCRecruitingNCT07174908
Fulzerasib (IBI351) + ivonescimabShanghai Chest Hospital2First lineNSCLCNot yet recruitingNCT06936644
JDQ443 vs docetaxelNovartis3Previously treatedNSCLCActive, not recruitingNCT05132075 (KontRASt-02)
HRS-7058 vs investigator-chosen G12C inhibitorShandong Suncadia3After standard therapyNSCLCNot yet recruitingNCT07601048
JMKX001899 vs docetaxelJemincare3Previously treatedNSCLCNot yet recruitingNCT07242274
Sosimerasib + IN10018Shanghai Jeyou3First line, non-squamousNSCLCNot yet recruitingNCT07833202
HS-10370 + chemo ± adebrelimabJiangsu Hansoh3First line, non-squamousNSCLCNot yet recruitingNCT07822542
D3S-001 alone or in combinationD3 Bio (Wuxi)1/2Advanced, dose escalation and expansionSolid tumorsRecruitingNCT05410145
Elironrasib ± daraxonrasibRevolution Medicines1b/2AdvancedSolid tumorsRecruitingNCT06128551

Not shown: phase 3 trials in which G12C is one arm among several, such as the B-FAST umbrella study (NCT03178552), and pan-RAS trials such as RASolve 301 (NCT06881784). Also not shown: further calderasib phase 3 trials (NCT07190248, NCT07554339) and many investigator-led phase 2 studies.

Registry records change often. Check each record's status before you rely on it.

Step 2: Normalize drug names and de-duplicate

The same drug appears under several names. Map every code name to one generic name before you count anything.

Code name in recordsGeneric nameWhere the link is stated
AMG 510SotorasibNCT04625647 title and interventions
MRTX849AdagrasibNCT04685135 intervention names
LY3537982OlomorasibNCT06119581 intervention names
MK-1084CalderasibNCT06345729
JAB-21822GlecirasibNCT05009329 intervention names
IBI351, GFH925FulzerasibNCT05504278, NCT06936644
D-1553GarsorasibPhase 2 publication (D1553-102 Study Group)
RMC-6291ElironrasibNCT06128551 intervention names

Then de-duplicate:

  • One trial can appear in several registries. Match on the sponsor protocol number and secondary IDs, not on titles.
  • Drop or flag trials where KRAS G12C is a single cohort in a multi-drug or umbrella study.
  • Separate G12C-selective inhibitors from pan-RAS or RAS(ON) multi-selective drugs. Both appear in a KRAS G12C search.
  • Exclude observational and real-world studies unless the committee wants them.

Step 3: Check EU registries

Some trials run in Europe are not on ClinicalTrials.gov, or appear there late. Check both EU sources:

Search both by drug name and code name. Record the EU trial number alongside the NCT number so the same trial is not counted twice.

For KRAS G12C, many early trials come from Chinese sponsors. Check the national registry of each sponsor's home country too.

Step 4: Add conference abstracts and publications

Registries tell you what is running. They rarely tell you how it is going. Results first appear at major oncology meetings, often months before a journal paper.

  • Search abstracts from the main lung and gastrointestinal oncology meetings by drug name and code name.
  • Link each abstract to its trial by NCT or EU number. Abstracts often cite the number.
  • Record the data cut, the number of patients and the endpoints reported. Do not compare response rates across trials with different lines and populations.
  • Search PubMed for full papers, such as the garsorasib phase 2 trial in Lancet Respiratory Medicine (2024).

How TrialMind's trial landscape workflow does this

TrialMind, from Keiji AI, has a trial landscape workflow for competitive and portfolio intelligence teams. According to the solution page:

  • It pulls trials from ClinicalTrials.gov, the EU Clinical Trials Register (EudraCT) and other registries, and organizes them by phase, sponsor and status.
  • It reads publications and conference abstracts directly, and links every claim to the trial record, publication or abstract it came from.
  • You describe the task, review the proposed plan before anything runs, then check the result with its sources.
  • The solution page shows an example KRAS G12C landscape of 166 trials across 12 drugs, built from ClinicalTrials.gov (searched October 2026). Its search scope differs from the query above, so the counts are not directly comparable.

The table on this page was built by hand from the public API, not by TrialMind.

Frequently asked questions

How many KRAS G12C trials are active right now?
On 1 November 2026, a ClinicalTrials.gov search for "KRAS G12C" returned 144 trials that were recruiting, active but not recruiting, or not yet recruiting. Twenty-four were phase 3. Counts change weekly, so rerun the query.
Which KRAS G12C inhibitors are in phase 3?
In the registry on 1 November 2026, phase 3 trials included sotorasib, adagrasib, divarasib, olomorasib, calderasib, glecirasib, garsorasib, JDQ443, HRS-7058, JMKX001899, sosimerasib and HS-10370.
Is ClinicalTrials.gov enough for a competitive landscape?
It is the best place to start, but not enough on its own. Check CTIS and the EU Clinical Trials Register, national registries for sponsors outside the US and EU, and recent conference abstracts for results.
How do I avoid counting the same trial twice?
Match trials across registries by sponsor protocol number and secondary IDs, map drug code names to generic names, and flag umbrella studies where G12C is one cohort among many.
Can an AI tool build this landscape for me?
Yes. Tools such as TrialMind read the registries, publications and abstracts and return a sourced landscape. Whatever tool you use, keep a link to each record so the result can be checked.

See TrialMind on your own question

A 30-minute walkthrough on a study like yours, with sources and code you can check.