Laparoscopic repeat liver resection after open liver resection: A comparative study from a single-centre.

BACKGROUND: Technological innovations have made it possible to use laparoscopic liver resection in cases with pre-existing adhesions or cicatricial changes. However, laparoscopic repeat liver resection (LRLR) still represents a challenge for surgeons, especially in case of previous open liver surgery. This study evaluated the outcomes of LRLR after open liver resection (OLR) in cases of recurrent liver cancer.
MATERIALS AND METHODS: A total of 62 patients who underwent laparoscopic minor liver resection at our institution between September 2012 and September 2016 were retrospectively divided into an LRLR group (n = 13) and a laparoscopic primary liver resection group (LPLR; n = 49). The two groups were compared in terms of patient demographics, surgical procedures and short-term outcomes. Recurrence-free survival (RFS) and overall survival (OS) were compared for patients with hepatocellular carcinoma (HCC) or colorectal liver metastases (CRLMs).
RESULTS: There was a significant intergroup difference in the hepatitis virus background, although the two groups' primary histology and pre-operative liver function were comparable. The two groups had statistically similar values for extent of resection, operative time, estimated blood loss, transfusion requirement, conversion to laparotomy, post-operative complications, surgical margins, time to oral intake and hospital stay. No significant differences were detected when we stratified the cases according to low and intermediate difficulty. Furthermore, there were no intergroup differences in RFS or OS in the two groups for patients with HCC and CRLM.
CONCLUSIONS: The findings suggest that minor LRLR after OLR is safe and comparable with minor LPLR in the present study.

INTRODUCTION

The number of laparoscopic liver resections (LLRs) has increased steadily since the First International Consensus Conference on laparoscopic liver surgery (Louisville, 2008). Moreover, the Second International Consensus Conference on LLR (Morioka, 2014) defined the current role of LLR and developed relevant recommendations.[1] Although laparoscopic repeat liver resection (LRLR) was not discussed at these conferences, improvements in treatment outcomes among patients with liver cancer have increased the number of cases that require repeat liver resections, which highlight the importance of LRLR. Two reports have indicated that open repeat liver resection (ORLR) is effective for recurrent hepatocellular carcinoma (HCC)[2] and colorectal liver metastases (CRLMs),[3] and it is possible that LRLR could be useful in those cases, although few reports have described this approach. This may be because LRLR has inherent surgical challenges (e.g., adhesions, cicatricial changes and altered anatomy as a result of the previous resection), which could delay the implementation of LRLR in clinical practice. A recent report has provided some evidence regarding the utility of LRLR,[4] and a difficulty scoring system from the International Consensus Conference on LLR[5] may be useful for determining the indication for laparoscopic surgery during primary and repeat resections. This study aimed to evaluate the safety and feasibility for LRLR after previous open liver resection (OLR), based on the difficulty scores, and to compare the outcomes of LRLR and laparoscopic primary liver resection (LPLR).

MATERIALS AND METHODS

This single-centre retrospective study evaluated patients who were treated at the Keio University School of Medicine (Tokyo, Japan), and the study protocol was approved by the Institutional Review Board. The retrospective search identified 157 patients who underwent LLR for liver tumours between September 2012 and September 2016. However, we excluded 39 patients with diseases other than HCC and CRLM, 10 patients with only previous LLR, 28 patients who underwent liver resection that was more extensive than sectionectomy (with the exception of the left lateral sectionectomy [LLS]), 15 patients who underwent combined resection of other organs and 3 patients with an incomplete workup or insufficient follow-up (<6 months). Thus, 62 patients were included in the analyses and divided into an LRLR group (13 patients who had previously undergone at least one OLR) and an LPLR group (49 patients who had not undergone OLR). The informed consent to publish the current article was obtained from each patient.

The patients' records were searched to collect information regarding age, sex, primary histology, hepatitis virus background, body mass index, liver function test results (Child–Pugh score, indocyanine green retention at 15 min [ICG15R], and presence of liver cirrhosis), previous abdominal surgery and diagnosis of recurrence of liver cancer. The surgical procedures were analysed based on the extent of the liver resection, purely laparoscopic or laparoscopy-assisted, the use of the Pringle manoeuvre, and the difficulty scores. The extent of liver resection was defined according to the Brisbane 2000 terminology,[6] and partial resection in this study was defined as including small anatomical resection. The difficulty score was calculated based on the tumour's size, extent of liver resection, tumour location, proximity to major vessels and fibrosis severity.

For cases of LRLR, we also reviewed the extent of the previous liver resection, the previous non-resection liver therapies, the surgery interval and post-operative outcomes (primary histology, procedure, complications, operative time, blood loss and hospital stay). For the analysis of the short-term post-operative outcomes, except two cases of high difficulty, 60 cases were categorised as low or intermediate difficulty based on their difficulty scores, and the two difficulty subgroups of the LRLR and LPLR groups were compared in terms of operative time, blood loss, transfusion requirement, conversion to laparotomy, post-operative complications, positive surgical margins, time to oral intake and hospital stay.

Post-operative complications were classified using the Clavien-Dindo grading system, and ≥CDIIIa and

Surgical devices and techniques

All dissections of intra-abdominal adhesions were performed using a laparoscopic ultrasonically activated scalpel. Hepatic parenchymal transection was performed using a laparoscopic ultrasonically activated scalpel, CUSA EXcel Plus™ (Integra Lifesciences Corp., Plainsboro, New Jersey, USA), and BiClamp™ (Erbe Elektromedizin Gmbh, Tübingen, Germany). The Pringle manoeuvre was performed extracorporeally using silicon tape and a soft catheter, if necessary. Conversion to a laparoscopy-assisted surgery was performed if necessary (e.g., in cases with severe adhesions or cicatricial changes).

Statistical analysis

All analyses were performed using SPSS software (version 23; IBM Corp., Armonk, NY, USA). Comparisons were performed using Student's t-test for continuous variables, and the Chi-squared test or Fisher's exact test for categorical variables. Categorical data were reported as number and frequency (%), and continuous data were reported as median (range). The Kaplan–Meier method and log-rank test were used to compare RFS and OS outcomes. Two-tailed P < 0.05 was considered statistically significant.

RESULTS

During the study, we identified 15 cases of open minor liver resection. These cases were not treated using LLR because of previous haemihepatectomy (four cases), multiple tumours (one case), combined resection of other organs (six cases) or unknown reasons (two cases).

The patients' demographic characteristics are shown in Table 1. The median age in the LRLR group was 70 years (range: 54–88 years), compared to 72 years (45–87 years) in the LPLR group (P = 0.839). The primary histology results in the LRLR group were HCC (n = 10) and CRLM (n = 3), compared to HCC (n = 40) and CRLM (n = 9) in the LPLR group (P = 0.702). There was a significant inter-group difference in the hepatitis virus background, although no significant differences were observed for Child–Pugh score, ICG15R, or the presence of liver cirrhosis. Four patients in the LRLR group (31%) had undergone previous abdominal surgery (except previous liver surgery), compared to 16 patients (33%) in the LPLR group (P = 0.897). Eleven patients in the LPLR group (22%) were diagnosed with a recurrence of liver cancer, compared to all patients in the LRLR group (100%) (P < 0.001).

Table 1

Patient demographics

	LRLR group (n=13)	LPLR group (n=49)	P
Basic characteristics
Age (years)	70 (54-88)	72 (45-87)	0.839
Sex, male	11 (85)	35 (71)	0.334
Primary histology			0.702
HCC	10 (77)	40 (82)
CRLM	3 (23)	9 (18)
Hepatitis background			0.036*
HBV	5 (38)	5 (10)
HCV	4 (31)	15 (31)
Neither	4 (31)	29 (59)
Body mass index, kg/m2	24 (17-27)	24 (16-40)	0.209
Liver function
Child-Pugh score			0.459
A	13 (100)	47 (96)
B	0	2 (4)
ICG15R, %	9.4 (2.8-27.1)	14.1 (2.7-81.5)	0.085
Liver cirrhosis	3 (23)	17 (35)	0.426
Previous abdominal surgery
Cases	4 (31)	16 (33)	0.897
Surgeries	5	18
Recurrence of liver cancer	13 (100)	11 (22)	<0.001*

Values are reported as median (range) or number (%). *Indicates a significant difference (P<0.05). LRLR: Laparoscopic repeat liver resection, LPLR: Laparoscopic primary liver resection, HCC: Hepatocellular carcinoma, CRLM: Colorectal liver metastasis, HBV: Hepatitis B virus, HCV: Hepatitis C virus, ICG15R: Indocyanine green retention at 15 min

Table 2 shows the details of the surgical procedures. In the LRLR group, one patient (8%) underwent LLS and 12 patients (92%) underwent partial resection, compared to LLS in two cases (4%) and partial resection in 47 cases (96%) in the LPLR group (P = 0.437). Both groups had similar laparoscopic approaches and numbers of cases that involved purely laparoscopic or laparoscopy-assisted surgery and the Pringle manoeuvre. There was no significant difference between the two groups' difficulty scores (P = 0.483).

Table 2

Surgical characteristics

	LRLR group (n=13)	LPLR group (n=49)	P
Extent of resection
Left lateral sectionectomy	1 (8)	2 (4)	0.590
Partial	12 (92)	47 (96)
Techniques
Purely laparoscopic	11 (85)	48 (98)	0.109
Laparoscopy-assisted	2 (15)	1 (2)
Pringle manoeuvre	7 (54)	27 (55)	0.936
Difficulty score
Low	5 (38)	25 (51)	0.483
Intermediate	7 (54)	23 (47)
High	1 (8)	1 (2)

Values are reported as number of patients (%).LRLR: Laparoscopic repeat liver resection, LPLR: Laparoscopic primary liver resection

The details of the 13 LRLR cases are shown in Table 3. The previous liver resection had involved haemihepatectomy in one case, sectionectomy in five cases, and partial liver resection in seven cases. The post-operative complication rate was 15% (two bile leaks), although the mortality rate was 0%. Bile leak (CD–IIIa) was detected in two cases, which were successfully treated using only endoscopic nasobiliary drainage.

Table 3

Details of the laparoscopic repeat liver resection cases

Case	Sex	Age	Primary histology	PLR procedure	Tumour location in PLR	Previous liver therapy	Surgery interval (months)	LRLR procedure	LA	Post-operative complication	Difficulty	Operative time (min)	Blood loss (g)	Hospital stay (days)
1	Male	70	HCC	ex. RPS	S6/7	3 × Cryo, MCT, RFA	13	S5 par	P		I	333	350	11
2	Male	62	HCC	par	S5/6	-	17	S3, S5 par	P		L	297	350	10
3	Male	54	HCC	LLS	S2/3	-	72	S4 par	P		L	396	200	11
4	Female	80	HCC	par	S4/5	RFA	54	LLS	P	Bile leak	I	389	100	24
5	Male	60	HCC	par	S8	-	24	S2 par	P		L	212	5	7
6	Male	70	CRLM	par	S4/8	-	49	S4 par	P		L	276	5	8
7	Male	69	HCC	par and ex. RPS	S1, S5/6	TACE, MCT, RFA	24	S4 par	P		L	377	50	6
8	Male	88	CRLM	par	S5/6, S8	-	37	S8 par	H		I	480	350	46
9	Female	70	HCC	LH	S2/3	PEIT, TACE, 2 × RFA	66	S5 par	P		L	317	0	7
10	Male	78	HCC	par	S1	PEIT, TACE, RFA	39	S4/8 par	H	Bile leak	I	556	870	40
11	Male	65	HCC	ex. RPS	S6/7, S5/8	-	95	S3/4, S5 par	P		I	553	650	8
12	Male	72	HCC	par	S2, S4	TACE, RFA	129	S3/4 par	P		H	346	100	19
13	Male	65	CRLM	ex. RPS	S5/6		18	S8 par	P		I	278	10	9

HCC: Hepatocellular carcinoma, CRLM: Colorectal Liver Metastasis, PLR: Previous liver resection, Ex. RPS: Extended Right Posterior Sectionectomy, Par: Partial resection, LLS: Left Lateral Sectionectomy, LH: Left Haemihepatectomy, Cryo: Cryoablation, MCT: Microwave Coagulation Therapy, RFA: Radiofrequency ablation, TACE: Transcatheter Arterial Chemoembolisation, PEIT: Percutaneous Ethanol Injection Therapy, LRLR: Laparoscopic Repeat Liver Resection, LA: Laparoscopic approach (H: Hybrid, P: Purely Laparoscopic Resection), I: Intermediate, L: Low, H: High

Table 4 shows the short-term outcomes in the LRLR and LPLR groups. When we considered all cases, both groups had similar values for operative times, estimated blood loss, transfusion requirement, conversion to laparotomy, post-operative complications, surgical margin, time to oral intake and hospital stay. When we considered the low difficulty cases (LRLR: 5 patients and LPLR: 25 patients), there were no significant inter-group differences in those parameters. Similarly, no significant differences were observed when we considered the intermediate difficulty cases (LRLR: 7 cases and LPLR: 23 patients).

Table 4

Short-term surgical outcomes

All cases	LRLR (n=13)	LPLR (n=49)	P
Operative time (min)	346 (212-556)	297 (84-831)	0.712
Estimated blood loss (g)	100 (0-870)	150 (0-2400)	0.611
Transfusion requirement	1 (8)	7 (14)	1.000
Conversion to laparotomy (%)	0	0	NA
Complications (≥ CDIIIa)	2 (15)	5 (10)	0.630
Complications (< CDIIIa)	0	5 (10)	0.574
Positive surgical margin	2 (15)	10 (20)	0.684
Time to oral intake (days)	2 (2-4)	2 (1-12)	0.814
Hospital stay (days)	10 (6-46)	9 (5-50)	0.351
Low difficulty cases	LRLR (n=5)	LPLR (n=25)	P
Operative time (min)	297 (212-396)	282 (84-641)	0.852
Estimated blood loss (g)	50 (5-350)	50 (0-2400)	0.603
Transfusion requirement	0	3 (6)	1.000
Conversion to laparotomy (%)	0	0	NA
Complications (≥CDIIIa)	0	1 (2)	1.000
Complications (<CDIIIa)	0	2 (4)	1.000
Positive surgical margin	1 (8)	4 (8)	1.000
Time to oral intake (days)	2 (2-2)	2 (1-5)	0.753
Hospital stay (days)	8 (6-11)	8 (5-16)	0.760
Intermediate difficulty cases	LRLR (n=7)	LPLR (n=23)	P
Operative time (min)	389 (278-556)	416 (202-831)	0.868
Estimated blood loss (g)	350 (0-870)	200 (0-2250)	0.861
Transfusion requirement	1 (7)	3 (6)	1.000
Conversion to laparotomy (%)	0	0	NA
Complications (≥CDIIIa)	2 (14)	3 (6)	0.565
Complications (<CDIIIa)	0	3 (6)	1.000
Positive surgical margin	1 (7)	6 (12)	1.000
Time to oral intake (days)	3 (2-4)	3 (1-12)	0.864
Hospital stay (days)	11 (7-46)	12 (6-50)	0.264

Values are reported as median (range) or n (%). NA: Not available

Post-operative major complications (≥CDIIIa) were observed in two LRLR cases (both were bile leaks) and in five LPLR cases (two bile leaks, two liver failures and one intraperitoneal haematoma that required abdominal puncture and drainage). Post-operative minor complications (

In the LRLR group, patient 4 (the first case with a bile leak) underwent LLS and dissection of Glisson's sheath at segments 2 and 3. Although the bile leak was not detected during post-operative cholangiography, we inferred that a bile leak was present at the biliary tract of segment 2, based on the post-operative abdominal CT findings. Patient 10 (the second patient with a bile leak) had a bile leak from the peripheral biliary tract of segment 8, which drained into the left branch of the hepatic duct. In that case, we attempted S4/8 partial resection beside a previous scar from radiofrequency ablation, as the pre-operative CT revealed a small nodule in the vicinity of the main tumour. However, bleeding developed from the Glisson's sheath of segment 8, which we stopped using a Z-shaped suture. Post-operative cholangiography revealed a bile leak from the biliary tract of this segment. Both patients had high total bilirubin levels in their drained fluid on the first post-operative day.

Among patients with HCC (LRLR: 10 patients and LPLR: 40 patients), the 1-year RFSs were 60% and 82%, respectively, and the 3-year RFSs were 36% and 64%, respectively. Furthermore, the 1-year PRSs were 90% and 97%, respectively, while the 3-year PRSs were 90% and 86%, respectively. When we compared the HCC cases in the LRLR and LPLR groups, no significant differences were observed in RFS (P = 0.118), PRS (P = 0.659) or OS (P = 0.693) [Figure 1a–c]. Among patients with CRLM (LRLR: three patients, LPLR: nine patients), the 1-year RFSs were 50% and 67%, respectively, and the 3-year RFSs were 50% and 53%, respectively. Furthermore, the 1-year PRSs were 100% and 89%, respectively, while the 3-year PRSs were 100% and 74%, respectively. When we compared the CRLM cases in the LRLR and LPLR groups, no significant differences were observed in RFS (P = 0.926), PRS (P = 0.533) or OS (P = 0.362) [Figure 2a–c], although the sample size for this analysis was small.

Figure 1

Recurrence-free survival, procedure-related survival and overall survival among cases with hepatocellular carcinoma. Among cases with hepatocellular carcinoma, no significant differences between the two groups were observed for (a) recurrence-free survival (P = 0.118) or (b) procedure-related survival (P = 0.659) and (c) overall survival (P = 0.693)

Recurrence-free survival, procedure-related survival and overall survival among cases with colorectal liver metastases. Among cases with colorectal liver metastases, no significant differences between the two groups were observed for (a) recurrence-free survival (P = 0.926) or (b) procedure-related survival (P = 0.533) and (c) overall survival (P = 0.362)

DISCUSSION

Improved survival after primary liver resection or treatment has led to an increased frequency of recurrent liver disease and repeat liver resection is a potentially curative treatment for these recurrences. However, repeat liver resection involves technical challenges and a risk of complications, based on the presence of adhesions and anatomical modifications from the earlier surgery. Previous reports have indicated that the most common indication for ORLR is a regional recurrence of CRLM,[78] and an international multicentre analysis revealed that perioperative morbidity was similar after primary OLR and ORLR for CRLM.[9]

Pre-operative imaging improvements, technological innovations and increased surgeon experience have led to noticeable improvements in LLR during recent decades,[1011] although no randomised controlled trials have confirmed that this approach is superior to the open approach. Nevertheless, large propensity score-matching studies have revealed that LLR and OLR had equivalent oncological outcomes, but that LLR provided less blood loss, fewer post-operative complications and shorter hospital stays.[1213] A difficulty scoring system has also recently been created and used to clarify the indications for LLR based on the patient's pre-operative variables and the surgeon's skill level.[5] The initial concerns regarding LLR were that it would provide insufficient tumour resection and promote tumour seeding, although Belli et al. demonstrated that LLR had similar oncological outcomes to those of OLR, even after stratification for tumour characteristics.[14] In addition, other studies have confirmed that there were no differences in the rates of margin-free resections between the LLR and OLR groups.[1516]

Despite the existing evidence regarding the benefits of LLR, it remains unclear whether LRLR is suitable for the recurrent liver disease after previous curative liver resection. Shafaee et al.'s large study revealed that the optimal candidates for laparoscopic repeat resections are those with previous laparoscopic resections, rather than open resections.[17] In the present study, we examined the relative difficulties of LRLR after previous open resections versus LPLR, based on the difficulty scoring, and examined the related short-term post-operative outcomes. Our findings revealed an overall complication rate of 15% and a mortality rate of 0%, which are comparable to the findings from previous studies.[1819202122232425] Furthermore, the two groups had statistically equivalent short-term outcomes, and no statistical differences were observed in the analysis even when we stratified the cases based on low or intermediate difficulty. In the intermediate LRLR group, two patients developed bile leaks and we speculate that at least one bile leak occurred because of the scarring effect of previous local therapy, rather than pre-existing adhesions and anatomical changes.

Recent studies have also confirmed that LLR provides the short-term benefits of LLR without impaired long-term results. However, the oncological outcomes of LRLR have only been described in a few reports. For example, Shafaee et al. reported that the 3- and 5-year survival rates were 83% and 55%, respectively, after LRLR for patients with CRLM. In addition, other reports have revealed no significant differences in the 1-year recurrence rate (16.1% vs. 21.2%) and RFS rate for LRLR and ORLR,[26] and no significant differences in the 3-year OS rate (60% vs. 89.3%) and DFS rate (18.9% vs. 45.7%) when LRLR was compared to ORLR for patients with HCC.[27] In the present study, the comparison revealed no significant differences in RFS and OS when we compared LRLR and LPLR for HCC and CRLM. However, this cohort is relatively small and in the future, properly powered comparisons are needed to clarify whether LRLR does not compromise the oncological outcomes compared to LPLR.

The present study has several limitations. First, we only examined a limited number of cases and the high likelihood of Type II error possibly exists, which is even more pronounced in the subgroup analyses performed. For instance, for patients with CRLM, the two comparison groups had three and nine patients, respectively. Importantly, these data cannot be used to establish equivalence between the groups in the absence of a properly powered comparison. Furthermore, the 3-year RFS in the study group for HCC was 36% as opposed to 64% for the control group. Figure 1a shows visually consistent with a significant difference between the groups, with P = 0.118 likely secondary to a limited sample size. However, this study is similar to a preliminary report, and we should validate the conclusion by adding cases and prolonged follow-ups. Second, we excluded procedures that were more extensive than sectionectomy in the present study. In addition, we have not performed such major hepatectomy in patients who underwent previous OLR in our institute. Therefore, additional data are needed to determine if similar patterns are observed among patients who undergo major LRLR. Third, selection bias is present, as we select favourable cases for LRLR (i.e., no previous hemihepatectomy, ≤2 tumours, and no resection of other organs), which may partially explain the comparable outcomes in this study. Finally, comparing the oncologic outcomes of primary laparoscopic resections to repeat resections is not an appropriately controlled comparison, as the latter have, by definition, recurrent liver disease and therefore, are expected to have worse survival outcomes. Thus, we calculate the OS rate in two ways with regard to patients who underwent LRLR to minimise survival bias.

CONCLUSIONS

It appears that LRLR after OLR is relatively safe and effective when compared to LPLR in the present study. However, further studies with a larger cohort are needed to strengthen our findings.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.