Prognostic impact of <i>PIK3CA</i> amplifications and co-occurring mutations detected through circulating tumor DNA (ctDNA) in metastatic breast cancer (MBC).

Abstract

1036 Background: Activating PIK3CA mutations are established oncogenic drivers and predictive biomarkers in MBC. In contrast, PIK3CA amplifications are rare, poorly characterized, and of unclear clinical significance. Tissue-based studies report a prevalence of 1–7%, but their detection and relevance in ctDNA are largely unexplored. Moreover, whether amplifications interact with co-occurring mutations to define a distinct high-risk subset is unknown. We therefore aimed to characterize the impact of PIK3CA amplifications detected in ctDNA through a multi-institutional cohort of patients (pts) with MBC. Methods: This retrospective study analyzed a multi-institutional cohort of 1579 pts with MBC and baseline ctDNA testing with the Guardant360 NGS panel within a large academic consortium (PMAC). Hormone Receptor positivity (HR+) and HER2 status were defined based on the most recent biopsy. Associations between SNVs, CNVs, and clinical characteristics were assessed using multivariable logistic regression. The impact of prognosis, adjusted for the number of prior treatment lines, was evaluated through Cox regression for overall survival (OS), defined from time of baseline ctDNA collection. Results: Among 1,579 pts, 1,121 (71%) were HR+/HER2–, 214 (13.5%) HER2+, and 244 (15.5%) had triple-negative breast cancer (TNBC). PIK3CA CNVs were detected in 7.7% of pts, and 44.7% harbored a concomitant PIK3CA SNVs. PIK3CA CNVs were significantly enriched in TNBC compared with other subtypes (Odds Ratio [OR] 2.02, p=0.011). In the overall population, PIK3CA CNVs were associated with significantly worse OS (HR 2.46, p&lt;0.001), an effect observed across all subtypes, including HR+/HER2– (HR 2.21, p&lt;0.001), HER2+ (HR 5.23, p&lt;0.001), and TNBC (HR 1.81, p=0.018). In multivariable analysis, PIK3CA CNVs remained independently associated with inferior OS in the overall cohort (HR 1.61, p=0.015) and in HR+/HER2– disease (HR 1.38, p=0.048). Notably, the coexistence of a PIK3CA CNV and mutation identified a subset with particularly poor prognosis. Compared with patients harboring PIK3CA SNVs alone, those with concurrent SNVs and CNVs had significantly worse OS in the overall population (HR 1.80, p=0.002), in HR+/HER2– disease (HR 1.51, p=0.037), and in TNBC (HR 5.72, p&lt;0.001). Conclusions: PIK3CA amplifications detected in ctDNA are relatively frequent in MBC and are associated with significantly worse survival across molecular subtypes. The coexistence of PIK3CA amplification and mutation identifies a distinct high-risk subset with particularly poor prognosis, beyond the effect of mutations alone. Further mechanistic understanding and real-world outcome analyses are needed to determine whether PIK3CA amplifications, alone or in combination with mutations, have predictive relevance for response or resistance to PI3K-pathway inhibitors.