Panitumumab

Patients with RAS and BRAF wt mCRC, treated with R or T versus anti-EGFR-based treatment in third-line, were retrospectively included in the study. Patients were enrolled by four Italian Medical Oncology Units (Comprehensive Cancer Center, Fondazione Policlinico Universitario Agostino Gemelli-IRCCS, Università Cattolica del Sacro Cuore, Rome; Ospedale Fatebenefratelli Isola Tiberina - Gemelli Isola, Rome; Department of Medical Oncology, Campus Bio-Medico University, Rome; Ospedale F. Spaziani - ASL Frosinone)
Patients had to have received two prior regimens of standard chemotherapy (oxaliplatin, irinotecan, fluoropyrimidine) for metastatic disease. Previous treatments could include bevacizumab. Patients who received cetuximab and/or panitumumab in first- or second-line were excluded from the anti-EGFR group; on the contrary, they could be enrolled in the R/T group. Prior therapy with R or T was not allowed.
The R-sided tumor was defined as cancer from the cecum to the transverse colon, L-sided tumor was defined as cancer from the splenic flexure to the rectum. For each patient we collected the following available variables: baseline ECOG performance status (PS), gender, age, synchronous vs metachronous disease, previous anticancer treatments, and number of metastatic sites (single vs multiple).

Patient-derived organoid generation was attempted from baseline biopsies of all patients. A total of 10 colorectal tumor organoid lines were successfully generated from tumor baseline biopsies of 10 patients enrolled in the BRAF/MEK/PD-1 inhibition trial (patients 1, 2, 4, 10, 11, 14, 16, 18, 21 and 24). Tumor biopsies were transported in ice-cold RPMI with 10% human serum and transferred into a petri dish on ice before processing. Tumor biopsies were minced and subjected to enzymatic dissociation in 4.75 ml minimum essential medium for suspension cultures (Gibco) supplemented with 250 µl Liberase for 45 min at 37 °C using a heater-shaker. Following the dissociation, tumor biopsies were centrifuged at 300g for 5 min, seeded into Matrigel in a prewarmed 24-well plate and cultured with 500 µl of basal growth media. For passaging, organoids were mechanically pipetted out of Matrigel using Corning Cell Recovery Solution (Corning), followed by a 1-h incubation at 4 °C. Organoid fragments were then centrifuged and subjected to enzymatic dissociation in Tryple E (Gibco) for 5 min at 37 °C. A 20-gauge needle was used to further disrupt the organoids mechanically. DMEM/F12 media supplemented with 10% FBS was added to the conical tube to stop the enzymatic reaction. Dissociated organoids were collected by centrifugation and seeded in Matrigel as above; 10 µM Rko-Kinase inhibitor was added to the basal growth media for organoid passaging. For drug treatment, dissociated organoids were resuspended in basal growth media supplemented with 2% Matrigel and plated in a 24-well plate coated with 250 µl Matrigel. The next day, organoids were treated with dabrafenib (100 nM) + trametinib (10 nM), dabrafenib (100 nM) + ERAS007 (100 nM), or dabrafenib (100 nM) + panitumumab (3 µg ml⁻¹; McKesson) for 72 h. The drugs were refreshed every 24 h. After the treatment, organoids were collected and subjected to RNA extraction. For cell viability experiments, dissociated organoids supplemented with 2% Matrigel were plated in a 96-well plate coated with 70 µl Matrigel; 48 h later, organoids were treated with various doses of dabrafenib + trametinib or dabrafenib + ERAS007 for 72 h and subjected to cell viability measurement using CellTiter-Glo 3D (Promega).

Bulk RNA sequencing data in 71 patients (including 45 paired day 1 and day 15 biopsy samples and 26 separate biopsy samples from baseline) enrolled in the previous BRAF/EGFRi ± MEKi trial with dabrafenib, panitumumab and trametinib in patients with BRAF^V600E CRC were obtained from Novartis^{25 (link)}. RNA sequencing data were trimmed mean of M values normalized^{41 (link)}. Normalized expression data were then corrected for varying levels of liver gene expression using the expression of a 22-gene score (Supplementary Table 5) in a linear model to reduce the impact of biopsy location on the expression data. All expression values are log₂ of liver-corrected counts per million. Gene signature expression levels are the mean log₂ of corrected counts for all genes in the signature. Genes used for gene signature score calculation are listed in Supplementary Table 5.