Chemotherapy Before Liver Resection of Colorectal Metastases

Friend or Foe?

Kuno Lehmann, MD; Andreas Rickenbacher, MD; Achim Weber, MD; Bernhard C. Pestalozzi, MD; Pierre-Alain Clavien, MD, PhD, FACS


Annals of Surgery. 2012;255(2):237-247. 

In This Article


Search Strategy

The literature was scanned for publications related to liver surgery and chemotherapy according to the methodology recommended by the Cochrane Collaboration. The database (PubMed) was searched using exploded Medical Subject headings (MeSH) terms and specific text-word terms ("colorectal liver metastases," "liver surgery," "chemotherapy," "portal vein embolization," "portal vein ligation," "liver regeneration," "chemotherapy-associated steatohepatitis," "sinusoidal obstruction syndrome," "downsizing," "liver injury," "complications," and "neoadjuvant"). All search results were stored in an EndNote file and duplicates were excluded. Two authors (KL and AR) independently assessed all the abstracts for eligibility with predefined selection criteria (see below). The reference lists of retrieved papers were also screened for missing publications and publications from a manual search were added to the list.

Selection Criteria

Publications reporting the use of chemotherapy in combination with liver surgery were included in this review. Exclusion criteria were nonhuman studies, and articles not published in the English language. Only full original papers were retained, whereas editorials, letters, practice guidelines, reviews, case reports, comments, and expert opinions were also excluded. Publications concerning neoadjuvant chemotherapy were only considered if neoadjuvant chemotherapy was used preoperatively in patients with resectable liver metastases, which had to be described in the publication.

We identified 1815 studies combining the search terms "colorectal liver metastases," "chemotherapy," and "liver surgery." After exclusion of nonhuman and non-English studies and the types of study described above we identified 805 studies. Specific search terms as listed above together with studies identified by hand search lead to the articles indicated in each specific section (Fig. 1). Publications with missing endpoints (outcome, incidence of parenchymal alterations, response rate) were excluded. The level of evidence of each publication was ranked in accordance to a modified Sackett's classification.[23]

Figure 1.

Search strategy. Flow diagram showing the selection and screening process for eligible studies.

Downsizing Therapy: Defining the Optimal Regimen

The interest in downsizing unresectable metastases to enable a R0 resection has emerged with the availability of novel and effective chemotherapy regimens. The success of this approach should be mirrored by an increased rate of resectable metastases. Several recent RCTs compared different chemotherapy regimens in a palliative setting, including unselected patients with various degrees of dissemination of metastatic CRC.[20,24,25] Despite the palliative nature of the treatment, 5% to 15% of patients were later able to undergo curative resection (Table 1). In studies looking at selected groups of patients with less advanced disease, which were considered potentially resectable after chemotherapy, downsizing therapy enabled higher resection rates, often exceeding 30% (Table 2). The higher resection rates in those studies are certainly the consequence of a selection bias, the availability of more aggressive surgical strategies in expert centers reporting on small nonrandomized series.

Another crucial question is to define the most effective regimen to achieve successful downsizing. Resectability rates up to 21% were obtained in 2 randomized trials comparing the effect of 5FU with or without oxaliplatin.[18,26] The addition of antibodies to both cytotoxic agents may seem logical, but there are little data with resectability rate as a primary endpoint. Two RCTs, the CYSTAL[27] and the OPUS trials[24] showed improved resection rates from 3.7% (22/599) to 7% (42/599), and from 2.4% (4/168) to 4.7% (8/169), when cetuximab was added to irinotecan or oxaliplatin based regimens, respectively. The OPUS trial also documented an increased resectability from 4% to 10% with cetuximab in tumors without KRAS mutations.[24] However, these results are based on very few patients (3/73 in FOLFOX vs. 6/61 in FOLFOX plus cetuximab) and both resectability rates are similar to those in control groups without cetuximab in other RCTs.[18,20,25] Similar, data from a subgroup analysis of the CRYSTAL trial suggests that survival is improved in patients with a major tumor response to cetuximab.[28]

This benefit of cetuximab in KRAS wild type tumors has been questioned by the COIN trial.[29] Survival of patients without KRAS mutations was not improved when this antibody was added to an oxaliplatin-based neoadjuvant chemotherapy.[29] Recent data from genetic assessment of tumor tissues suggests that response rates and patient outcome may depend on other specific mutations of the EGFR signaling cascade.[30,31] Thus, a yet poorly defined subgroup of patients may profit from a therapy with cetuximab.

In the CELIM randomized phase II trial, cetuximab was added to both FOLFOX or FOLFIRI, resulting in a high overall resectability rate of 34% (23/68).[32] Of note, this study reported similar resection rates for patients with mutant and with wild type KRAS.[32] Because the CELIM trial used cetuximab in both arms, it failed to provide clear evidence on the benefit of adding a targeted agent to a cytotoxic agent.

Panitumumab added to oxaliplatin was assessed in the PRIME trial, including only patients with KRAS wildtype tumors.[33] This regimen did not increase resectability. A similar finding was documented for bevacizumab, that only moderately improved resectability rates from 6.1% (43/700) to 8.4% (59/700), when added to FOLFOX in a large RCT.[20] According to this study, only 16 of 700 patients presenting with unresectable liver metastases could benefit from the addition of preoperative bevacizumab. A combination of 2 antibodies, bevacizumab, and panitumumab, to cytotoxic regimens failed to add any benefit, as shown by the PACCE[34] and CAIRO-2[35] trials.

Another approach is the upfront use of all 3 effective cytotoxic drugs, ie, 5FU, irinotecan plus oxaliplatin. In a large RCT, such a regimen has led to improved progression free survival, and an improved overall resection rate of 15%, and 36% for liver-only metastases.[25] This regimen was associated with many side effects, such as diarrhea and fatigue.

We conclude that preoperative downsizing chemotherapy with oxaliplatin or irinotecan-based combinations for nonresectable metastases enables subsequent radical resections in about a third of selected groups of patients. The addition of a single targeted agent such as cetuximab or panitumumab (for KRAS wildtype tumors) or bevacizumab to cytotoxic combinations may offer a benefit for selected patients in the downsizing scenario. Given the additional costs of about 40,000 $ per patient (cetuximab for 16 weeks), the routine use of combination chemotherapy such as FOLFOX (or XELOX) or FOLFIRI in combination with cetuximab, panitumumab, or bevacizumab for downsizing liver metastases is currently not warranted. In the future, additional genotyping may help to select those patients that will respond well to a therapy with cetuximab. In young and healthy patients, the triple combination FOLFOXIRI is an appealing alternative, although the side effects are higher.

Can Response and Resectability Rates be Further Increased by Selective Hepatic Intraarterial Chemotherapy (HAI)?

HAI was designed to selectively deliver chemotherapeutic drugs to the target site. It aims at reducing side effects by reducing the overall systemic exposure to the drugs. Selective HAI may offer superior efficacy along with reduced systemic toxicity, compared to systemic chemotherapy.[36] Furthermore, placement of HAI catheters during open surgery can be combined with other procedures, such as portal vein ligation to induce hypertrophy of the contralateral liver lobe.[37,38] To avoid disease progression, the future remnant part of the liver, usually the left hemiliver or just segments II to III, can be concomitantly cleared from metastases.[13,39] After an interval of about 4 weeks, a curative hepatectomy can be attempted.[37]

Most studies on HAI have used continuous infusion of floxuridine (FUDR), a pyrimidine antimetabolite considered superior to 5FU because of its short half-life and high hepatic extraction rate. A recent meta-analysis has suggested a higher response rate to HAI than to systemic 5FU-based therapy (43% vs. 18%, respectively). However, there was no benefit on survival.[40] The evidence on the benefit of the use of the second generation of drugs is more limited for HAI, although promising data are emerging. In a recent phase II study, HAI with oxaliplatin was used in combination with systemic treatment, resulting in an impressive response rate of 64%.[41] The same regimen enabled successful downsizing and resection in 18% (8/44) of patients with metastases refractory to aggressive systemic oxaliplatin or irinotecan treatment.[42] HAI with oxaliplatin resulted in an impressive 5-year-survival of 56% in patients that were downsized and resected, compared to 0% in the nonsurgical group.[43] In addition to improved resectability, HAI might reduce relapse rates, as liver metastases, that become undetectable after HAI, were reported to remain silent in nearly two thirds of the cases.[44] This benefit of HAI is supported by a recent study that identified HAI with FUDR as a significant predictor for a complete pathologic response.[45]

We conclude that hepatic intraarterial chemotherapy may achieve more complete responses, and higher resectability rates than systemic chemotherapy. HAI should be considered in situations with a high tumor load, and in lesions refractory to systemic chemotherapy.

Is Neoadjuvant Chemotherapy Justified for Resectable Lesions?

The justification for the use of neoadjuvant therapy should rely on a significant improvement in disease free and long-term patient survival. Currently, the only randomized data available are from the EORTC Intergroup trial, in which perioperative FOLFOX (3 cycles given preoperatively, and 3 cycles postoperatively) was compared to surgery alone in a population of patients with resectable disease (Table 3). This study on 364 patients met its primary endpoint, showing a significant increase in progression-free survival at 3 years for the FOLFOX group (35.4% vs. 28.1%).[46] Yet, overall survival was not increased and the incidence of postoperative complications was higher in patients treated with neoadjuvant therapy. Unfortunately, this study did not include a third arm using adjuvant chemotherapy only. This shortcoming was recently partially addressed in a multicentric cohort study including 169 patients treated with neoadjuvant chemotherapy. Those patients were compared to 1302 patients, who underwent upfront surgery.[47] In this retrospective analysis, adjuvant chemotherapy was associated with improved survival, whereas neodadjuvant chemotherapy had no impact on disease free and overall survival. In addition, postoperative complications were increased from 24% to 37% in the group receiving neoadjuvant chemotherapy. Another retrospective study evaluating 297 patients with pre- versus postoperative chemotherapy found an improved outcome when chemotherapy (mostly irinotecan) was administered after curative liver resection.[48]

To analyze survival after neodajuvant chemotherapy, further retrospective studies have been assembled (Table 3). For example, 5-year-survival was similar in 1 study where patients received preoperative 5FU monotherapy (n = 52) versus surgery alone (n = 54). However, when patients with tumor progression under preoperative chemotherapy were excluded, the analysis indicated favorable survival in patients receiving neoadjuvant chemotherapy.[49] Similarly, another retrospective study of 131 patients found improved survival rates, when the disease was controlled under neo-adjuvant chemotherapy.[50] Thus, the most important benefit of neoadjuvant chemotherapy is probably to identify and select patients with favorable tumor biology.[51]

Complete tumor response, occurring in 2% to 14% of patients after neoadjuvant treatment (Table 3) is unwanted for resectable lesions, and may even be detrimental due to undetectable remnant tumor cells. Careful histological examination after preoperative chemotherapy has revealed the presence of vital cancer cells in surrounding halos of former tumor nodules in most cases.[52] For example in one study, 80% of metastases that had disappeared radiologically after chemotherapy contained viable tumor cells at the time of resection.[53] Together with the difficulty to localize the tumor during surgery, undetectable tumor nodules may predispose to incomplete resections and early recurrence.[53]

An undisputed drawback of neoadjuvant chemotherapy is the possibility of progressive disease on chemotherapy. In the EORTC Intergroup trial,[27] progression was observed in 7% of the patients undergoing neoadjuvant chemotherapy. Higher rates of progression up to 37% were observed in other studies (Table 2 and Table 3). Tumors with primary resistance are likely to represent a subgroup with particularly aggressive tumor biology. Although it remains unclear how these patients should be managed, proceeding immediately to surgery is an option. If not possible, other chemotherapeutic regimen could be applied, such as the addition of cetuximab.[54] Hepatic intraarterial chemotherapy with FUDR or oxaliplatin might be another successful strategy.[42] Maintaining radical tumor clearance must be the goal, as it is the only chance for long-term survival.

At this time, routine use of neoadjuvant chemotherapy for patients with clearly resectable lesions limited to the liver is not recommended due to a lack of benefit on survival.

Chemotherapy-induced Liver Injury: Many Open Questions

Chemotherapy induces various histological changes of the liver parenchyma including steatosis, chemotherapy-associated steatohepatitis (CASH), or sinusoidal injury (sinusoidal obstruction syndrome, SOS).[55–57] CASH has been found in patients treated with regimens including irinotecan.[58] It is characterized by steatosis, lobular inflammation and ballooning of hepatocytes similar to the characteristics observed in nonalcoholic steatohepatitis (Fig. 2). SOS, on the other hand, is classically associated with the use of oxaliplatin.[59] This syndrome has also been called the "blue liver syndrome" due to the characteristic bluish-red macroscopic aspect of severely damaged livers. Histological changes include nodular regenerative hyperplasia (NRH), sinusoidal dilatation, congestion and hemorrhage, leading to hepatocyte loss, perisinusoidal fibrosis and sinusoidal occlusion (Fig. 3),[60] which can lead to veno-occlusive lesions.

Figure 2.

Chemotherapy-associated steatohepatitis (CASH). A, Low power view revealing severe steatosis accentuated in centrilobular zone (H&E, scale bar: 200 μm). B, High power view showing hepatocellular steatosis and hepatocellular degeneration (ballooning, arrowhead) as well as an inflammatory infiltrate with polymorphonuclear cells (arrow) (H&E, scale bar: 50 μm).

Figure 3.

Oxaliplatin-associated liver damage. A, Low power view showing abnormal nontumoral liver parenchyma with high degree of nodularity and distinct bands of congested sinus (H&E, scale bar: 5 μm). B, High power view demonstrating dilated sinus with congestion (H&E, scale bar: 5 μm). C, More severe case revealing additional extensive centrilobular hepatocyte loss (H&E, scale bar: 200 μm, arrowhead). D, Elastica van Gieson stain highlights fibrous obliteration and occlusion of the central vein (arrow) (EvG, scale bar: 100 μm).

The pathogenesis of CASH and SOS remains unclear, and no animal models are available to study the underlying mechanisms. In analogy to the "2-hit model" of nonalcoholic steatohepatitis,[61,62] the development of CASH might only occur in patients with preexisting steatosis, when irinotecan acts as the second hit. Thus, chemotherapy-induced histological alterations might be limited to a still poorly defined patient subpopulation with either preexisting steatosis or other inherent susceptibility.

The reported incidence of SOS and CASH are highly variable (Table 4). CASH after irinotecan based chemotherapy, is only reported by few studies, whereas several investigators failed to identify a correlation with preoperative chemotherapy.[63] The reported incidence of SOS ranges from 5% to 50%. The likely reason for this large variation is the lack of a uniform classification system and the limited awareness among pathologists of chemotherapy-related changes.[64]

Does Preoperative Chemotherapy Increase the Morbidity/Mortality After Surgery?

The association of CASH and SOS with postoperative outcome has been addressed in several recent reviews.[55,56] Briefly, agreement exists on a link between the chemotherapy-associated changes and poor postoperative outcomes. A subgroup analysis of a large study on 167 major liver resections identified irinotecan-induced CASH as an independent risk factor for postoperative mortality including death from liver failure.[58] Others have identified oxaliplatin-mediated SOS as a risk factor for postoperative complications and prolonged hospital stay (Table 5). Administration of more than 6 cycles of oxaliplatin increased the risk of developing SOS and was associated with a poorer outcome.[65] Further observations suggest that prolonged administration of chemotherapy over more than 9 to 12 cycles increases the rate of reoperations and prolonged hospital stay.[66,67] An alternative to prolonged administration of chemotherapy may be the addition of cetuximab. In selected patients, a rapid tumor response after cetuximab may reduce the number of cycles required for tumor downsizing.[28]

In the EORTC trial, 6 cycles of FOLFOX before surgery increased the overall complication rate from 16% to 25%, and the rate of hepatic complications from 9% to 15%. However, mortality was not affected, resulting in 1 death per group.[46] Similar results were obtained by a recent multicentric cohort study showing a significantly higher overall complication rate (24% vs. 37%) in patients subjected to neoadjuvant chemotherapy for resectable lesions.[68]

The use of intraarterial chemotherapy was associated with an increased rate of steatosis without negative impact on postoperative complication rate.[69] However, in patients in which bilirubin levels increased during treatment, continuation of selective intraarterial application of FUDR increased the rate of biliary complications, mimicking sclerosing cholangitis.[70] The use of targeted therapies has been evaluated in several studies, and none of them reported an increase in postoperative complication rates.[54,71–76] The same conclusion was drawn for 37 resected patients after downstaging chemotherapy with FOLFOXIRI.[77]

In summary, preoperative chemotherapy increases postoperative complication rates. In the setting of extended surgical resection, preoperative chemotherapy may contribute to the development of a small-for-size syndrome and fatal liver failure.[78]

Can we Diagnose Chemotherapy Induced Liver Damage Before Surgery?

Many studies tried to identify predictive factors for chemotherapy induced liver damage.[55,56] For example, a high GGT,[79] low platelet counts, high aspartate aminotransferase to platelet ratios,[80] and an enlarged spleen[81] could be associated with SOS. The relevance of these factors should be confirmed in prospective studies, and combination of parameters will probably provide us the missing tool, ie, by scores enabling a diagnosis of SOS before liver surgery. Bevacizumab may prevent, or offer a potential treatment for SOS by reducing the incidence from 46% to 5% when added to oxaliplatin-based preoperative therapy.[60,67,82] A potential explanation for this comes from murine models of SOS, showing that the matrix-metalloprotease-9 (MMP-9), secreted by endothelial cells, was involved in the pathogenesis of SOS.[83] Bevacizumab might improve SOS by inhibiting VEGF dependent induction of MMP-9 and subsequent matrix degradation.[60]

Does Preoperative Chemotherapy Impair Liver Regeneration After Portal Vein Occlusion or Surgery?

In staged procedures, liver regeneration during the interval between the 2 procedures is crucial.[13] In humans, there is no reliable marker to assess the state of liver regeneration, only sequential imaging is used for volumetric assessments. The difficulty here is the choice of the timepoint to assess at regeneration. For example, a retrospective study of 100 patients after portal vein embolization (PVE) compared patients with or without chemotherapy and found no significant difference in liver volume after 1 month.[84] The majority of patients received 5FU in combination with either oxaliplatin or irinotecan. Similar findings were shown in other PVE studies,[85,86] and for regimens containing bevacizumab after PVE.[87] This is likely related to the considerable time interval separating the last dose of bevacizumab from PVE or PVL. Likewise, another study observed no impairment of liver regeneration, 3 months after neoadjuvant bevacizumab and liver resection.[88] These studies do not exclude the possibility that liver regeneration may be delayed by chemotherapy. For example, bevacizumab decreased liver hypertrophy after portal vein occlusion when the liver volume was assessed at an early time point of 4 weeks.[89] This seems reasonable because bevacizumab selectively targets vascular endothelial growth factor (VEGF), which is a known mediator of liver regeneration.[90,91]

If chemotherapy is used in the interval between PVE/PVL and major liver resections, patients should be carefully assessed for sufficient liver regeneration by volumetry to avoid small-for-size syndrome, and surgeons should be aware that regeneration could be delayed.[13]

When Can We Operate Patients Subjected to Chemotherapy?

In clinical practice, many patients with hepatic metastases from CRC receive chemotherapy before the evaluation by an experienced liver surgeon. Caution is needed regarding the number of cycles beyond a certain threshold that will fail to improve the response, but will elevate the likelihood of postoperative complications.[67] The improved safety of liver surgery after an extended pause after chemotherapy has been demonstrated in several studies.[92] The randomized EORTC trial, in which the mean time to surgery was 4 weeks after the last cycle of FOLFOX, was associated with increased complications, but not mortality.[46] In the majority of patients, liver function tests normalize after a period of about 4 weeks.[93] Administration of an oxaliplatin-based therapy combined with bevacizumab seems to be safe, if surgery follows 5 weeks after the last dose of bevacizumab.[73,88] Similar findings were obtained regarding wound healing in a population of patients with metastatic CRC. Wound healing was impaired during the bevacizumab treatment, but normal after discontinuation for at least 4 weeks.[94]

The level of evidence concerning the latency between the last drug application and liver resection is low. A reasonable recommendation is to avoid prolonged exposure to chemotherapy for more than 9 cycles, and to stop chemotherapy 4 weeks before surgery.[13,73]


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.