Hepatectomy is the only potentially curative treatment for colorectal liver metastases (CRLM); however, only 15 to 20 percent of patients with liver-only metastases are eligible for resection. Of those patients who undergo resection, 70 to 80 percent ultimately develop recurrence.1 Approximately 50 percent of patients experience recurrence locally in the residual liver, thought to be secondary to microscopic residual disease.2 The optimal treatment strategy that achieves locoregional control, improves overall survival, and preserves quality of life remains controversial. The inclusion and sequence of chemotherapy for the treatment of liver-only metastatic disease often is disputed and remains a source of debate in the surgical oncology community.
Proponents of chemotherapy contend that this form of treatment offers the potential to eliminate residual micrometastatic disease in this cohort of patients.1 However, previous studies have yet to convincingly demonstrated this advantage. Evidence to date has suggested that perioperative and posthepatectomy chemotherapy strategies may improve disease-free survival but do not affect overall survival (OS).2
Only one randomized trial, European Organisation for Research and Treatment of Cancer (EORTC) 40983, has investigated the role of perioperative chemotherapy in patients with resectable CRLM. In this study, patients were randomized to receive either perioperative chemotherapy and surgery or surgery alone. Patients in the perioperative chemotherapy arm experienced improved three-year progression-free survival (36.2 percent versus 28.1 percent), albeit with a higher rate of perioperative complications compared with patients in the surgery-only study arm (25 percent versus 16 percent).2 No significant difference in OS was observed between cohorts, although the trial was not designed to detect a difference in OS. Until recently, EORTC 40983 has provided the only prospective, randomized data in this setting; however, because of the conflicting results, oncologists and surgeons continue to debate the best treatment strategy.
JCOG 0603 randomized clinical trial
A recently reported phase III trial (Japan Clinical Oncology Group [JCOG] 0603) in Japan sought to investigate the efficacy of posthepatectomy adjuvant 5-FU, oxaliplatin, and leucovorin (FOLFOX) compared with hepatectomy alone for patients with resectable CRLM. A total of 300 patients were randomized to surgery (n = 149) or surgery followed by adjuvant chemotherapy (n = 151). The study was powered to detect a difference in disease-free survival (DFS) of 12 percent between the two study arms; secondary outcomes included OS and adverse event rates.1 Of those enrolled, most patients had up to three metastases, and approximately 56 percent presented with synchronous liver-only metastases. At the time of randomization, 73 percent of patients in the hepatectomy-only cohort and 81 percent in the posthepatectomy chemotherapy group were treatment-naïve.
Five-year DFS was significantly higher in patients treated with adjuvant chemotherapy compared with surgery alone (50.1 percent versus 37.3 percent; hazard ratio [HR] 0.63, 95 percent confidence interval [CI] 0.45 to 0.89). Because the trial met the prespecified criteria for superiority during interim analysis, the study was terminated early. Subgroup analysis demonstrated an advantage in DFS when posthepatectomy chemotherapy was administered to patients with synchronous and limited (one to three) metastases, left-sided primary tumors, and who were treatment-naïve.1 However, no difference in five-year OS was observed between treatment groups (hepatectomy only: 83.1 percent versus hepatectomy and adjuvant chemotherapy: 71.2 percent, p = 0.42). Moreover, no clinical characteristics were identified to favor hepatectomy with or without chemotherapy with respect to OS.1
Optimizing outcomes by quantifying minimal residual disease
Although these trials have demonstrated improved DFS with perioperative or adjuvant chemotherapy, the absence of an OS benefit continues to cast doubt on the long-term benefit of systemic therapy for resectable CRLM. Some experts suggest that waiting until patients develop recurrent disease may be a more favorable approach, as it may minimize cumulative toxicity without compromising survival. Others suggest that better patient selection is necessary to prevent recurrence in high-risk patients, while avoiding unnecessary treatment in patients with a more favorable prognosis.
Accordingly, enthusiasm is growing for the potential use of posthepatectomy circulating tumor DNA (ctDNA) to stratify which patients may benefit from posthepatectomy chemotherapy and as a dynamic biomarker for treatment response. This biomarker-directed strategy could identify patients with minimal residual disease (MRD), defined by the presence of posthepatectomy ctDNA positivity, and could allow clinicians to tailor the use of adjuvant chemotherapy based on the presence or absence of MRD.
To date, one completed phase II randomized clinical trial has evaluated the role of ctDNA analysis in the neoadjuvant setting. Partenariat de Recherche en Oncologie Digestive (PRODIGE)-14 randomized patients with potentially resectable CRLM to first-line triplet (FOLFIRINOX) or doublet chemotherapy (5-FU, FOLFOX, or 5-FU, leucovorin, irinotecan [FOLFIRI]) combined with targeted therapy. At baseline, 91 percent of patients with KRAS (Kirsten rat sarcoma viral oncogene homolog) mutant tumors had detectable ctDNA, which decreased to 63 percent at four weeks, and 19 percent before surgery. Patients with persistently detectable ctDNA levels after four weeks of therapy had a lower margin negative resection compared with patients with no ctDNA detected (36 percent versus 85 percent, p = 0.01).3 Furthermore, among patients who underwent margin negative resection, those with detectable ctDNA levels before surgery had a worse overall survival, suggesting that the presence of detectable ctDNA after four weeks of systemic chemotherapy or at completion of preoperative therapy could ultimately play a role in patient selection for resection of CRLM.
Loupakis and colleagues have described the utility of ctDNA in the adjuvant setting of metastectomy. The authors report a 54 percent MRD rate among patients who underwent curative-intent resection of liver (58 percent), lung (20 percent), peritoneum (14 percent), and other (8 percent) metastases.4 MRD was associated with inferior OS and was the most significant factor associated with DFS (HR 5.78; 95 percent CI 3.3 to 10.0). In addition, patients who were MRD-negative and who did not receive chemotherapy experienced OS of 100 percent, demonstrating that complete clearance of disease with surgery may not necessitate additional chemotherapy.4
Another study of 54 patients with resectable CRLM demonstrated detectable ctDNA in 85 percent of patients before treatment and 24 percent of patients following resection.5 Patients with detectable postoperative ctDNA had lower RFS and OS compared with patients without detectable ctDNA. Of note, end-of-treatment ctDNA detection was associated with five-year RFS of 0 percent versus 76 percent for patients with an undetectable ctDNA, regardless of whether adjuvant chemotherapy was given. Additional studies report posthepatectomy MRD of 44 percent to 67 percent and have likewise demonstrated an association of MRD with RFS, illustrating a promising role of ctDNA in interpreting the dynamic response to therapy, informing prognosis, and potentially guiding adjuvant therapy decisions.6,7
Additional investigation of ctDNA in the clinical trial setting will be important to define the future of individualized treatment strategies, an approach that could improve survival for patients at high risk of recurrence, while minimizing toxicity in patients unlikely to benefit from further therapy. A nonrandomized trial led by MD Anderson Cancer Center has recently opened and will stratify patients to adjuvant chemotherapy based on ctDNA risk category following curative-intent resection of CRLM after preoperative chemotherapy. High-risk patients (ctDNA positive) posthepatectomy will receive multiagent chemotherapy (FOLFOX or FOLFIRI), whereas low-risk patients (ctDNA negative) will receive single-agent 5-FU or Xeloda or surveillance per provider judgment, with a primary endpoint of one-year recurrence-free survival. The trial design allows for crossover depending on dynamic changes in ctDNA during survivorship. The results of this and similar trials are highly anticipated and will likely shape the future of personalized, risk-stratified treatment in patients with CRLM.
