Identifying the optimal post-surgical timing of molecular residual disease (MRD) detection in colorectal cancer (CRC) using an ultra-sensitive assay: Interim results from the VICTORI study.

275 Background: While detection of MRD using ctDNA is prognostic for recurrence in CRC, some patients still recur prior to MRD detection. VICTORI is prospectively investigating NeXT Personal, an ultra-sensitive NGS-based MRD assay, to profile patients with resected CRC. Methods: Patients with CRC treated with curative intent (all stages) are tested for MRD using NeXT Personal, a bespoke assay with up to ~1,800 tumor-informed single nucleotide variants (SNVs) identified from whole-genome sequencing. Plasma is collected prior to surgery, every 2 weeks post-surgery up to week 8 (MRD landmark window), and every 3 months for up to 3 years (surveillance). We present preliminary results on 397 samples from the first 62 patients. Results: A total of 62 patients (N=36 rectal [58%], N=26 colon [42%]; N=46 stage I-III [74%], N=16 stage IV [26%]) were included in our analysis. Baseline pre-surgical sensitivity (treatment naive) was 93.5% [N=29/31]. Pre-surgical positivity rate in patients who had received neoadjuvant therapy and had residual cancer at the time of surgery was 67% [N=16/24]. 60 patients were evaluable for clinical outcomes. At a median follow-up of 355 days, 15 patients (25%) had a recurrence. Of these, all patients were ctDNA-positive prior to recurrence (100%, 14/14; 1 pt excluded due to lack of samples prior to recurrence). ctDNA detection preceded clinical relapse by a median of 194 days [range: 5-397]); ctDNA for 78.6% (N=11/14) were first detected in the MRD landmark window. All landmark-positive recurrences occurred within one year of surgery. The 14 recurrent cancers with samples were first detected at a median ctDNA concentration of 28.7 parts per million (PPM) (range 2.4-111,120), with 64.3% (9/14) of those detections in the ultra-low range of <100 PPM. MRD detection at week 4 and week 8 had the greatest reduction in RFS (HR 12.86 [2.74-60.28], p=0.0012 week 4; HR 16.14 [3.52-74.09], p=0.0004 week 8), with weeks 4, 6 and 8 having similar higher prevalence of ctDNA detection (36.0% [18/50], 35.3% [18/51], 38.5% [20/52] respectively). Week 2 detection rate was 17.0% (8/46). cfDNA concentration was highest at week 2 (4.63ng/ml vs. 2.22 at baseline, p=0.00028) and 4 (3.68 vs. 2.22, p=0.0081), returning to baseline levels at week 6 (2.28 vs. 2.22, p=0.67) and 8 (2.48 vs. 2.22, p=0.64). Conclusions: In this interim report after a median ~1 year follow-up, NeXT Personal detected MRD for all patients prior to disease recurrence. Most initial MRD detection was in the ultra-sensitive range <100ppm and detection at 4-8 weeks after surgery was highly prognostic for recurrence.