Neoadjuvant Chemotherapy for Colorectal Liver Metastases

Over the last decade, since the development of improved systemic agents for colorectal cancer, there has been increasing exploration of the use of neoadjuvant strategies for colorectal liver metastases (CRLM). In some cases, upfront chemotherapy enables downsizing, in order to enhance resectability. In other cases, neoadjuvant chemotherapy is a means to expedite delivery of systemic therapy and perhaps also to select patients for subsequent treatments.


Introduction
Over the last decade, since the development of improved systemic agents for colorectal cancer, there has been increasing exploration of the use of neoadjuvant strategies for colorectal liver metastases (CRLM).In some cases, upfront chemotherapy enables downsizing, in order to enhance resectability.In other cases, neoadjuvant chemotherapy is a means to expedite delivery of systemic therapy and perhaps also to select patients for subsequent treatments.

Surgical management of colorectal liver metastases
Since the 1980s, hepatectomy has been increasingly performed for patients with liver-only metastases from colorectal cancer.The first large series were published in the 1990s, demonstrating 5-year survival of 25-40% for patients who underwent R0 resections.These compared favorably with the 0-5% survival rates for patients with technically resectable disease who did not undergo resection, or who had positive margins at resection (R1 or R2) (Scheele & Altendorf-Hofmann, 1999).The patients included in these early series where a highly select group, mostly with solitary lesions less than 5 cm in diameter.However, from these data the initial guiding principles of surgery for CRLM were developed: a) exclusion of patients with any extra-hepatic metastases; b) removal of all detectable liver metastases while providing sufficient residual liver volume for postoperative function; and c) achievement of negative margins (R0 resection; Fong et al., 1997).With advancements in preoperative staging, surgical techniques, and multimodality treatment since that time, this definition of resectability has been greatly expanded.Current approaches extend the possibility of surgical resection to patients with CRLM involving: both lobes of the liver; large/multiple lesions requiring extended hepatectomies; and extra-hepatic sites, particularly pulmonary (Yang et al., 2010).For practical purposes, the only remaining limits on the definition of "resectable" are the achievement of negative margins, preservation of adequate remnant liver (minimum 20% with a healthy liver) with vascular and biliary inflow/outflow, and capacity of the patient to tolerate the planned, often multimodality, treatment (Charnsangavej et al., 2006;Nordlinger et al., 2009).Even with these expanded resection criteria and increasingly aggressive surgical strategies, survival rates have continued to improve.Recent large series report 5-year overall survival of 53 -58% and 5-year recurrence-free survival of 28 -36% (Abdalla et al., 2004;Choti et al., 2002;Fernandez et al., 2004;Figueras et al., 2001;Pawlik et al., 2005).These series are Fig. 1.Portal vein embolization is used to induce hypertrophy of the liver remnant.The lobe or segments containing the liver metastases (and to be resected) is embolized, if it is anticipated that the remaining liver will be too small to provide adequate functional hepatic reserve (A, B).After 4 -6 weeks, the non-embolized segments of liver will hypertrophy (C), allowing an extensive liver resection with a greater margin of safety.outcomes and long-term survival similar to that in patients who did not require PVE (Abdalla et al., 2006;Yang et al., 2010).One contraindication to PVE is the presence of lesion(s) in the FLR, due to the risk of progression with hypertrophy.In these cases, a staged approach to resection may be possible.At the initial operation, the lesion(s) in the FLR are resected and the portal vein to the remainder of the diseased liver is ligated.If the patient recovers well and sufficient hypertrophy is achieved, a second laparotomy with extended hepatectomy is undertaken 6 -12 weeks later.Perioperative morbidity and long-term survival with this approach appears to be similar to that of patients with a similar burden of disease who achieve complete resection with a single hepatectomy (Chun et al., 2007;Wicherts et al., 2008;Yang et al., 2010).The administration of chemotherapy has been successfully implemented with all of these approaches -preoperatively, between staged operations, and postoperatively.The relative contribution of surgical and chemotherapeutic components of these complex management algorithms is unknown.However, as in the more straightforward clinical scenarios discussed in detail below, any evidence of disease progression while on chemotherapy is usually cause for re-evaluation of any further surgical intervention.When complete surgical resection cannot be achieved, even with the use of adjuncts such as PVE and staged hepatectomies, radiofrequency ablation (RFA) can be used in conjunction with hepatectomy.When compared head-to-head for single lesions, RFA has a significantly higher recurrence rate and is inferior to resection (Aloia et al., 2006a).However, in the era of modern chemotherapy, patients with multiple lesions treated with a combination of resection and RFA, long-term survival rates similar to those of patients with complete resection have been achieved (Nikfarjam et al., 2009).This suggests that in patients with CRLM that are stable or regressing on chemotherapy, but which remain technically unresectable, an approach that includes resection and RFA may be considered.The final surgical scenario worth noting in the context of CRLM is the patient with synchronous disease who stabilizes or responds to neoadjuvant chemotherapy but whose liver disease remains unresectable.These patients may have an extended survival and the question often arises whether there is value in resecting their primary lesion.Data supporting a survival benefit from such a resection come from series that pre-date modern chemotherapy (Cook et al., 2005;Ruo et al., 2003).With the prolonged survival offered by current chemotherapy regimens, any added benefit is more likely to be related to prevention of local morbidity.The likelihood of developing obstruction and/or pain from local invasion appears to be greatest in patients with rectosigmoid lesions.Therefore, prophylactic resection, particularly when it can be achieved with minimal morbidity, is likely of benefit in this group of patients (Scheer et al., 2008).In all others, the chance of requiring emergent palliative surgery appears to be minimal and decisions regarding resection are made on an individual basis (Cellini et al., 2010).In general, the surgical management of CRLM is becoming more complicated and more aggressive.Partly, this is due to technical advances in hepatobiliary surgery.Perhaps of greater import is the development of newer systemic agents with greater activity against metastatic colorectal cancer.

Systemic agents used for metastatic colorectal cancer
Modern chemotherapy for metastatic colorectal cancer is typically a multi-drug regimen.Doublet regimens of a fluoropyrimidine combined with either oxaliplatin or irinotecan are highly effective in metastatic and adjuvant colorectal cancer.The addition of a targeted biologic agent further increases response rates in appropriate metastatic patients (Compton et al., 2008;Douillard et al., 2010;Folprecht et al, 2010).
Fluorouracil is a pyrimidine analog developed in the 1950s.It functions to interrupt DNA synthesis by inhibiting DNA methylation.Fluorouracil can be given in an intravenous form, 5-FU, or orally via capecitabine, which is subsequently metabolized to an active molecule.When given in IV formulation, fluorouracil is commonly combined with the folate derivative, leucovorin, to enhance its function.Irinotecan (also known as CPT-11) was approved for use in metastatic colorectal cancer in the mid-1990s.It is an inhibitor of topoisomerase I, a complex that reduces the torsional strain on DNA by breaking, detorting and reconnecting single strands of DNA.Irinotecan's action on topoisomerase I allows single strand breaks to accumulate, ultimately leading to cell cycle arrest and cell death.Oxaliplatin was introduced to the global market in the late 1990s.It is a platinum-based compound that cross-links DNA, preventing its replication and transcription.The 5-FU, leucovorin, and oxaliplatin combination, FOLFOX, is highly effective in the metastatic setting, but its long-term use is limited by a cumulative dose-dependent side effect of peripheral sensory neuropathy.Novel biologic agents have been recently introduced into clinical use and are monoclonal antibodies to proteins key to tumor growth pathways.Bevacizumab is a fully humanized antibody against vascular endothelial growth factor (VEGF), and results in inhibition of tumor angiogenesis.Cetuximab and panitumumab are antibodies to the human epidermal growth factor receptor (EGFR; also called HER1).EGFR blockade inhibits cell growth and induces apoptosis.Mutations of the KRAS gene, downstream from EGFR, negate the potential effects of EGFR inhibitors, and so only patients with wild-type KRAS benefit from this treatment (Karapetis et al., 2008).These medications are generally well tolerated, although rare but concerning side effects have been documented (Table 1B).As all of these agents are new to clinical use, their side effect profiles are likely to be refined over time.

Rationale for neoadjuvant chemotherapy
Neoadjuvant chemotherapy could be considered in instances where the CRLM are resectable, or when CRLM are not technically resectable.The rationale for giving upfront chemotherapy for each of these circumstances differs, which bears some discussion.

Unresectable colorectal liver metastasis
The utility of using chemotherapy to downstage CRLM which are unresectable to a resectable status was first described by Bismuth et al (1996).Since then, in the setting of unresectable CRLM, unless some adjunctive approach such as PVE or two-stage hepatectomy is considered, all patients with unresectable CRLM should be considered for chemotherapy.After detailed imaging, chemotherapy is administered for 2 -6 cycles, and then repeat detailed imaging is performed to evaluate for response.Patients with a response significant enough to make them candidates for resection are then reconsidered for surgery.Patients whose liver lesions do not respond or progress on chemotherapy continue on palliative-intent chemotherapy, with consideration for second-line and/or clinical trial agents.This approach has several benefits.Firstly, it defines a group of responders whose tumor biology is more favorable, and in whom aggressive resection is most likely to be of benefit.
Secondly, it provides an in vivo measurement of the effectiveness of the given chemotherapy regimen for a given patient.Depending on the response, decisions can then be made about continuing with the same agents or switching regimens -either as adjunctive therapy following surgery, or as palliative-intent chemotherapy when surgery is not possible.

Resectable colorectal liver metastasis
In patients with initially resectable CRLM, the decision to administer upfront chemotherapy is much more complicated.A neoadjuvant chemotherapy approach has some advantages.First, it allows the delivery of systemic treatment early in the patient's treatment course.
Occult or micrometastatic disease is treated before it becomes clinically visible, when the burden of disease is low.Neoadjuvant therapy further allows the patient to receive chemotherapy during optimal health, as undoubtedly some surgical patients will experience morbidity that will delay or preclude the administration of cytotoxic drugs.Secondly, neoadjuvant chemotherapy allows the clinician to assess tumor biology.Response to chemotherapy can be measured on serial imaging.Progression of disease, whether in or outside the liver, indicates resistance to the chemotherapy regimen used and thus provides the opportunity to select an alternative combination that may be more effective.Progression of disease while on chemotherapy is an independent predictor of worse survival and liver surgery may not be beneficial in this population (Adam et al., 2004b).Thus, this information further aids in selection of appropriate liver surgery patients.
There are also some disadvantages that accompany neoadjuvant chemotherapy for CRLM.First, current chemotherapy regimens have the potential for liver damage.This risk is mitigated by careful monitoring and by limited durations of pre-operative chemotherapy.
Secondly, disease progression on chemotherapy may preclude some patients who were initial surgical candidates from liver resection.This has been a rare event in trials of neodjuvant chemotherapy, and likely portends a worse overall prognosis (Nordlinger et al., 2008;Bathe et al., 2009).Thus, it may actually spare patients from surgery if they are unlikely to achieve a survival benefit.Finally, up-front chemotherapy may cause some tumors to disappear completely, making subsequent decision-making more difficult.These challenges are discussed in more detail in Section 5.

The role of neoadjuvant chemotherapy in unresectable liver-only metastases
Chemotherapy has long been used as the sole treatment modality for patients with CRLM, with the goal of palliation and the prolongation of life by a few months.With the finding that hepatectomy can lead to long-term survival in 30 -40% of patients, this goal changed.In initially unresectable patients, chemotherapy is now used with the aim of downstaging lesions to the point of resectability.If this is not achievable, then the goal shifts towards more traditional palliative-intent treatment.Clinical trials in the setting of liver-only metastases have thus begun to focus on determining the optimal chemotherapy regimen, in terms of maximizing response and resection rates, and also in terms of minimizing perioperative morbidity associated with the effects of intensive preoperative chemotherapy.In early cohort and phase II studies of unresectable patients, treatment with 5-FU, leucovorin and oxaliplatin and/or irinotecan (FOLFOX/FOLFIRI) led to a measurable response rate of 59 -72% and achievement of margin-negative resections in 12-38% of patients (Giachetti et al., 1999;Adam et al., 2004a;Masi et al., 2009).In patients undergoing resection after downstaging, median survival from the start of chemotherapy was 37 -48 months, compared to 14 -15 months in those who did not respond sufficiently to allow resection and/or were not candidates for resection following chemotherapy.These results were obtained with the administration of, on average, 10 -12 cycles of chemotherapy over a 5 -6 month period.No perioperative deaths were reported in these studies, although Masi et al. (2009) reported a perioperative morbidity rate of 27% including 8% transient liver failure.Two phase II trials have evaluated tumor response and secondary resection rates with the addition of biologic agents to FOLFOX/FOLFIRI chemotherapy.In initially unresectable patients, Wong et al. (2011) reports an objective response rate of 40% and a margin-negative resection rate of 10% after administration of capecitabine, oxaliplatin, and bevacizumab (CAPOX+B).With only 12 months of follow-up reported, median survival times have not been reached in this study.Of note, patients in this study received a median of only 4 cycles of preoperative CAPOX+B and 2 patients who achieved a complete response did not undergo resection.Bevacizumab was stopped, on average, 10 weeks prior to hepatectomy but a 4% rate of grade 2 or 3 intestinal perforation was reported.Overall, the perioperative morbidity rate was 28%, with no liver failure events reported.
In the CELIM trial (Folprecht et al., 2011), patients with unresectable CRLM were randomized to cetuximab plus either FOLFOX or FOLFIRI.The objective response rate was 62%, with a 70% response rate in the subset of patients who were K-ras wildtype -a known predictor of responsiveness to cetuximab.The overall margin-negative resection rate was 34% and patients received a median of 8 treatment cycles over a 5-month period prior to surgery.Survival data have not yet been published.Perioperative mortality was 4% and major morbidities occurred in 35%.
The addition of bevacizumab has not been conclusively shown to improve response rate or resectability rate, either in the phase II trial reviewed here or, as a secondary end-point in phase III trials (Saltz et al., 2008).On the other hand, the response rates reported in the CELIM trial are quite compelling reasons to consider cetuximab-containing regimens when attempting to downstage unresectable liver metastases, particularly in K-ras wildtype patients.
With modern chemotherapeutic regimens, usually incorporating oxaliplatin or irinotecan and biologic agents, responses to chemotherapy are well over 50% (Folprecht et al., 2010;Gruenberger, 2008aGruenberger, , 2008b)).Downstaging of initially unresectable disease to resectable disease occurs in 13 -40% of patients with liver-only metastases who have received systemic chemotherapy (Table 2 and Table 3).Higher rates of conversion to resectability are generally reported in surgical series (Table 3).The wide range of incidences of conversion of unresectable liver metastases to a resectable status is partly a function of the variations in definitions of resectability among surgeons and other oncologic specialties.This phenomenon was well illustrated in the series by Folprecht et al. (2010).Table 3.Rates of successful liver resection in surgical series where neoadjuvant chemotherapy was used to downstage tumour bulk.

The role of neoadjuvant chemotherapy in resectable colorectal liver metastases
There are few clinical trials of neodjuvant chemotherapy in the upfront resectable setting.
The EORTC 40983 trial provides the only level I evidence for its use (Nordlinger et al., 2008).This trial enrolled 364 patients with up to four colorectal liver metastases to receive perioperative chemotherapy and liver resection versus surgery alone.The chemotherapy consisted of 12 cycles of FOLFOX4, with half given pre-operatively and half postoperatively.In the chemotherapy arm, 80 percent of patients completed the preoperative schedule and 70% completed the entire regimen.At the time of surgery, 3% had a complete response, 40% a partial response, and 38% stable disease.Only 8 patients had progression that precluded surgery.Ultimately, 3 year progression-free survival went from 28.1 months in the surgery arm to 35.4 months in the multimodality arm, for a hazard ratio of 0.79.Final analysis of overall survival is still pending.
A similar trend has been seen in nonrandomized trials.2009) published a multi-institutional study of patients undergoing surgery for initially resectable, synchronous colorectal liver metastasis.A variety of treatment regimens were used, including pre-operative chemotherapy, post-operative chemotherapy, pre-and post-operative treatment, and no chemotherapy.This study found post-operative treatment for six months or longer was associated with improved overall survival in the synchronous liver metastases setting.The heterogeneity of the treatment regimens in terms of timing relative to surgery, duration of therapy, and drugs used makes this analysis difficult to interpret.
A summary of randomized and observational studies of neoadjuvant chemotherapy was recently published (Chua et al., 2010).With a heterogeneous group of chemotherapy protocols and patients, they estimated an objective response rate of 64% (range 44 -100%) and a median overall survival of 46 months (range 20-67 months).Alternative approaches to neoadjuvant chemotherapy are available.Portier et al. (1996) studied chemotherapy in the adjuvant setting with positive results.They randomized 173 patients to surgery alone versus surgery with adjuvant 5-FU/leucovorin for six months.The 5 year disease-free survival was 26.7% for the surgery alone arm and 33.5% for the combined treatment.Hepatic artery infusional chemotherapy has been advocated by some groups as well.Kemeny et al. (1999) randomized 156 patients undergoing liver resection to post-operative systemic 5-FU based chemotherapy versus 5-FU-based hepatic artery infusional chemotherapy along with systemic chemotherapy.The median survival was 59.3 months in the systemic arm and 72.2 months in the combined arm.Two other trials have found similar results (Lygidakis et al., 2001;Kemeny et al., 2002).Despite promising results, this technique has not gained popularity.The technical difficulty and expertise needed to perform the procedure safely has limited its use to a few highly specialized cancer centres, and would pose difficulty in expanding its use to the extraordinary numbers of patients with colorectal cancer.

Special problems related to neoadjuvant chemotherapy for colorectal liver metastases 6.1 Hepatotoxicity
As the administration of neoadjuvant chemotherapy has become more popular, reports have emerged that clearly demonstrate associated hepatotoxicity, which may adversely affect operative outcomes.The type of hepatic injury depends on the agents administered, while the degree of hepatotoxicity is related to duration of therapy, as well as the chemotherapeutic agents used.
Steatosis was one of the first lesions observed in association with chemotherapy.In a phase II clinical trial where irinotecan-based chemotherapy was administered prior to surgery, we observed that 66% of patients had hepatic steatosis (Bathe et al., 2009).Others have observed lower rates of steatosis, ranging from 9 -30% (Vauthey et al., 2006).This large range may be due to differences in definitions or severity of steatosis, as well as to population-specific factors such as incidence of diabetes and obesity.Vauthey et al. (2006) reported that no specific chemotherapy regimen was particularly associated with steatosis.However, steatohepatitis, which involves a monomorphic and neutrophilic inflammatory response in addition to steatosis, occurs in a minority of individuals, and it is particularly associated with exposure to irinotecan containing regimens-20.2% vs. 4.2% with no chemotherapy (Vauthey et al., 2006).Vascular changes including hepatic sinusoidal dilatation, peliosis, hemorrhagic centrilobular necrosis and regenerative nodular hyperplasia have also been observed.These lesions are particularly common in individuals who have received oxaliplatin-based regimens (Aloia et al., 2006b;Kandutsch et al., 2008;Karoui et al., 2006;Vauthey et al., 2006).Sinusoidal congestion and dilatation is present in 18 -23% of individuals who have had oxaliplatin (Aloia et al., 2006b;Vauthey et al., 2006;Wicherts et al., 2011a).Regenerative nodules appear in individuals who have had a prolonged exposure to chemotherapy, and they appear to represent an end-stage vascular injury (Aloia et al., 2006b;Wicherts et al., 2011a).
The effects of each of these lesions on liver function and clinical outcomes following liver resection vary.Steatosis has been associated with an increase in postoperative morbidity although mortality rates do not appear to be adversely affected (Belghiti et al., 1998;Kooby et al., 2003).Steatohepatitis is clearly associated with increased mortality.In the series described by Vauthey and colleagues, steatohepatitis was associated with a 90-day mortality rate of 14.7%, whereas the mortality was only 1.6% in individuals without steatohepatitis (2006).The impact of sinusoidal dilatation and other vascular injuries from chemotherapy on operative outcomes is not clear.In a number of series, sinusoidal dilatation was not associated with increased morbidity and mortality following liver resection (Kishi et al., 2010;Nordlinger et al., 2008;Kandutsch et al., 2008;Vauthey et al., 2006).However, the more severe vascular lesions, such as hemorrhagic centrilobular necrosis and regenerative nodular hyperplasia, may be associated with increased transfusion requirements and liver dysfunction (Aloia et al., 2006b;Wicherts et al., 2011a).The influence of chemotherapy on operative morbidity is proportional to the duration of neoadjuvant chemotherapy (Aloia et al., 2006b;Karoui et al., 2006;Kishi et al., 2010).This has led to a general recommendation to limit the degree of exposure to chemotherapy in the preoperative phase to a period of two to three months, if possible.In addition, lower morbidity rates have been reported when liver resection is performed more than 4 weeks after stopping chemotherapy (Welsh et al., 2007).
There is evidence that some of the biological agents modify the degree of hepatic injury induced by cytotoxic agents.In one report, the prevalence of sinusoidal injury and fibrosis was lower in patients who received cetuximab, and the prevalence of steatohepatitis was lower in patients who received bevacizumab (Pessaux et al., 2010).The addition of bevacizumab or cetuximab to neoadjuvant chemotherapy did not appear to increase the morbidity rates after hepatectomy, and was not associated with any additional histopathologic evidence of hepatic injury.In individuals exposed to 5-FU and oxaliplatin, bevacizumab appears to diminish the incidence and severity of sinusoidal dilatation (Klinger et al., 2009;Ribero et al., 2007).Bevacizumab does not appear to adversely affect postoperative liver function (Wicherts et al., 2011b), but the effects of cetuximab on liver function require further study.

Thromboembolic complications
Any systemic therapy regimen that increases the risk for thromboembolic complications should be viewed with caution when administered in the preoperative setting.Thromboembolic events that occur preoperatively may delay surgery, and patients on anticoagulants will have an increased risk of bleeding from liver resection.
In our own experience, individuals who had irinotecan-based neoadjuvant chemotherapy had a particularly high risk of thromboembolic complications.Significant thromboembolic events have previously been reported with irinotecan and bolus 5-FU/leucovorin (Pan et al., 2005;Rothenberg et al., 2001).Another study utilizing a similar chemotherapy regimen prior to liver resection did not demonstrate such a high thromboembolic event rate (Pozzo et al., 2004).Therefore it is difficult to determine whether the chemotherapy itself represents a risk factor for thromboembolic complications.A number of contributory risk factors are also present in this patient population in addition to the underlying malignancy.In particular, the insertion of indwelling central venous catheters can be associated with increased risk of thromboembolic complications (Seddighzadeh, Shetty, & Goldhaber, 2007).Given the spurious nature of reports on thromboembolic complications in irinotecan-containing neoadjuvant chemotherapy regimens, it is premature to completely dismiss their role in the management of colorectal liver metastases.
Bevacizumab is also associated with a risk of thromboembolic complications including arterial thrombosis (Kozloff et al., 2010;Schutz et al., 2010).Therefore, caution should be utilized when using bevacizumab containing regimens in the preoperative setting.Having said this, a number of series have been reported in which preoperative bevacizumab was not associated with a particularly high rate of thromboembolic complications, and surgery done more than 8 weeks after the last dose of bevacizumab was considered safe (Gruenberger et al., 2008b;Wicherts et al., 2011b).

Planning the liver resection following a significant response to chemotherapy
In this chapter we have already shown that unresectable tumors can be downstaged with chemotherapy to a point where they are rendered resectable.However, the question remains whether it is oncologically appropriate to remove less leiver in the case of liver metastases that have shrunk with chemotherapy.A liver-sparing approach would be adequate if the metastasis shrank in a concentric fashion.However, if intratumoral cell death from chemotherapy had a more random distribution, then doing a liver-sparing resection may result in leaving islands of viable tumor in the space previously occupied by the tumor.
Our group studied the histologic patterns of response to chemotherapy to define whether lesions actually shrank in a centripetal fashion (Ng et al, 2008).Our detailed histopathologic analysis demonstrated that tumor did indeed shrink centripedally.However, there were also regional differences in the degree of chemotherapy-induced cell death and fibrosis, resulting in the appearance of islands of viable tumor outside of the confines of the main tumor.Fortunately, these islands of viable tumor always resided close to the gross residual tumor.These observations provided support to the practice of removing only the residual tumor (and possibly preserving liver parenchyma), although a margin of > 1 cm might be desirable to avoid leaving behind more peripheral islands of viable tumor.

Management of a complete response to chemotherapy
Management of patients who have sustained a complete response to chemotherapy is also controversial.A series of 15 patients with complete radiographic response to chemotherapy was reported by Elias et al. (2004).All were submitted to surgery.In four patients, the lesions could be found at laparotomy, and were therefore resected.In the other 11 patients, the site of liver metastases could not be found at the time of laparotomy and were therefore left in situ.Three of these lesions eventually recurred within a median follow-up of 31.3 months.Benoist and colleagues reported another series of 66 patients with complete disappearance of metastases on CT.Thirty-one patients were observed.Within a year, 23 of these lesions reappeared on CT scan.Of the patients who went for surgery, 20 had macroscopically visible tumor at surgery, 15 had invisible metastases that were resected, and viable tumor cells were seen in 12 of the final pathologic specimens (Benoist et al., 2006).These observations suggest that, in the majority of individuals who have experienced a radiographic complete response to chemotherapy, residual tumor is present.Therefore, there is a rationale to remove all segments of liver in which tumor had resided prior to chemotherapy.Alternatively, patients can be treated expectantly with extremely close follow-up, and ablative treatments can be administered as soon as tumor recurs.

Response as a prognostic marker
One rationale for neoadjuvant chemotherapy is that it is utilized as a means of selecting patients for resection.In particular, if extrahepatic disease appears during preoperative chemotherapy, then it is not likely that the patient will benefit from hepatic metastasectomy.Clinical management is not as well defined in patients who experience disease progression while on chemotherapy yet who still have resectable disease confined to the liver.Most data suggest that progression on chemotherapy is associated with a worse prognosis.A large retrospective series reported by Adam and coworkers demonstrated that patients who progressed on chemotherapy prior to liver resection had a 5-year overall survival and diseasefree survival of only 8% and 3%, respectively (Adam et al., 2004).Similarly, Gruenberger's group reported a median recurrence-free survival of 24.7 months in patients who had a response to chemotherapy, 8.2 months in patients who had stable disease, and only 3.0 months in patients who had progressive disease (Gruenberger et al., 2008a).Others have also demonstrated the prognostic value of response to chemotherapy (Chan et al., 2010;Small et al., 2009).These observations have prompted reflection on whether individuals who have progressed on chemotherapy should be considered candidates for liver resection.On the other hand, some have observed that response to chemotherapy is not prognostic in certain circumstances, such as in synchronous metastases (Gallagher et al., 2009).Moreover, it is not clear whether resection could still provide some clinical advantage in those patients with progression.More research is required to determine the best treatment algorithm for patients who have progressed on chemotherapy yet still have technically resectable disease.

The need for a working multidisciplinary tumor conference
If neoadjuvant chemotherapy is to be considered in the management of colorectal liver metastases, then a coordinated and well-functioning multidisciplinary group is essential.It begins with review of imaging by a radiologist with specialty in the area.In consultation with the surgical team, a decision on resectability should be made from initial imaging.This stratifies the patient into: unresectable/pallative treatment; unresectable but suitable for chemotherapy in an attempt at conversion to resectable status; and resectable at presentation.With this information, the medical oncologist is given clear goals for treatment and an appropriate systemic regimen is chosen.It is important that the medical oncologist understands the concerns of the surgeon with respect to the potential hepatotoxicity of any chemotherapy, and the need to limit the course of treatment to only that which is necessary in the neoadjuvant setting.Administration of drugs such as bevacizumab in the perioperative period is also concerning and further highlights the need to coordinate the timing of chemotherapy and surgery.The surgeon should have a role in ongoing monitoring of patients undergoing a "conversion to resectable" approach in order to decide when the patient has reached a resectable state.The role of an experienced liver surgeon, as well as medical oncologist and others, in the delivery of care cannot be overemphasized.This approach to care has been endorsed by the National Comprehensive Cancer Network and others (NCCN, 2011;Vickers et al., 2010).

Future considerations
The administration of chemotherapy prior to resection of liver metastases is gaining popularity.There are a number advantages to this approach but there are also some disadvantages, as we have outlined in this article.Trials will be required, trials will be required that compare outcomes related to preoperative or perioperative chemotherapy versus postoperative chemotherapy.The design of these trials will be particularly important and will require some forethought.For example, there are some agents, such as bevacizumab and irinotecan, that are unlikely to be successful in an adjuvant setting given results from adjuvant trials in the non-metastatic setting (Saltz et al., 2007;Allegra et al., 2011).More research is required to understand the underlying cause of hepatotoxicity associated with certain chemotherapy agents.Understanding the mechanisms of hepatotoxicity may aid in developing strategies to reduce it, and ultimately enhance the safety of liver resection following chemotherapy.As biological or targeted therapies become more frequently utilized, it may be that current criteria for measuring response (Response Evaluation Criteria In Solid Tumors; RECIST) are insufficient to determine whether there is a benefit to any chemotherapy.This is due to the phenomenon that many of these agents induce more of a cytostatic rather than cytocidal response.Therefore, new methods of determining biological response (including the use of biomarkers and metabolic measures such as PET) will have to be investigated when gauging response.Eventually biomarkers will be developed to predict the likelihood of response to a particular chemotherapy and the identification of such biomarkers would result in truly personalized cancer care.Furthermore, prognostic biomarkers would help to select patients who would most likely benefit from surgery.Ultimately a combination of predictive and prognostic biomarkers would be very useful for this field and will constitute an important part of decision-making in the future.

Fluorouracil Irinotecan Oxaliplatin Common Side Effects
particular interest is the potential of chemotherapy to induce specific liver toxicity in the setting of future liver surgery, discussed further below.

Table 1B .
Side effects of biologic agents used in metastatic colorectal cancer.

Table 2 .
Incidence of successful hepatic metastasectomy in first-line chemotherapy trials.Overall, these experiences have served to make response and resection rates important end points in trials of new chemotherapeutic agents for CRLM.Downstaging unresectable liver-only disease followed by surgery is accompanied by reasonable perioperative complication rates and has become the accepted standard.