Introduction

Colorectal cancer is the third leading cause of cancer-related mortality in the USA [1]. Death of these patients usually results from uncontrolled metastatic disease. The liver is the most common site of metastasis for colorectal cancer with poor prognosis associated with a median survival of 6–12 months in untreated cases [2–4]. As shown in the last years, surgical resection is the only curative treatment option with an overall 5-year survival rate of approximately 40% [5–11]. Modern surgical strategies from major hepatobiliary centers have demonstrated that hepatectomy of as much as 70% of the liver can be performed with a mortality rate of less than 5% [12–15].

Although it is well recognized that the liver regenerates completely after major hepatectomy, the effect of hepatic regeneration on intra- and extrahepatic tumor growth is still controversially discussed. Experimental studies demonstrate enhanced growth of colorectal liver metastasis after both minor (30%) and major (60–70%) hepatectomy [16–20]. Furthermore, Rupertus et al. showed in a standardized model that growth of extrahepatic tumors correlates with the extent of liver resection due to enhancement of neovascularization and tumor cell migration [21]. These studies conclude that in experimental models, the magnitude of tumor stimulation of tumor growth is thought to be proportional to the volume of the resected liver tissue, whereas the cause for enhanced tumor growth in the regenerating liver is likely multifactorial [19, 20].

In contrast to older clinical trials recommending a standard right or left hemihepatectomy for the treatment of colorectal liver metastases [5–7], numerous different surgical approaches to hepatectomy have been developed, allowing a resection appropriate to the pathology being treated with the aim of leaving maximal residual functional liver parenchyma [22, 23]. In this context, several studies analyzed the impact of the resected liver volume on survival of patients with colorectal liver metastases leading to the conclusion that patient survival correlates negatively with the extent of liver resection [8–11, 22–24]. Therefore, parenchyma-preserving resection has been introduced as standard for patients with colorectal liver metastases at our center since March 2001. The aim of this study was to analyze the potential benefit of a low hemihepatectomy rate of less than 15% in the context of the new multi-modal therapy concepts for colorectal cancer stage IV.

Patients and methods

Data from all patients undergoing colorectal surgery and liver resection at our hepatobiliary center were prospectively entered in an ISH-Med (GSD, Berlin, Germany) database running on a SAP platform (SAP, St. Leon-Roth, Germany). For this study, patients undergoing liver resections for colorectal metastases during a 6-year period until December 2006 were reviewed. In all patients, resection of the primary colorectal cancer was categorized as formally curative defined by removal of all macroscopically detectable disease and microscopically clear resection margins. Colorectal operations were performed in combination with or without chemotherapy and radiotherapy using neoadjuvant and adjuvant protocols. Overall survival data for patients with colorectal cancer stage IV according to the international union against colorectal cancer (UICC) were obtained from the Saarland Cancer Registry [25].

All patients received intravenous antibiotic prophylaxis before surgery. The operative procedure selected to ensure adequate oncological resection margins and leave a maximal volume of hepatic parenchyma was based on the result of preoperative diagnostics, findings after laparotomy, and intraoperative ultrasound. The criteria for non-resectability were infiltration of all three liver veins, diffuse liver metastases, and non-resectable extrahepatic tumor manifestations. Partial hepatectomy (Phx) was performed as anatomical resection according to Couinaud, non-anatomical or wedge resection, and combination of anatomical and non-anatomical resections with or without “Pringle” maneuver, selective vascular clamping, or selective vascular occlusion. Major hepatectomy was defined as resection of three or more anatomical liver segments [15]. Tissue destruction within the parenchymal dissection line was usually created by ultrasonic dissection, and the resection margins of the remnant liver were coagulated by argon plasma beamer. Cholecystectomy was performed in all cases if the gallbladder was still in situ. If necessary, resection of extrahepatic manifestations was performed in all included cases. Lymph node dissection of the hepatoduodenal ligament/retropancreatic area (area 1) and along the common hepatic artery/celiac axis (area 2) was only performed if increased size and firmness of the lymph nodes as well as preoperative radiologic imaging were suspicious for malignant infiltration. Para-aortic lymph nodes (area 3) were not dissected.

Data were recorded prospectively in our database system including all demographic details, disease-related data, medical data, and data from the perioperative and postoperative course. Recurrence and follow-up information of the patients were determined from the medical records or were assessed retrospectively.

Statistical analysis

Data are expressed as absolute numbers or mean ± standard error of the mean (SEM) unless indicated otherwise. The length of follow-up was calculated from the date of liver resection at our institution. Comparisons of categorical and continuous variables were performed using the χ² test (Fisher's exact test) and the Wilcoxon rank–sum test as applicable. Differences between more than two groups were calculated by ANOVA, followed by the recommended post hoc test. To clarify and structure our data, patients with pretreatment of liver metastases before liver surgery in our center and loss of follow-up were excluded from univariate and survival analyses. Survival analyses were estimated according to the Kaplan–Meier method and compared with the log–rank test using the software package SPSS 14.0® (SPSS GmbH Software, Munich, Germany). Patients who died from unknown disease were also counted as an “event” in the Kaplan–Meier analysis as well as other patients who died from tumor recurrence. Correlations were calculated between clinical–pathological variables and the endpoints of overall survival after primary colorectal cancer operation, survival after first liver resection, recurrence-free survival, and liver recurrence-free survival. The p values of <0.05 were considered significant.

Results

Between March 2001 and December 2006, 163 patients underwent 198 liver resections for colorectal liver metastases. These 198 resections accounted for 40% of a total of 498 curative hepatic resections for benign and malign liver diseases over the same 6-year time period within our center. With respect to demographic data (Table 1), there were 102 male (62.6%) and 61 female (37.4%) patients with a mean ± SEM [range] age of 62.9 ± 0.8 years [range, 27–84] at the time of liver surgery. Previous abdominal operations other than for primary colorectal cancer were performed in 23.3% of the patients. Regarding the primary tumor (Table 2), 46% of the patients suffered from rectal cancer and 54% from colon cancer. A pT4 stage determined according to the UICC was diagnosed in 19 patients (11.7%), and positive lymph nodes were detectable in 65.6% of the resected specimens. Synchronous liver metastases, defined as liver metastases occurring within 6 months after diagnosis of the colorectal primary tumor, were evident in 42.9% of the patients. Twenty-two patients had already previous treatment of liver metastases (Table 2). Bilobular liver metastases were treated in 30.1% of the patients. Multiple different chemotherapeutic protocols were applied in 104 patients as an adjuvant regime for the primary colorectal cancer (Table 2). Thirty-four patients received neoadjuvant chemotherapy for their liver metastases before resection, including patients with a neoadjuvant protocol for their primary tumor. Furthermore, a total of 95 patients had neoadjuvant or adjuvant chemotherapy related to the liver resection procedure (Table 2).

Table 1 Demographics and characteristics (n = 163) of patients undergoing liver resection for colorectal metastases at the time of the first procedure
Full size table
Table 2 Histological classification of the primary tumor as well as characteristics and pretreatment of colorectal liver metastases of 163 patients undergoing liver resection at the time of the first procedure
Full size table

The surgical procedures performed are listed in Tables 3 and 4. Twenty-six major hepatectomies with three or more liver segments (13.1%), 65 minor anatomical resections, 78 non-anatomical resections, and 29 combinations of minor anatomical and non-anatomical procedures were performed. Twenty-six out of the 163 patients underwent repeat liver resection procedures (61 procedures overall). Additional surgical procedures were performed in 48 out of 198 liver resection procedures. Simultaneous resection of the primary colorectal cancer together with the liver metastases was performed in 19 (9.6%) patients (Table 4).

Table 3 Operative technique of liver resections for colorectal liver metastases in all cases (n = 198) and at the time of the first procedure (n = 134)
Full size table
Table 4 Additional surgical procedures during liver resections for colorectal liver metastases (n = 48 of 198 operations)
Full size table

The operative and perioperative data are shown in Tables 5 and 6. Operative time was 199.3 ± 5.2 minutes [range, 40–491]. Mean blood loss was 694 ± 65 ml. The maximum diameter of the resection area was 9.3 ± 0.4 cm [range, 2–29]. The mortality rate of the 198 procedures was 2%. Two patients died after extended liver resection due to liver insufficiency and further two patients of multi-organ failure (Table 6). The overall rate of major and minor postoperative complication was low in all 198 liver resection procedures. Twenty-nine patients developed pleural effusion requiring interventional drainage.

Table 5 Operative parameters of 198 liver resections in 163 patients
Full size table
Table 6 Length of hospital stay as well as morbidity and mortality of 163 patients undergoing 198 liver resections for colorectal metastases
Full size table

Comparing the type of liver resection, univariate analysis showed that non-anatomical resections as well as combined resections were significantly (p < 0.001) used more often in patients with bilobular metastases than anatomical resections, accompanied with a significant higher incidence of a positive resection margin (p = 0.015). Comparing anatomical and combined resections with non-anatomic resections, non-anatomic resection was associated with significant shorter operation time (p = 0.006), lower blood loss (p < 0.001), and lower number of red blood cell transfusions (p < 0.001). Anatomical resections had a significant higher incidence of pleural effusions requiring interventional drainage than non-anatomic and combined resection (p = 0.027). Univariate analysis of patients with a major hepatectomy versus minor resections showed that CEA levels >100 ng/ml were significantly (p = 0.025) higher in patients with extended resections. Major resections required significant more red blood cell transfusions (p = 0.017) due to a higher blood loss (p = 0.003) and were accompanied by a higher incidence of interventional pleural drainages (p = 0.005) compared to minor resections. Patients undergoing repeated liver resections had significantly (p = 0.005) more rectal than colon cancer and significantly more synchronous (p = 0.021) as well as bilobular metastases (p = 0.015) compared to patients with single liver resection procedure only. Comparing the time interval between diagnosis of the primary cancer and occurrence of the liver metastases (≤6 vs. >6 months or ≤12 vs. >12 months, respectively), no significant differences could be found using univariate analysis.

For survival analyses, all patients with pretreatment of liver metastases before liver surgery (n = 22) and loss of follow-up (n = 7) were excluded. Out of these 134 patients, the mean follow-up interval was 31.7 ± 1.6 months. The overall 5-year survival rate for the entire cohort was 61% after operation of the primary colorectal cancer tumor and 40% after the first liver resection procedure, with an overall recurrence-free survival rate of 30% and a liver-related recurrence-free survival rate of 58% (Fig. 1). For subgroup analyses, already known predictors of poor long-term outcome were analyzed [6, 9, 13, 15]. Morphological predictors and factors related to the primary tumor are presented in Table 7. Patients with liver metastases from colon cancer had a significantly longer overall survival after resection of the primary tumor (p = 0.048) and a longer overall recurrence-free survival (p = 0.030) than patients with rectal cancer. Patients with a pT4 and a nodal positive stage of their primary tumor had a shorter survival rate after resection of the primary tumor as well as after liver resection, but these data were not statistically significant. The temporal relationship (Table 7) between diagnosis of the primary tumor and occurrence of the liver metastases (synchronous vs. metachronous and <12 vs. >12 months, respectively) was found as a significant predictor for the overall survival after operation of the primary tumor (p < 0.001 and p < 0.001, respectively) and the recurrence-free survival (p = 0.011 and p = 0.005, respectively) as well as the liver-related recurrence-free survival (p < 0.001, p < 0.001). The hepatic distribution of metastases (Table 7) showed that patients with unilobular metastases had a higher survival rate than patients with bilobular metastases, however, without statistical significance. Size (≤5 vs. >5 cm) and number of liver metastases (≤3 vs. >3) had no influence on the survival rates of the patients (Table 7).

Fig. 1
figure 1

Patients' (n = 134) overall survival after operation of the primary colorectal tumor (a) and after the first resection procedure of liver metastases (b). Overall tumor-free survival (c); liver-related tumor-free survival after liver resection (d)

Full size image
Table 7 Survival analyses of 134 patients with hepatic colorectal cancer metastases undergoing liver resection related to variables of the primary tumor and their liver metastases
Full size table

Additionally, possible predictors of poor long-term outcome related to surgical techniques for the resection of liver tumors were analyzed (Table 8). Comparing the type of liver resection for the first procedure, patients undergoing non-anatomical resections had a longer survival after liver resection compared to anatomical or combined procedures, without showing statistical significance (Fig. 2, Table 8). If multiple non-anatomic resections due to the number of metastases were needed, these patients had a shorter survival after the liver resection procedure than those with only one non-anatomic resection (Table 8). In case of intrahepatic recurrence, 23 of 134 patients underwent at least a second hepatectomy, with a maximum of four repeated procedures during the time period reported (Table 8). Following repeated hepatectomy, these patients had a comparable 3-year survival rate after liver resection (66% versus 61%) with a notable high 5-year survival rate (27% versus 45%) compared to patients with only one liver resection procedure (Fig. 3). Patients undergoing major hepatectomy had a lower 3-year related survival rate (54% versus 64%) than those with minor resections (Fig. 4, Table 8). Interestingly, a diameter of the resection area of ≥10 cm (Fig. 5), as an indicator for major liver resection surgery, had a significant influence on the overall survival of the patients (p = 0.043) and the survival after the liver resection procedure (p = 0.015). Additionally, although 38 patients had a positive resection margin of <1 mm with a lower 5-year survival rate (20%) compared to patients with R0 resection, these data failed to be statistically significant (Table 8). Additional resection of known extrahepatic diseases during the liver resection procedure positively influenced the 5-year survival in these patients, however, again without statistical significance (Table 8).

Table 8 Survival analyses of 134 patients with hepatic colorectal cancer metastases undergoing liver resection related to surgical techniques
Full size table
Fig. 2
figure 2

Patients' (n = 134) overall survival after the first resection procedure of liver metastases (a) and the liver-related tumor-free survival after liver resection (b), according to the type of liver resection (anatomical vs. non-anatomical vs. combined anatomical and non-anatomical procedures)

Full size image
Fig. 3
figure 3

Patients' (n = 134) overall survival after operation of the primary colorectal tumor (a) and after the first resection procedure of liver metastases (b), according to the number of liver resection procedures (1 procedure vs. >1 procedure). Overall tumor-free survival (p < 0.001) (c) and liver-related tumor-free survival after liver resection (p < 0.001) (d)

Full size image
Fig. 4
figure 4

Patients' (n = 134) overall survival after operation of the primary colorectal tumor (a) and after the first resection procedure of liver metastases (b), according to the extent of hepatectomy (<3 segments vs. ≥3 segments). Overall tumor-free survival (c) and liver-related tumor-free survival after liver resection (d)

Full size image
Fig. 5
figure 5

Patients' (n = 134) overall survival after the first resection procedure of liver metastases (a, p = 0.015)) and the liver-related tumor-free survival after liver resection (b), according to the extent of the dissection area (<10 vs. ≥10cm)

Full size image

Discussion

A variety of possible explanations for the improved outcome of patients with colorectal liver metastases within the last decades can be considered. Advantages in preoperative imaging, sophisticated surgical techniques, as well as neoadjuvant and adjuvant chemotherapeutic protocols may have contributed to the improvement in long-term outcome. Nevertheless, patients with stage IV disease of colorectal cancer still have a poor prognosis. During March 2001 and December 2006, 225 patients with colorectal cancer UICC stage IV were registered in the Saarland Cancer Registry [25]. Survival analysis of these 225 patients according to the data from the Saarland Cancer Registry showed a relative survival rate of 77% after 1 year, 38% after 3 years, and 14% after 5 years. If surgical resection for hepatic colorectal metastases with curative intent is possible, long-term survival and potential to cure could be demonstrated by our study presenting a 5-year survival rate of up to 40% which is in line with the new literature [5–11, 22]. Patients with bilateral colorectal metastases usually have either large or multiple tumors. Multiple tumors (≥4 tumors) and large tumor size (≥5 cm) are the most important independent adverse predictors of survival in patients undergoing hepatic resection for colorectal liver metastases [9, 26, 27]. Recurrence after resection of bilateral colorectal metastases is common with only 18% remaining recurrence free at 5 years [9].

The optimal therapeutic strategy for synchronous resectable colorectal liver metastases—defined as metastases occurring within 6 months after diagnosis of the primary tumor—has not been defined yet. Comparing patients with synchronous vs. metachronous colorectal liver metastases, Bockhorn et al. demonstrated similar 5-year survival rates after liver resection [28]. As shown by the present study, the temporal relationship between the occurrence of the primary tumor and liver metastasis seems to be an important prognostic factor for survival. Survival data after the first liver resection showed longer 5-year survival after resection of metachronous metastases, however without reaching statistical significance. The liver metastases and colorectal tumors may be removed in a simultaneous or staged fashion. As shown by Martin et al. simultaneous resection avoids a second laparotomy and reduces the overall complication rate without increasing operative mortality [29]. Therefore simultaneous resection is recommended at our institution. If the liver and colorectal resections are both extensive (i.e., extended hepatic resection and low anterior resection), a staged approach with neoadjuvant chemotherapy will be preferred according to the recommendations of Bentrem et al. [30].

The role of liver resection for colorectal metastases in the setting of associated hilar or para-aortic lymph node metastasis still remains controversial. As lymph node dissection was not routinely performed; the current study could not define a potential benefit of surgical resection. As shown by Pulitano C. et al., the presence of limited resectable lymph node metastases is not an absolute contraindication for the liver resection procedure, as subgroups of patients have a long-term survival benefit [31]. Especially, lymph nodes of areas 1 and 2 should always be removed if involved by metastases, whereas long-term survival of patients with metastases within area 3 is rare [31, 32].

Within the last decade, a shift toward parenchyma-sparing liver surgery can be observed in the reports of most hepatobiliary centers [8–11, 22–24]. Several studies report that perioperative mortality can be decreased using a segment-oriented approach to hepatic resection, obviating the need for resection of large amounts of uninvolved parenchyma [12–15]. As shown by Gold et al., this concept has become evident by the decreased use of major and extended hepatectomies and by the decreasing number of segments resected in a large series of patients [9]. To spare uninvolved surrounding parenchyma, similarly wedge resections and ablations were used more often. Interestingly, this change in management was not related to tumor size and tumor numbers. Wanebo et al. showed with a multivariate analysis that bilobular distribution of metastases and extent of liver resection (wedge and segmental versus hemihepatectomy and trisegmentectomy) directly correlate with patient survival [24]. Furthermore, Stewart et al. showed a median survival of 51 months after standard hepatectomies, of 23 months following extended resections, and of 59 months after segmental resection; however, there was no statistical difference between the three groups regarding the 5-year survival [22]. Interestingly, Jonas et al. demonstrated that survival after minor resections was significantly longer than after major resections (both groups were not defined in this analysis) [7], but the authors point out that in solitary liver metastases, the extent of the liver resection was not always determined by the extent of the tumor. Although only a randomized controlled trial could provide exact data regarding the impact of liver resection volume on patient survival taking into account the aspects of tumor size, distribution, and localization, the current analysis and the known literature recommend parenchyma-sparing liver surgery for patients with colorectal liver metastases.

In contrast to published data [7], in the present study, the extent of the liver resection procedure was strictly determined by the extent of the tumor. No standard hemihepatectomy was performed, so that the concept of parenchyma-preserving resection as a standard for patients with colorectal liver metastases resulted in an overall major hepatectomy rate of 13.1% (Table 9). Although no statistical significance of patient survival could be observed between patients undergoing major vs. minor liver resection, the data of the dissection area as an indicator for a large trauma of parenchyma indicate that major resections with high parenchymal loss (dissection diameter ≥10 cm) are associated with a shorter 5-year survival. Furthermore, the predicted and current survival rates compare favorably with results from other centers. In context with experimental findings, demonstrating that the extent of liver resection correlates with growth of residual liver or extrahepatic metastases [21], our clinical data indicate a correlation between loss of hepatic parenchyma and overall survival after liver resection.

Table 9 Review of the literature of the last decade indicating the rate of major hepatectomy (≥3 segments) for resection of colorectal liver metastases
Full size table

The effect of surgical techniques for the resection of colorectal liver metastases comparing anatomical segment resections with non-anatomical wedge resections is still controversially discussed in the literature [33, 34]. Whereas some authors showed that anatomical resections are superior to wedge resections according to a better tumor clearance and improved survival [35], others demonstrated equivalent results for both techniques [11, 36–39]. Additionally, the present study shows that non-anatomical resections are not associated with impaired survival after liver resection. The non-anatomical liver resection is an oncological equivalent alternative and should remain an integral component of surgical treatment of colorectal liver metastases. Furthermore, Kokudo et al. demonstrate that major hepatectomy was unnecessary in >80% of the cases when the tumors were resectable by non-anatomical limited resections [36]. Taken together, the type of resection should be based on the anatomy of the lesion and the goal of preserving an adequate volume of functional liver parenchyma. Several studies recommend repeated liver resection as an option in multi-modal treatment of recurrent colorectal cancer [40–43]. In this context, evaluation for repeated hepatectomy should be performed at first. As demonstrated by the present study, repeated liver resection is safe and provides prolonged survival.

R0 resection is a positive prognostic factor for the outcome of patients with colorectal liver metastases. Colorectal liver metastases are well circumscribed, and micro- or satellite-metastases are not common. Therefore, a surgical resection margin of 2–5 mm has become a clinically acceptable minimum requirement reducing the risk of margin-related recurrence [7, 26, 44, 45]. Patients with positive margins were more likely to have surgical margin recurrence [45]. Otherwise, as liver resection provides the only chance of cure, complete removal of the tumor with a minimal margin is justified when technically unavoidable. This is not always achievable when multiple lesions are present or when they are located deep inside the liver and close to major vessels. Interestingly, this point is of major interest in patients with colorectal cancer metastases involving the caudate lobe. Due to its special anatomy, these cases of cancer-involved resection margins of over 50% were reported in the literature, leading to a 1-year overall survival of up to 90% with a 1-year recurrence-free survival of 60–70% [46]. Furthermore, as described by de Hass et al., optimal surgical techniques allowed an overall 10-year survival of 37% of patients with an R1 status compared to 43% of patients with an R0 resection [47]. In this context, the results of the present study with an R1 status of 28% are in line with the literature. In our center, all patients undergoing liver resection received treatment of the resection margin using the argon plasma coagulation device for further parenchymal cell destruction to achieve a deeper clear resection margin. Cryosurgery is only indicated to obtain deep parenchymal destruction next to major vessels. Together with the actual literature and the new multi-modal chemotherapy treatment options, a predicted positive surgical resection margin no longer should be an absolute contraindication for aggressive or advanced surgery of liver metastases. Although an R1 resection should clearly be avoided, the tumor biology is the more important predictor for intrahepatic recurrence and survival rather than millimeters of the resection margin [48, 49].

In conclusion, the availability of a highly effective systemic chemotherapy enables patients with stage IV disease to survive long enough to develop new metastases, making them candidates for repeated hepatectomy. Therefore, maximizing the amount of residual liver tissue is of considerable importance, and non-anatomical, parenchyma-preserving liver resections pursue this aim. Patients undergoing minor liver resections have the longest survival suggesting adequate oncological resection margins, whereas major hepatectomy has its role in selected patients with bilobular or large solitary colorectal metastases.