Patients with colorectal cancer (CRC) frequently develop liver metastases1–3. The prognosis of these patients is skewed by the histopathological heterogeneity of their liver metastases4,5. Patients with ‘replacement’ metastases have a 5-year overall survival of less than 44.2%, compared with 73.4% in patients with ‘encapsulated’ (previously known as desmoplastic) metastases5; yet there are currently no approved therapies targeting replacement liver metastases. Here we show that treatment-naive patients with CRC with liver steatosis have an increased occurrence of replacement metastases compared with patients without steatosis. Mechanistically, we find that steatosis-promoted fatty acid oxidation increases formation of replacement metastases by increasing MYC stability through acetylation. In turn, MYC activates proline synthesis, fuelling collagen production, enabling growth of replacement metastases. Targeting MYC, P5CS or COL1A1 suppresses the occurrence and growth of replacement metastases in patient-derived organoids, mouse or patient-derived xenograft models. Spatial metabolite and protein analyses of liver metastases from patients with CRC further support this mechanism. In conclusion, we provide a mechanistic understanding of the emergence of liver metastases with poor prognosis in treatment-naive patients with CRC, identifying potential targets for therapeutic intervention.
