ctDNA-guided anti-EGFR rechallenge in metastatic colorectal cancer: where PURSUIT evidence stops

What earlier evidence established about molecular selection for rechallenge

The rationale for anti-EGFR rechallenge rests on clonal evolution: RAS-mutant subclones that expand under EGFR blockade can decay after drug withdrawal, potentially restoring sensitivity [6]. Foundational prospective and retrospective studies established that patients with ctDNA RAS/BRAF wild-type status derive the greatest benefit, with response rates up to 30% and median PFS exceeding 4 months in molecularly selected populations [6]. A systematic review and meta-analysis of 13 studies, mostly phase II, reported a pooled ORR of 20.5% and DCR of 67.4%, with ctDNA RAS wild-type status associated with improved OS (HR 0.41, 95% CI 0.28–0.60) [8]. The RASINTRO prospective study further showed that ctDNA RAS/BRAF wild-type at cycle 1 predicted longer PFS (3.3 vs 1.9 months; HR 0.43) and OS (7.9 vs 4.9 months; HR 0.46) compared with mutated status [9]. These data collectively made pre-rechallenge ctDNA testing a reasonable selection standard, but they did not resolve whether sampling immediately after prior anti-EGFR progression—before the drug-free interval—adds independent information [1].

PURSUIT's efficacy signal and the ctDNA dynamics it uncovered

PURSUIT enrolled 50 patients with tissue RAS/BRAF V600E wild-type mCRC who had prior response to anti-EGFR therapy, plasma RAS negativity before rechallenge, and an anti-EGFR-free interval of at least 4 months [1]. Panitumumab plus irinotecan every 2 weeks yielded a confirmed ORR of 14.0% and DCR of 80.0%, with median PFS 3.6 months and median OS 12.0 months [1]. These figures are numerically lower than the 20.5% pooled ORR from earlier meta-analyses [8] and the 25% ORR target prespecified in PURSUIT's own statistical design, though the trial population was chemorefractory with limited tumor-shrinking options [1]. The most distinctive biomarker finding was that no responses occurred among seven patients with detectable plasma RAS immediately after progression on prior anti-EGFR therapy, even when they later converted to plasma RAS-negative before rechallenge [1]. Plasma RAS positivity at cycle 3 occurred in 10.6% of patients with available samples and was associated with shorter PFS and OS [1]. At treatment discontinuation, 36.0% of patients had detectable plasma RAS, and patients who were plasma RAS-positive immediately after prior anti-EGFR therapy had higher conversion rates at cycle 3 (42.9% vs 6.3%) and at discontinuation (85.7% vs 32.3%) [10]. Exploratory NGS showed that resistance alterations beyond RAS—including EGFR ectodomain mutations, BRAF V600E, MAP2K1/2 alterations, and ERBB2/MET amplification—were associated with no responses (0/13 vs 5/23 without such alterations) [10].

How PURSUIT compares with randomized and pooled rechallenge evidence

The strongest comparative evidence comes from a systematic review and meta-analysis of three randomized phase II trials with 320 patients, which found that anti-EGFR rechallenge significantly improved DCR (OR 3.39, 95% CI 2.13–5.39), ORR (OR 5.13, 95% CI 2.30–11.41), and PFS (HR 0.674, 95% CI 0.499–0.909) versus standard of care, but detected no overall survival benefit (HR 0.895, 95% CI 0.736–1.087) [2]. This is a critical boundary: PURSUIT's median OS of 12.0 months is consistent with the meta-analysis's median OS of 9.8 months [8], but the absence of a randomized control in PURSUIT means its survival data cannot establish that rechallenge extends life compared with alternative later-line therapies. The VELO randomized phase II trial, which tested panitumumab plus trifluridine/tipiracil versus trifluridine/tipiracil alone, showed no OS difference between arms (13.1 vs 11.6 months; HR 0.96) but a significant PFS and OS benefit for subsequent anti-EGFR rechallenge in patients who received fourth-line therapy after progression on the control arm (PFS HR 0.29; OS HR 0.30) [5]. The A-REPEAT phase II study, which used panitumumab with chemotherapy based on tissue RAS testing without ctDNA selection, reported a 13% response rate and 52% DCR in 23 patients, with similar outcomes regardless of liquid biopsy RAS status—though the small sample limited interpretation [3]. The CAVE-2 trial demonstrated that liquid biopsy comprehensive genomic profiling can identify actionable alterations beyond RAS/BRAF in over half of screened patients, including high tumor mutational burden, BRAF V600E, KRAS G12C, and RET fusions, suggesting that broader profiling may refine rechallenge selection further [7]. PURSUIT's contribution is not superior efficacy but the prospective demonstration that the timing of ctDNA sampling—specifically immediately after prior anti-EGFR progression—captures persistent resistance biology that a single pre-rechallenge test misses [1].

Validation from large-scale ctDNA dynamics and prognostic context

The prognostic significance of ctDNA RAS/BRAF dynamics is supported by a multi-institutional study of 1,391 patients from the SCRUM-Japan GOZILA cohort, which classified patients into persistent RAS wild-type, acquired RAS mutant, Neo RAS wild-type, and persistent RAS mutant groups [4]. In multivariable analysis, ctDNA fraction ≥1.0% was independently associated with shorter OS (HR 1.58, 95% CI 1.31–1.91), as were BRAF mutation in tissue (HR 2.75) and RAS mutation in tissue (HR 1.31) [4]. Neo BRAF wild-type status was associated with longer OS (HR 0.37, 95% CI 0.16–0.85), suggesting that loss of detectable BRAF mutation after treatment may reflect favorable biology [4]. These findings validate the principle that ctDNA dynamics carry prognostic information, but they also highlight a limitation: the GOZILA analysis was retrospective, with heterogeneity in prior treatment history and ctDNA sampling timing, and it did not test whether acting on these dynamics improves outcomes [4]. PURSUIT's prospective design partially addresses this by linking serial ctDNA assessment to a defined treatment protocol, but the absence of a control group means the prognostic associations cannot be translated into predictive utility [1].

Where PURSUIT evidence stops: design, sample, and interpretation limits

PURSUIT is a single-arm phase II trial with 50 treated patients, designed to detect a 25% ORR against a 10% null hypothesis with 85% power [1]. The observed 14.0% ORR did not meet this prespecified threshold, though the confidence interval overlaps with clinically meaningful activity in a chemorefractory population [1]. The absence of randomization means no conclusion can be drawn about whether ctDNA-guided rechallenge is superior to standard molecular selection or to other later-line therapies [1][2]. The small sample size also limits subgroup analyses: the finding that no patients with detectable plasma RAS immediately after prior anti-EGFR progression responded is based on only seven patients, and the authors explicitly state this should not be considered an absolute exclusion criterion without further prospective validation [1]. The anti-EGFR-free interval of ≥365 days was associated with a numerically higher ORR (37.5% vs 9.5% for <365 days; P=0.037), but this was an exploratory analysis in a small subgroup [1]. The exploratory NGS analyses were hypothesis-generating and not used for eligibility, and the digital PCR platform covered only RAS alterations, not BRAF or other resistance genes [1]. The meta-analysis of randomized trials found no OS benefit for rechallenge versus standard of care [2], and PURSUIT's median OS of 12.0 months cannot overturn that finding because of its design. The clinical question that remains open is whether prospective randomized testing of a two-step ctDNA assessment strategy—sampling immediately after prior anti-EGFR progression and again before rechallenge—can improve patient selection beyond a single pre-rechallenge test, and whether such selection translates into meaningful survival gains [1][6].

About These Sources

This research page is built on 10 peer-reviewed studies — published from 2023 to 2026, 9 from 2024 or later, collectively cited 79 times — selected as the most relevant from 13 studies that passed quality screening, drawn from 83 papers retrieved from a database of over 500 million.

Sources used in this answer