Introduction

Colorectal cancer (CRC) is a common cancer, which often metastasises to the liver. Approximately 20–30 % of patients will have liver metastases at the time of diagnosis, and subsequently, up to 50 % of patients with CRC will develop colorectal liver metastases (CRLM) [1]. For CRLM, liver resection has become a potentially curative strategy, in which the safety has been improved by the use of preoperative portal vein embolization to enhance the hypertrophy of the future liver remnant, keeping a low central venous pressure to decrease bleeding, and intraoperative ultrasound to define the location of intra-parenchymal tumours and vascular structures. However, only about 20 % of patients present with metastases confined to the liver that are initially resectable [2], and strategies to downsize and downstage the metastases are usually required to facilitate their removal. Successful surgical R0 resection confers a 5-year overall survival (OS) of 35–40 % and a 10-year OS rate of 25 % [3, 4], but further disease progression is likely to occur in 75 % of patients [5], mostly in the liver or lungs, within the 2 years of surgery, raising the question of whether adjuvant therapy should be used in this setting [6]. Several early studies have suggested a benefit from this approach [7–9].

Even with an R1 resection, the effectiveness of current chemotherapy may achieve similar outcomes [10•]. This survival rate after liver resection compares with a 75 % 5-year OS rate in patients with stage III disease following removal of the primary tumour following adjuvant chemotherapy and <10 % in unselected patients with stage IV disease treated with chemotherapy alone [11, 12].

However, the majority of patients with CRLM have a disease which is too advanced to allow initial surgical treatment or are assessed as having borderline resectable lesions. Both technical resectability (and maintaining at least a 30 % future liver remnant) and prognostic aspects (whether resection is in the interest of the patient) should be considered [13], and hence, there is no consensus regarding resectability. ESMO guidelines [14] list both technical and oncological contraindications to hepatic resection in patients with CLRM.

Neoadjuvant chemotherapy (NACT) has a long established role in downstaging initially borderline or unresectable liver metastases [15], but remains controversial in easily resectable tumours/single metastases. Response is a prognostic factor, as tumour progression during neoadjuvant chemotherapy (NACT) is associated with a worse outcome, even when surgical resection can still be performed with a curative intent.

In colorectal cancer, chemotherapy strategies and drug combinations are usually considered in two separate categories either as ‘adjuvant’ following resection of the primary—or in contrast as ‘advanced/metastatic’ (with or without the primary in situ). For metastatic disease, standard chemotherapy agents (5-fluorouracil, capecitabine, oxaliplatin, irinotecan) have been supplemented by targeted therapies directed against the epidermal growth factor receptor (EGFR) (cetuximab, panitumumab) or angiogenesis (bevacizumab, aflibercept, regorafenib). Currently, there is no proven role for the addition of biological agents in the postoperative adjuvant setting.

The treatment of liver metastases lies uneasily between these two settings (adjuvant and advanced/metastatic) and raises specific questions which remain to be answered regarding the optimal chemotherapy schedules, the duration of chemotherapy and the role of biological agents both prior to and following resection. Chemotherapy for CRLM can be administered before (neoadjuvant or induction), after (postoperative adjuvant) or before and after (perioperative) surgical resection. Trials in CRLM have been difficult to perform and generally underpowered because of the paucity of patients randomised.

For easily resectable CRLM, there is no robust evidence in favour of either adjuvant or neoadjuvant systemic chemotherapy in addition to surgery. The EORTC trial (EPOC trial) in patients with resectable liver-limited CRC randomly assigned to treatment with surgery alone or perioperative chemotherapy (neoadjuvant and postoperative) with FOLFOX—infusional fluorouracil, leucovorin and oxaliplatin. Eligible patients had a relatively favourable prognosis as only a single metastasis was present in 51 % of patients, the majority of whom had metachronous metastases. The risk of subsequent relapse after surgery in patients randomly assigned to receive perioperative FOLFOX was reduced by 25 % [16], but improvement in OS was not observed. In contrast, the ‘new EPOC’ trial reported inferior outcomes for KRAS WT patients with resectable and borderline resectable liver metastases who received cetuximab in combination with oxaliplatin-based chemotherapy when compared with chemotherapy alone [17••].

For borderline or unresectable metastatic disease, chemotherapy is started upfront. Response will be observed in about 50 % of patients, but the duration of response is usually short and tumours tend to develop resistance within a few months.

There are also questions regarding the optimal sequence of surgery for primary and liver disease and the role of radiotherapy in rectal cancer. A number of different investigational strategies are tabulated in Table 1.

Table 1 Chemotherapy/biological agent strategies currently being integrated to deliver systemic doses of chemotherapy in the treatment of CRLM
Full size table

Uptake of further postoperative adjuvant chemotherapy after liver resection is generally poor. Hence, neoadjuvant chemotherapy (NACT) has significant potential advantages, which include better compliance, compared to chemotherapy given in the adjuvant setting, and offers the possibility of measuring early in vivo response to systemic treatment (see Table 2). For resectable patients, the prime objective of NACT was to provide a time interval before surgery for assessment of the tumour biology, to treat potentially occult disease and to avoid surgery in those patients with rapidly progressive disease as a result of primary resistance to chemotherapy. A second objective in these resectable patients was to achieve cytoreduction both to limit the extent of liver resection and potentially postoperative morbidity and to facilitate a margin-free R0 liver resection.

Table 2 Advantages and disadvantages of preoperative/neoadjuvant chemotherapy in CRLM
Full size table

Currently, systemic chemotherapy in combination with liver resection for CRLM is accepted as the standard of care: the National Comprehensive Cancer Network (NCCN) [18] recommends 6 months of perioperative systemic chemotherapy, and in the UK, the National Institute for Health and Care Excellence (NICE) [19] recommends considering the use of systemic chemotherapy prior to liver resection. In borderline or unresectable cases, standard chemotherapy agents in combination with biological agents appear to have increased the resection rate of these patients. ESMO guidelines [14] recommend that patients with clearly resectable disease and favourable prognostic criteria can be treated with upfront resection and perioperative treatment may not be necessary. In contrast, in patients with technically resectable disease where the prognosis is unclear or unfavourable, perioperative combination chemotherapy is recommended. ESMO guidelines also acknowledge that in resected patients with favourable oncological and technical (surgical) criteria, who have not received perioperative chemotherapy, there is no strong evidence to support the use of adjuvant chemotherapy. Hence, there is consensus that combination chemotherapy should be part of such neoadjuvant regimens—usually oxaliplatin and a fluoropyrimidine, but there is no consensus regarding the timing, nor the selection of targeted therapy in this regard—nor whether the targeted therapy, if administered prior to resection, should continue following successful resection.

The aim of this review was to examine past and ongoing trials employing systemic (rather than intra-hepatic) chemotherapy strategies—neoadjuvant, perioperative or adjuvant—for resectable or borderline resectable liver metastases.

Methods

A systematic review of available literature was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [20]. The review criteria included randomised phase II or phase III controlled trials reporting on the outcomes of overall survival (OS), progression-free survival (PFS)/disease-free survival (DFS) and grade 3–4 complications in patients with resectable CRLM. Potentially relevant trials were selected by a search using Medical Subject Headings (MeSH) terms and specific text-words terms: colorectal cancer liver metastases, neoadjuvant therapy, neoadjuvant chemotherapy, adjuvant chemotherapy and perioperative chemotherapy. Relevant studies were identified by reviewing the titles and abstracts.

Current trials were identified using ClinicalTrials.gov to include currently recruiting and recently terminated trials investigating treatments for metastatic colorectal cancer in the neoadjuvant and adjuvant setting. The following terms were used as in the search criteria: neoadjuvant, adjuvant, metastatic, colorectal and cancer.

Results

We found only 4 randomised phase III controlled trials with a total of 1098 patients and 14 ongoing randomised phase II and prospective phase III or IV trials. The main details of the published randomised trials are tabulated in Table 1.

Ongoing or Planned Trials

  1. 1.

    CHARISMA: Neo-adjuvant chemotherapy followed by surgery versus surgery alone in high-risk patients with resectable colorectal liver metastases: the CHARISMA randomized multicenter clinical trial

This multicentre phase III randomised control trial aims to evaluate the impact of neoadjuvant chemotherapy in high-risk patients (Fong’s clinical risk score 3–5) with resectable colorectal liver metastases, without extrahepatic disease. Such high-risk patients will be randomised to receive surgery alone versus neoadjuvant oxaliplatin-based chemotherapy prior to surgery. The primary study endpoint is OS. Secondary endpoints are progression-free survival (PFS), quality of life, morbidity of resection, treatment response on neoadjuvant chemotherapy and whether CEA levels can predict treatment response.

  1. 2.

    NCT02510378—Short Course Radiotherapy Combined With Chemotherapy in Stage IV Rectal Cancer With Resectable Liver Metastases

In this phase II trial, patients with rectal cancer and resectable liver metastases are treated neoadjuvantly with short-course radiotherapy 25 Gy in five fractions to the pelvis followed by at least four cycles of consolidation XELOX chemotherapy. Patients will be evaluated after neoadjuvant therapy and those with resectable rectal cancer and liver disease will undergo surgery. Those patients with unresectable lesions will receive chemotherapy. The primary outcome measure is R0 resection rate. The secondary endpoint is radiotherapy toxicity rate.

  1. 3.

    NCT01923987—Short Course Radiotherapy and Biochemotherapy With Delayed Surgery for Rectal Cancer With Synchronous Distant Metastasis

Primary rectal cancer presenting with synchronous metastastatic disease is often a locally advanced disease and needs downsizing before surgery. In such patients, it is reported that pelvic recurrence rates and distant metastasis rates outside the liver are 30–35 and 60 %, respectively. Therefore, combined treatment with radiotherapy and chemotherapy is used. However, the sequence of treatment modalities is not yet definitely established and preoperative chemoradiotherapy and surgical resection is accepted as an option of treatment. Conventional long-course chemoradiotherapy delays administration of full-dose chemotherapy, and metastatic lesions could progress during chemoradiotherapy. In this multicentre phase II study, the efficacy of short-course radiotherapy followed by full-dose chemotherapy with delayed surgical resection of the primary tumour and metastases was investigated. The primary outcome measure is R0 resection rate of primary and metastatic lesions. Secondary endpoints include OS and PFS rates, tumour regression grade and toxicity.

  1. 4.

    NCT01722903—Detection of CTCs in Patients Undergoing Surgery for Stage IV Colorectal Cancer

CTCs are believed to cause metastasis and may provide a non-invasive alternative to organ biopsies for the detection, characterisation and monitoring of solid cancers. CTC numbers have been shown to be a strong predictor of progression-free survival and overall survival for mCRC patients. The CellSearch system is currently the only FDA-approved test for the evaluation of CTC numbers in metastatic colorectal cancer. In this protocol, the CellSearch system will be compared to a new technology, called the Flexible Micro Spring Array (FMSA) device.

In this prospective observational study, CTCs will be harvested from patients undergoing liver or lung metastatectomy for colorectal cancer. The primary outcome measure will be to ascertain the quantity of CTCs isolated during liver and/or lung metastatectomy. The secondary outcome measure will be 3-year overall survival rate.

  1. 5.

    NCT00630045—Phase 3 Study of Surgery Combined With Neoadjuvant Chemotherapy(XELOX) in Colorectal Cancer With Resectable Liver Metastasis

In this randomised phase III study, patients with colorectal cancer and resectable liver metastases patients were randomised to receive two to three cycles of neoadjuvant Xelox chemotherapy versus no neoadjuvant chemotherapy prior to resection of liver metastases. The primary aim of the study is to establish if neoadjuvant chemotherapy improves 3-year disease-free survival rate in this setting. Secondary endpoints include R0 resection rate and 5-year overall survival rate.

  1. 6.

    NCT01505166—Randomized Phase II Adjuvant Chemotherapy ± FANG™ in Colorectal Carcinoma With Liver Metastases (FANG-CLM)

In this randomised phase II study, patients with colorectal carcinoma with either synchronous or metachronous liver metastases will receive either sandwich/adjuvant chemotherapy and an intradermal autologous Vigil™ cancer vaccine or sandwich/adjuvant chemotherapy and placebo following resection ± ablation of primary tumour and liver metastases with curative intent. The primary aim of the study is to investigate if overall survival rate is increased with the addition of the Vigil cancer vaccine to chemotherapy in patients with colorectal cancer and liver metastases being treated with curative intent.

  1. 7.

    NCT00070265—Neoadjuvant and Adjuvant Capecitabine and Oxaliplatin in Treating Patients With Resectable Liver Metastases Secondary to Colorectal Cancer

This phase II trial is studying the efficacy of capecitabine and oxaliplatin when given in combination before and after surgery in patients with resectable CRLMs. The primary outcome measure is rate of R0 resection. Secondary outcome measures include response rate, improvement in survival associated with downstaging based on metastatic colorectal prognostic score, disease-free and overall survival.

  1. 8.

    NCT00264979—Evaluation of 2 Resection Strategies of Synchronous Colorectal Cancer Metastases (METASYNC)

The surgical strategy for the treatment of synchronous colorectal cancer liver metastases has still not been defined. The purpose of this study is to compare two treatment strategies for liver resection in this setting. In the first arm, liver metastases are resected at the same time as the primary resection. In the second arm, liver metastases are resected 12–14 weeks after the primary resection. The primary endpoint is the rate of patients with at least one postoperative severe complication within 60 days after each surgery. Secondary endpoints include the rate of recurrence and survival.

  1. 9.

    NCT01269229—A Trial of Neoadjuvant FOLFOX6 With Short Course Radiotherapy in Patients With Unresectable Rectal Cancer and Liver Metastasis

The purpose of this phase II study is to investigate if neoadjuvant FOLFOX chemotherapy in combination with SCRT improves resection rates in patients with unresectable rectal cancer and liver metastases. The primary endpoint is R0 resection of rectal and liver lesions.

  1. 10.

    NCT01762813—ACROBATICC Study-Assessment of clinically related outcomes and biomarker analysis for translational integration in colorectal cancer: study protocol for a population-based, consecutive cohort of surgically treated colorectal cancers and resected colorectal liver metastasis

This is an observational study exploring prognostic and predictive biomarkers in a population-based, consecutive cohort of patients with surgically resected colorectal cancer and CRLMs. Long-term outcomes assessed will be cancer-specific survival, recurrence-free survival and overall survival at 5 years. The study incorporates the analysis of circulating tumour cells and novel biomarkers such as immune cells and microRNAs. The project aims to generate results that can help better discern prognostic groups in stage II/III cancers, explore prognostic and predictive biomarkers and help detail the biology of CRLM for better patient selection and tailored treatment.

  1. 11.

    NCT01564810—Cetuximab in combination with chemotherapy for the treatment of metastatic colorectal cancer

This phase IV study aims to assess the effect of cetuximab in combination with chemotherapy in the treatment of unresectable metastatic CRC. Patients are eligible for inclusion if they are KRAS wild type with synchronous liver-confined metastases deemed non-resectable and if their primary tumour has been resected. Patients are randomly assigned to chemotherapy plus cetuximab or chemotherapy alone. The primary endpoint is the conversion rate to radical resection for liver metastases. Secondary endpoints include PFS, OS and tumour response rate.

  1. 12.

    NCT01632722—A Randomized Phase II Study of Perioperative Chemotherapy Plus Bevacizumab Versus Postoperative Chemotherapy Plus Bevacizumab in Patients With Upfront Resectable Hepatic Colrectal Metastases (APPROACH)

This phase II study aims to compare the effectiveness of combination chemotherapy plus bevacizumab in the perioperative versus postoperative setting in patients with resectable CRLM. The primary endpoint is 2-year recurrence-free survival.

  1. 13.

    NCT01972490—Study of Avastin in combination with chemotherapy for the first treatment of metastatic CRC

This phase IV study of patients with CRC with synchronous unresectable liver metastases aims to assess if the addition of avastin to chemotherapy could improve the resection rate of liver metastases in patients with RAS mutant-type, unresectable colorectal liver-limited metastases compared with chemotherapy alone. The primary endpoint is the rate of patients converted to recection for liver metastases. Secondary endpoints include PFS, OS and response rates.

  1. 14.

    NCT02162563—Treatment Strategies in Colorectal Cancer Patients With Initially Unresectable Liver-only Metastases: CAIRO5 a Randomised Phase 3 Study of the Dutch Colorectal Cancer Group (DCCG)

In this study, CRC patients with initially unresectable liver-only metastases will be tested for RAS and BRAF tumour mutation status. Patients with RAS mutant tumours will be randomised between doublet chemotherapy (FOLFOX or FOLFIRI) plus bevacizumab (schedule 1) and triple chemotherapy (FOLFOXIRI) plus bevacizumab (schedule 2). Patients with RAS wild-type tumours will be randomised between doublet chemotherapy (FOLFOX or FOLFIRI) plus either bevacizumab (schedule 1) or panitumumab (schedule 3). The primary endpoint is PFS. Secondary endpoints include R0/1 resection rates, mOS, response rate, toxicity, pathological complete response rate and postoperative morbidity.

Discussion

Curative liver resection is a well-established therapy for CRLM in both the synchronous and metachronous settings and can impact on OS. It has been shown that resection rates correlate with response rates to chemotherapy and are higher for selected rather than unselected patients with CRLM [21]. Outcomes and the risk of recurrence after resection depend on factors such as the site of the primary (rectum or colon), the number and size of the liver metastases [22], the time from primary tumour treatment to hepatic metastases, the involvement of lymph nodes, the preoperative carcinoembryonic antigen level, any extrahepatic spread and a non-radical resection [23, 24], which probably reflects the innate tumour biology rather than poor surgical technique [25]. Hence, does response to chemotherapy simply identify patients who have a pre-determined favourable prognosis? Or can the response modify the course of the disease? Earlier trials were more selective with fewer liver metastases and hence better prognosis with less intensive and less rigorous follow-up.

For patients with widespread metastatic colorectal cancer, doublet chemotherapy regimens incorporating 5FU in combination with either irinotecan or oxaliplatin offer higher response rates than 5FU alone [26, 27]. Response can be even higher if biological agents are added. For this reason, vascular endothelial growth factor (VEGF) inhibitors and epidermal growth factor receptor (EGFR) blockers, when combined with cytotoxic chemotherapy agents, are considered currently the standard of care for those patients with appropriate molecular markers [14, 18]. But these biological agents have not offered added benefit in the adjuvant setting following resection of the primary colon cancer when combined with standard adjuvant chemotherapy schedules such as FOLFOX or 5FU/capecitabine. Nevertheless, retrospective subset analysis from first-line trials of cetuximab and panitumumab in combination with chemotherapy has drawn the conclusion that as a consequence of the responses, improved curative resection rates have been reported compared to chemotherapy alone [28, 29].

Some retrospective subset analyses from trials of bevacizumab in combination with chemotherapy have also reported improved curative resection rates compared to chemotherapy alone [30]. Intriguingly, bevacizumab added to standard cytotoxic chemotherapy may also be associated with higher complete pathological response rates in patients undergoing resection of CRLM [31] and can be safely administered without increasing postsurgical complications [32–34].

The use of triple chemotherapy regimens incorporating 5FU, oxaliplatin and irinotecan (FOLFOXIRI) also increases response rates, and retrospective analyses of prospective data suggest R0 resections may also be increased with triplet chemotherapy compared to doublet regimens [35–37]. In the METHEP randomised phase II trial, FOLFIRINOX showed high response rate and the best conversion rate in CRLM compared with standard chemotherapy [38•].

The addition of biological agents to FOLFOXIRI has also been investigated, leading to high response rates from panitumumab [39]. The combination of bevacizumab with FOLFOXIRI chemotherapy also significantly improved response rates in the TRIBE study compared to bevacizumab plus FOLFIRI (65 versus 53 %, respectively); however, the rate of successful secondary resections of liver and other metastases was similar in both arms [40].

In contrast, quadruple therapy, where both an EGFR inhibitor and a VEGF inhibitor are given in combination with doublet chemotherapy, has also been investigated, but has produced significantly inferior PFS and OS [41, 42].

For colon or rectal cancer, there are no proven differences in indications for chemotherapy or targeted EGFR inhibitors or antiangiogenic drugs [43, 44], although there are ongoing arguments regarding a differential response to biological agents depending on left- or right-sided tumours [45, 46]. The approach for rectal cancer and CRLM has also been complicated by the potential and debated requirement for radiotherapy/chemoradiotherapy.

NACT is a potentially useful therapeutic approach for locally advanced operable, primarily unresectable or borderline resectable cancers of many different primary origins. Although there remains considerable variability in treatment decisions on what qualifies as resectable, improved surgical techniques for liver resection in combination with downsizing CRLM with chemotherapy, interventions to induce liver hypertrophy before resection (portal vein embolization) and the use of locally destructive techniques (radiofrequency ablation) reduce the tumour and allow an increase of normal liver volume thereby increasing the number of patients who are candidates for liver resection. In addition, the cytotoxic activity of chemotherapy treatment can be supplemented with various interventional procedures as an addition or alternative, which include transarterial chemoembolization (TACE) (using standard chemotherapy or irinotecan-loaded drug-eluting beads), transarterial embolization (TAE), microwave ablation (MWA), radiofrequency ablation (RFA), cryotherapy and selective internal radiation therapy (SIRT).

Recent trials in metastatic disease have shown patients with Ras-wild-type tumours survive 30–36 months [47–49] with chemotherapy and biological agents. Yet, outcomes in CRLM depend on performance status, age and sex as well as the RAS, BRAF and MSI profiles. In addition, left- and right-sided tumours may exhibit innately different outcomes and different responses to targeted agents [45, 46, 50].

NACT in patients with resectable liver metastases may destroy micrometastatic disease and hence reduce the risk of subsequent liver and other distant failure. Several authors have reviewed the evidence [51–54]. The use of neoadjuvant chemotherapy has significant potential advantages including better compliance, compared to chemotherapy given in the adjuvant setting offering the possibility of measuring early in vivo response to systemic treatment. The EORTC trial (EPOC trial) in patients with resectable liver-limited CRC randomly assigned to treatment with surgery alone or perioperative chemotherapy (neoadjuvant and postoperative) with FOLFOX—infusional fluorouracil, leucovorin and oxaliplatin. The risk of subsequent relapse after surgery in patients randomly assigned to receive perioperative FOLFOX was reduced by 25 % [16], but improvement in OS has not been observed. However, there is little evidence that NACT and the chance of non-response can actually lead to non-resectable metastases [16, 55]. Postoperative complications were significantly more frequent after NACT in 40/159 (25 %) compared with surgery alone in 27/170 (16 %), but these problems did not impact on PFS [16]. However, these complications may depend both on the number of cycles of chemotherapy [56], the intensity of the chemotherapy regimen and the interval left between the end of chemotherapy and surgery. However, more ambitious resections such as major hepatectomy, defined as resection of four or more liver segments [57] and extended hemihepatectomy (i.e. right or left trisectionectomy), are now routinely performed. Yet, the incidence of complications correlates with the extent of liver surgery and the duration of chemotherapy [58–61]. Long-term intensive chemotherapy is therefore a two-edged sword for the surgeon.

Hence, for easily resectable colorectal liver metastases (CRLM), there is no robust evidence in favour of either adjuvant or neoadjuvant systemic chemotherapy in addition to surgery. Biological agents, such as vascular endothelial growth factor inhibitors and epidermal growth factor receptor blockers, when combined with cytotoxic chemotherapy agents are considered currently the standard of care for those patients with appropriate molecular markers in the treatment of metastatic colorectal cancer (CRC) [14, 18], but have not shown benefit in the adjuvant setting following resection in primary colon cancer when combined with standard adjuvant chemotherapy schedules such as FOLFOX or 5FU/capecitabine. However, many studies are underpowered to show a significant difference in overall survival. Although adjuvant chemotherapy appears of benefit in the adjuvant setting in stage III colon cancer [62–64], the MOSAIC trial [63] required more than 2000 patients with colon cancer to show an improvement in both DFS and OS (and most evident in high-risk pN2 patients).

Initial tumour response to chemotherapy is associated with improved disease-free survival after resection of CRLM, but for many patients, this RECIST response is transient and further disease progression occurs rapidly after stopping chemotherapy. In the randomised trials specifically investigating treatment-free intervals versus maintenance chemotherapy (COIN, AIO 0207 and CAIRO3), the median time from the end of chemotherapy treatment to progression without chemotherapy was 3.0–4.1 months [65–67]. This rebound effect may be more pronounced with the use of combined cytotoxic and biological agents to downstage the CRLM [17••]. Ultimately, considerable time is required for notification of completion of chemotherapy, any MDT discussion, clinic appointments and assessment and preparation for surgery. Some retrospective studies have shown the rate of growth of CRLM in this interval following chemotherapy is rapid (2.3 % per week), with a calculated tumour doubling time (DT) of 46 days [68]—as opposed to a DT of untreated CRLM between 63 and 112 days [69, 70]. It remains uncertain whether this rapid tumour regrowth can influence disease-free survival in resectable CRLM after resection or whether outcomes are simply determined by the initial response to chemotherapy.

Decrease in size of these lesions may indicate a distant effect in terms of dealing with micrometastatic disease, whereas the rebound growth of macroscopic hepatic lesions after stopping chemotherapy may not be associated with the recovery of these smaller peripheral micrometastases. However, if chemotherapy is used to downsize CRLM of borderline resectability, this rapid rebound after stopping chemotherapy may lead to the lesions becoming unresectable again, and liver resection at a sooner time point may be desirable in this group. Retrospective analyses suggest that selection of patients by clinical prognostic characteristics such as clinical risk score (CRS) (Fong score) may define a patient population expected to benefit more or less from chemotherapy [71]. Many previous studies of perioperative chemotherapy combined with liver surgery often excluded patients with a high CRS—who would have the highest risk of recurrence [72]. For instance, the EORTC trial included 51 % of patients with only a single liver metastasis, and >25 % of patients had only two metastases. The Liver Met Survey database examined patients with solitary, metachronous, primarily resectable metastases. These patients have more favourable tumour biology and the data suggest they do not benefit from perioperative chemotherapy [73]. Other investigators also report that chemotherapy is unlikely to impact on OS in patients with clearly resectable lesions limited to the liver [74]. Hence, patients with a relatively low risk of recurrence may not benefit so much from such chemotherapy, blurring the benefits to more high-risk patients. Many studies do not adequately stratify into risk groups for the different arms.

Systemic chemotherapy before and after the study period may also influence the end points either positively or negatively. For instance, some studies of chemotherapy in advanced disease (even though the study did not aim to convert patients with liver metastases into candidates for surgical resection) suggest prior adjuvant chemotherapy may be a disadvantage for an intensified upfront chemotherapy regimen [75]. In addition, the incremental benefit of further chemotherapy following NACT and liver resection remains unproven, but needs to be controlled for (Table 3).

Table 3 Randomised trials of adjuvant treatment for patients with CRLM
Full size table

Conclusions

Liver resection can be curative for CRLM, but can also lead to significant perioperative morbidity and mortality. The rationale for NACT in resectable CRLM derives partly from large adjuvant colon studies, which randomised several thousand patients. In contrast, most perioperative chemotherapy studies in CRLM randomised only a few hundred patients. With the lack of evidence of benefit for any targeted antibody in adjuvant treatment in stage III colon cancer and in the perioperative treatment of resectable metastatic disease, FOLFOX remains the standard of care for patients with resectable CRLM, although the benefits for one to three easily resectable metastases are probably small. The combination of FOLFOXIRI ± bevacizumab should be further investigated in more trials or borderline or not optimally resectable CRLM.

Yet, it remains unclear whether perioperative FOLFOX provides significant additional benefit in terms of either DFS or OS when compared to single-agent 5FU-based chemotherapy in the adjuvant treatment in patients with resectable CRLM. The appropriate trials have never been performed in the context of CRLM.

It is also unclear whether there should be different strategies for those with a solitary metastasis, compared with a few (1–3) but resectable metastases and finally those with borderline resectable metastases (some shrinkage required to allow sufficient liver volumes). Risk assessments in terms of Fong score are rarely applied, and factors such as performance status, age and sex as well as the RAS, BRAF and MSI profiles are not used as stratification factors. In addition, left- and right-sided primary tumours may exhibit innately different outcomes and different responses to targeted agents. One of the major challenges for the future is the determination of predictive markers of response or resistance to induction treatments. A comprehensive molecular and genetic analysis of specimens from all the phase III clinical trials to show any biological differences within subsites of right versus left colon is now required.

Also, no trial has been performed to determine the optimal duration of neoadjuvant therapy when resectability is achieved. Novel destructive, interventional and surgical techniques in combination with downstaging chemotherapy can improve resectability of borderline/unresectable CRLM but demand a close multidisciplinary cooperation. Open questions remain regarding the selection of patients who benefit from resection or local treatment, the optimal sequencing of the different modalities and the integration and duration of more effective treatments with higher response rates to treat remaining micrometastases.

However, the lessons learnt from the available phase III trials examining the addition of perioperative and adjuvant chemotherapy are that small underpowered trials, with heterogeneous inclusion criteria, performed with variable quality of surgery are likely to fail to produce robust results which can be translated into everyday practice.

There is a need for larger simple clinical trials. Larger trials with selection and stratification according to Fong scores and other risk factors should investigate new combinations of cytotoxic drugs and targeted agents in both the adjuvant and perioperative setting against both single-agent 5FU and FOLFOX.