Main

Surgical resection of colorectal liver metastasis (CRLM) is feasible in 10–20% of patients presenting with liver-only disease and confers a five-year survival of 35–40% (Kanas et al, 2012). Following resection disease progression is a challenge, occurring in 75% of patients (Nordlinger et al, 1996) of which half of the recurrences are in the liver (Lorenz et al, 1998; Giacchetti et al, 1999; Andre et al, 2004). Perioperative chemotherapy (CT) can reduce the risk of relapse (Nordlinger et al, 2008) as well as obviating surgery in those who progress rapidly because of poor biology, and has become standard of care in many centres. The New EPOC study sought to assess the benefit of adding cetuximab (CTX), an antibody to the epidermal growth factor receptor, to standard perioperative CT for patients with operable CRLM. Unexpectedly, the addition of CTX resulted in a significantly shorter progression-free survival (Primrose et al, 2014).

Following surgery for CRLM the most common site of first treatment failure is the liver with the lung being second (Nordlinger et al, 2008; de Jong et al, 2009). Data suggest that the pattern of recurrence is relevant to outcome with lung only recurrence possibly favouring a better prognosis (Govaert et al, 2014). In the context of significant progression-free intervals, selected patients can expect improved survival with further lung and liver resection (Luo et al, 2014). For those unsuitable for additional surgery, further systemic treatment offers good disease control, reflected in the favourable long-term survival for patients presenting with CRLM (Nordlinger et al, 2013).

We have examined the pattern of treatment failure in the New EPOC study to determine whether it can inform the unexpected study outcome. Potential differences in post-progression therapy may inform an important secondary end point such as overall survival.

Materials and methods

Study design, treatments and objectives

The New EPOC study (ISRCTN 22944367) design, eligibility criteria and treatment regimens have been described previously (Primrose et al, 2014). This was a multi-centre prospective randomised open label phase III controlled trial. The study was approved by the South West Research Ethics Committee and all patients provided written informed consent. Patients with KRAS exon 2 wild-type resectable or suboptimally resectable colorectal liver metastases were randomised 1 : 1 to receive CT with or without CTX before and after liver resection. The primary objective was to determine a difference in progression-free survival between the treatment groups. The trial was closed to recruitment on 1 November 2012 when protocol-defined futility criteria were met. The focus of the present analysis is to ascertain detailed data regarding the patients with progressive disease.

Data collection

Gender, age, site of primary disease, lymph node involvement, serum CEA level and dates of progression and death had already been obtained as part of New EPOC. For the 109 patients (CT alone group=48, CT plus CTX group=61) with progressive disease at the cut-off date for analysis of 1 November 2012, additional data regarding first site(s) of progressive disease and treatments undertaken for progressive disease were obtained using case report forms. One patient, who was previously reported as having progressed before the data cut-off, has since been confirmed as not having progressive disease. An additional 13 patients (CT alone group n=8, CT plus CTX group n=5) died but had not progressed.

Post-progression survival was defined as the time from date of progression to death. Patients still alive were censored at their date of last follow-up or 1 November 2012, whichever was earlier, so as not to release data that are at present immature but that will contribute to a final analysis of the trial in the future.

Statistical analysis

The comparison of post-progression survival between the treatment groups was done with Cox proportional-hazards regression and Kaplan–Meier curves. All analyses were done on a modified intention-to-treat basis, including all patients with known KRAS exon 2 wild-type genotype who had confirmed progressive disease at the cut-off date for analysis of 1 November 2012. Only the analyses on further lines of CT were pre-defined. SAS (version 9.2 or above, SAS Institute Inc., Cary, NC, USA) and STATA (version 11 or above, StataCorp LP, College Station, TX, USA) were used for all analyses.

Results

The clinico-pathologic features of the 110 patients with progressive disease identified before the cut-off for analysis of 1 November 2012 are shown in table one as previously described (Primrose et al, 2014). The baseline and clinico-pathologic characteristics of the patients with progressive disease were similar between the groups (Table 1).

Table 1 Characteristics of all randomised KRAS wild-type patients who progressed and/or died before 1 November 2012
Full size table

Of the 109 patients with progression events, the majority occurred after surgery (CT group 27% (13/48) vs 73% (35/48), CT and CTX group 25% (15/61) vs 75% (46/61); Table 2). Approximately half of all progression events involved the liver only (CT 60% (29/48), CTX group 49% (30/61)). When this analysis is restricted to those patients with post- rather than pre-operative progression, the same trend is apparent (CT 54% (19/35), CTX 48% (22/46)). Other single sites of progressive disease were similarly well balanced between the two groups of the trial, although a higher proportion of patients in the CTX group had multiple sites of progressive disease (CT 8%, 4/48; CTX 23%, 14/61, difference 15% 95% CI 2.0%–28.1% P=0.04).

Table 2 Site(s) of first progressive disease
Full size table

Over three quarters of patients (84/109) were reported to have undergone further treatment(s) for progressive disease (CT 75% (36/48); CTX 79% (48/61)). Of those the majority received further CT (CT 58%, 28/48; CTX 67%, 41/61), however, 31% (15/48) in the CT group and 23% (14/61) in the CTX group underwent surgical treatment of progressive disease with curative intent (Table 3). Four patients (4/48) in the CT group and six patients (6/61) patients in the CTX group received ablative therapies for progressive disease in the liver, lung or both. Radiotherapy was used to treat progressive disease in 6% (3/48) of patients in the CT group and 7% (4/61) in the CTX group. Twenty-seven percent (13/48) in the CT group and 41% (25/61) in the CTX group received CT alone as treatment for progressive disease.

Table 3 Treatment of progressive disease with curative intent
Full size table

Data were received on 69 patients (16 who had pre-operative progression and 53 who had post-operative progression) that received further CT for progressive disease (Table 4). The majority of second-line therapy given was irinotecan-based without any additional biological therapy. Twenty-two were given CTX as a further line treatment: 11 following CT and 11 following CTX. Four patients received CTX as a second line treatment for progressive disease having had CT plus CTX as part of the trial. Anti-vascular endothelial growth factor therapy was given as a treatment for progressive disease to 26 patients: 7 in the CT group and 19 in the CTX group.

Table 4 Further lines of chemotherapy for progressive disease
Full size table

Median follow-up post progression was 14.5 months (95% CI 5.4–19.6) in the CT alone group and 14.2 months (95% CI 6.3–19.5) in the CT plus CTX group. The median post-progression survival was 18.7 months (95% CI 13.1-not reached; Figure 1) in the CT group compared with 15.9 months (95% CI 8.6–26.1) in the CTX group (HR 1.69, 95% CI 0.86–3.34 P=0.130). When this analysis was restricted to those patients that underwent a R0 resection of the liver as part of the trial (29 in the CT group, 34 in the CTX group), the median post-progression survival was 29.0 months (95% CI 16.5-not reached) in the CT group compared with 19.9 months (95%CI 13.2-not reached) in the CTX group (HR 2.07 95% CI 0.70–6.09 P=0.187).

Figure 1
figure 1

Kaplan–Meier curve of post-progression survival by treatment group.

Full size image

A total of 38 patients have had surgery or ablation with curative intent for progressive disease. The post-progression survival of these individuals was compared with those who did not receive surgery or ablative therapy with curative intent for progressive disease. Median follow-up post progression was 12.8 months (95% CI 5.4–19.2) in the surgery/ablation with curative intent for progressive disease group and 14.5 months (95% CI 6.0–24.0) in the group with progressive disease not treated with curative intent. The median post-progression survival was not reached in the surgically/ablation treated group (lower limit of 95% CI 26.1 months; Figure 2) compared with 14.3 months (95% CI 8.4–17.6) in the group not treated with curative intent for progressive disease (HR 17.42, 95% CI 4.14–73.25, P<0.001).

Figure 2
figure 2

Kaplan–Meier curve of post-progression survival according to whether treatment of progressive disease was undertaken with curative or palliative intent.

Full size image

Discussion

This analysis demonstrates that the pattern of progression and subsequent treatment for patients do not differ substantially between the arms of the New EPOC study. There is one notable difference, a significantly higher proportion of multi-site recurrences in the CTX group, although the numbers are modest. In both arms of the study, the liver was the most frequent first site of failure. This pattern of disease progression is much the same as that observed in other studies including the original EPOC study (Nordlinger et al, 2008; de Jong et al, 2009).

The incidence of progressive disease in the liver was comparable between the two groups of the trial. This refutes the contention that inadequate surgery or differential use of ablation was responsible for the primary study outcome (Nordlinger et al, 2015; Primrose et al, 2015). It also allays concerns regarding the possibility of differential imaging being performed before trial entry. All patients were managed in high-volume liver centres and while the trial protocol only mandated computed tomography, the majority will have had additional imaging performed. As the randomisation is stratified by centre (hence MDT) bias is not possible. Indeed the large majority of progression events were post surgery and the greater proportion of multi-site progressions in the CT plus CTX group is more in keeping with a failure of systemic disease control.

The majority of patients with disease progression received additional off-trial treatments. The choice of further systemic therapy is as expected for such a cohort and reasonably similar in both groups. A number of patients received antibody therapies in the second or subsequent lines of treatment which is in keeping with their availability under the Cancer Drug Fund (introduced in the UK in 2011). Those treated with CT alone as part of the trial often received CTX post progression and indeed some that received it on trial also received it following progression. This pattern of post-progression treatment is consistent with standard practice in the UK.

Progression-free survival is widely accepted as a surrogate for overall survival in clinical trials of colorectal cancer (Giessen et al, 2013). It is however important to distinguish between curative and palliative intent. For liver only progression, one-third of those that progressed during the neoadjuvant phase of treatment went onto have a potentially curative liver resection and a similar proportion that progressed postoperatively underwent a further liver resection. These data are in line with a published series suggesting that approximately one-third of recurrences in the liver can be offered repeat hepatectomy (Jones et al, 2012).

Overall this analysis demonstrates that a quarter of patients went onto have further surgery with curative intent, predominantly lung and/or liver resections. This was balanced between the groups. Repeat liver surgery has been shown to be associated with favourable survival outcomes (Petrowsky et al, 2002; Lam et al, 2013; Luo et al, 2014), and indeed the post-progression survival for the cohort treated with curative intent looks most favourable, although it is noteworthy that the period of follow-up available is presently short. The relatively short follow-up (consequent in part on premature study cessation) restricts the ability of these data to inform whether the detrimental effect of CTX persists beyond relapse and a final analysis will obviate these issues.

In summary, the present study demonstrates that the pattern of failure associated with the addition of CTX to surgery and routine CT for operable colorectal liver metastases is as expected for such a cohort. The slightly higher proportion of multi-site recurrence in the CT and CTX group is the only notable difference. The pattern of disease progression seen is strongly suggestive of a systemic failure consistent with micrometastatic disease rather than local failure of disease control at the site of resection.