Background: The genetic landscape of CRC tumors may influence responses to targeted therapies. In CodeBreaK 300 (NCT05198934), sotorasib (soto) 960 mg plus panitumumab (pani) showed significant improvement in progression-free survival (PFS) compared with trifluridine/tipiracil (T/T) or regorafenib (rego) (hazard ratio, 0.48 [95% CI, 0.30-0.78]; P = 0.005) in patients with chemorefractory KRAS G12C-mutated mCRC. Here, we evaluated baseline genetic co-occurring alterations and their impact on the efficacy of soto plus pani in CodeBreaK 300.

Methods: Study procedures and eligibility criteria have been previously reported. Next-generation sequencing was performed on baseline tumor and plasma circulating tumor DNA (ctDNA) samples with Tempus xT and Guardant Infinity platforms, respectively. Biomarker status was correlated with clinical outcomes.

Results: A total of 140 available tumor samples and 154 plasma samples were examined. The prevalence of individual gene alterations was similar between arms for both tumor and ctDNA assessments The most common pathogenic KRAS G12C co-alterations occurred in APC (82.1%), TP53 (72.1%), SMAD4 (23.6%), and PIK3CA (17.9%) in tumor samples and in TP53 (73.4%), APC (65.5%), DNMT3A (39%), TET2 (22.8%), and SMAD4 (21.4%) in plasma samples. An exploratory analysis to determine associations between baseline molecular alterations with ≥10% prevalence and clinical outcomes showed that TGF-β pathway mutations were associated with shorter overall survival and time to progression (TTP) and TP53 mutation status was associated with shorter TTP. Among the most prevalent co-alterations, clinical benefit with soto 960 mg plus pani vs T/T or rego was observed irrespective of biomarker subgroup. A hypothesis-driven analysis indicated that alterations in ARID1A in a few patients were associated with shorter PFS. In patients receiving soto 960 mg plus pani, median PFS was 2.07 months (95% CI, 1.84-3.84; n = 3) for those with ARID1A alterations and 5.78 months (95% CI, 4.24-7.59; n = 49) for those without.

Conclusion: Genetic alteration rates in CodeBreaK 300 were similar across all treatment arms and comparable to those observed in CodeBreaK 100 and 101 and KRAS G12C+ cases in AACR Genie BPC v1.1. Soto 960 mg plus pani demonstrated consistent clinical benefits compared with T/T or rego across various molecularly defined subgroups. While no additional strong biomarkers associated with response were identified, this study supported the potential association of ARID1A alterations with shorter PFS as previously reported in CodeBreaK 101. However, the overall low prevalence of pathogenic ARID1A alterations largely limited the interpretation of these findings.