- Email: [email protected]
Laparoscopic versus open resection of hepatic colorectal metastases
Laparoscopic versus open resection of hepatic colorectal metastases Robert M. Cannon, MD, Charles R. Scoggins, MD, MBA, Glenda G. Callender, MD, Kelly M. McMasters, MD, PhD, and Robert C. G. Martin II, MD, PhD, Louisville, KY
Background. This study was undertaken to assess the safety and efficacy of laparoscopic versus open resection of hepatic colorectal metastases (CRM). Methods. We reviewed retrospectively of all patients undergoing initial resection of CRM at a single institution between 1995 and 2010. The study cohort consisted of all patients undergoing laparoscopic resection and a cohort of patients undergoing open resection matched on a 4:1 basis by propensity scoring. Variables analyzed included patient and tumor characteristics, short-term outcomes, and OS and disease-free (DFS) survivals. Results. The 35 patients in the laparoscopic cohort and 140 patients in the open cohort were equivalent in terms of age, Charlson Comorbidity Index, tumor characteristics, and Clinical Risk Score. Similar proportions of patients in the laparoscopic and open groups underwent synchronous colectomy (9% in both; P = .976) and major hepatectomy (54% vs 51%; P = .705). Blood loss (202 vs 385 mL; P < .001), complications (23% vs 50%; P = .004), and duration of stay (4.8 vs 8.3 days; P < .001) were less in the laparoscopic cohort. Five-year OS (36% vs 42%; P = .818) and DFS (15% vs 22%; P = .346) were also similar in the laparoscopic and open groups. Conclusion. Laparoscopic resection of hepatic CRM seems to be a beneficial alternative to open surgery in appropriately selected patients. (Surgery 2012;152:567-74.) From the University of Louisville, Department of Surgery, Division of Surgical Oncology, Louisville, KY
SINCE THE FIRST REPORTS of limited laparoscopic liver resection in the first half of the 1990s, the application of minimally invasive techniques to hepatic surgery has been relatively slow.1 Acceptance of the technique in specialized centers has been gaining ground steadily since Cherqui’s initial report of 30 laparoscopic hepatectomies.2 At the time of publication in 2000, Cherqui’s series was the largest in the literature. A review by Nguyen et al in 2009 demonstrated 2,804 cases reported in the world literature.3 Some experienced centers are now performing the majority of liver resections in a minimally invasive fashion,4 and several reports of experience with >300 patients have now been published.4-6 For example, in a report from Koffron et al4 of 300 laparoscopic liver resections, it is noted that the percentage of hepatic resections in their center increased from 10% in 2002 to 80% in 2007. With respect to the indication for hepatectomy, laparoscopic resection for benign lesions gained Accepted for publication July 10, 2012. Reprint requests: Robert C. G. Martin II, MD, PhD, University of Louisville, Department of Surgery, Division of Surgical Oncology, 315 E. Broadway - #312, Louisville, KY 40202. E-mail: [email protected]. 0039-6060/$ - see front matter Ó 2012 Mosby, Inc. All rights reserved. http://dx.doi.org/10.1016/j.surg.2012.07.013
acceptance relatively early on.7,8 Enthusiasm for laparoscopic resection of malignant lesions seems to have been slower to develop, owing to concerns about the oncologic adequacy of laparoscopic hepatectomy.9 In colorectal cancer, laparoscopic resection of the primary tumor has been demonstrated by randomized, controlled trials to be oncologically equivalent to open surgery, with the benefit of shorter postoperative duration of stay.10 We demonstrated recently the significant benefits of laparoscopic hepatic lobectomy in comparison with open hepatic lobectomy with benefits of decreases in adverse events, incision-related adverse events, and in hospital stay11 in a study with a mix of benign and malignant indications for operation. There have now been a handful of retrospective reports demonstrating the equivalence of laparoscopic and open hepatic resection for colorectal metastases as well.12-14 This study was undertaken to evaluate the safety and efficacy of laparoscopic resection of hepatic colorectal metastases compared with a matched group of patients undergoing open resection. PATIENTS AND METHODS With Institutional Review Board approval, a retrospective review of all patients undergoing laparoscopic first resection of hepatic colorectal metastases by the division of surgical oncology at SURGERY 567
568 Cannon et al
the University of Louisville from 1995 to 2010 was undertaken. These patients were compared with a cohort of patients undergoing open resection during the same period who were matched on a 4:1 basis. Matching was by propensity scoring, with scores based on patient age, size and number of lesions, performance of major hepatectomy (resection of $3 Couinaud segments) or synchronous colectomy, and Clinical Risk Score.15 These variables were chosen empirically based on factors we believed to be important contributors to selection bias and differences in oncologic outcomes. Matching on the Clinical Risk Score in particular was thought to be important given its validation and extensive use in predicting recurrence after resection for colorectal metastasis. The other variables are thought to contribute to operative difficulty or risk of complications/mortality. More granular measures such as specific lesion location were not included, because this approach would make achieving adequate matching difficult. Briefly, a propensity score represents the probability that a patient underwent laparoscopic resection based on a logistic regression model including the factors outlined as covariates. Open controls with the closest propensity score (and a maximum difference in propensity score of no greater than 0.1) to each laparoscopic case were chosen as the comparison group. The purpose of propensity score matching is to achieve balance (equal distributions) on the covariates included in the model between the laparoscopic and open cases. Two propensity-matched groups were constructed. In the ‘‘inclusive’’ cohort, controls undergoing open resection were selected from the group of all patients undergoing open resection from 1995 through 2010. A ‘‘restricted’’ cohort was also created in which the controls undergoing open resection were selected from a group limited to those undergoing open resection from 2004 (when the first laparoscopic resection was performed) through 2010. The rationale for analyzing the inclusive and restricted cohorts separately is to limit time selection bias. Use of controls from the earlier time period in the inclusive cohort has the limitation that there have been improvements in patient care and adjuvant therapy over the past 2 decades. Thus, utilization of controls undergoing open operation before the introduction of laparoscopy may have introduced bias owing to the improved adjuvant therapy in the laparoscopic cohort, which represents a more recent group of patients. In contrast, using controls who underwent operation before the introduction of laparoscopy minimizes
Surgery October 2012
selection bias, in that even the ‘‘easy’’ cases underwent open resection before the introduction of laparoscopy. In the restricted cohort, the patients undergoing open and laparoscopic resection are contemporaneous, thus minimizing the effect of changes in adjuvant therapy. In contrast, the restricted cohort is subject to greater selection bias because the more difficult cases receive preferentially an open operation. Comparisons were made between the matched groups in terms of patient demographics, tumor characteristics, operative factors, short-term outcomes, and overall (OS) and disease-free (DFS) survival. Resection margins were defined as either microscopically positive (R1) or negative (R0). Baseline comorbidities were assessed using the Charlson Comorbidity Index (CCI).16 Complications were graded according our standard scale which has been published previously.17 Grade 1 and 2 complications required oral or intravenous medical management, respectively. Grade 3 complications required intervention, either with interventional radiology or reoperation, or an escalation of care (reintubation, transfer to intensive care, etc). Grade 4 complications caused chronic disability, additional organ resection, or enteric diversion. Grade 5 complications were fatal. Continuous variables were summarized as mean values (standard deviation) or median values (interquartile range) and analyzed using the Student t-test or the Wilcoxon Mann–Whitney test, where appropriate. Categorical variables were summarized as count (percentage) and compared using the chi-square or Fisher exact test, where appropriate. OS was calculated from the time of resection to death or last follow-up according to Kaplan–Meier; DFS was the time from resection to recurrence, death, or last follow-up. Differences in survival curves were compared using the logrank test or Cox proportional hazards regression. All statistical analyses were performed using SAS version 9.3 (SAS Institute, Cary, NC). P < .05 was considered significant. Continuous variable are presented as mean values ± standard deviations. RESULTS Analysis of the inclusive cohort. There were 35 patients in the laparoscopic and 140 in the open cohort. The groups were well matched on age, tumor size and number, body mass index, and CCI16 (Table I). Patients in the laparoscopic resection cohort were more likely to have abnormal liver parenchyma (chemotherapy-associated steatohepatitis) compared with patients undergoing open resection (80% vs 58%; P = .016). In terms of
Cannon et al 569
Surgery Volume 152, Number 4
Table I. Baseline patient and tumor characteristics in the matched laparoscopic and open cohorts Inclusive cohort Open Age (yrs), mean (SD) CCI, median (IQR) BMI (kg/m2), mean (SD) Size of largest tumor (cm), mean (SD) Number of lesions, median (IQR) CEA at time of liver resection (ng/mL), mean (SD) Node-positive primary tumor, n (%) Clinical Risk Score, median (IQR)
62 4 27 5 1 77
Laparoscopic
(11) (1) (6) (3) (1) (163)
62 4 28 4 1 52
83 (59) 2 (1)
Restricted cohort P value
(10) (1) (8) (3) (1) (126)
20 (57) 2 (2)
Open
.972 .855 .626 .609 .381 .399
62 4 28 4 1 92
(11) (1) (6) (2) (2) (181)
.818 .372
88 (64) 2 (2)
Laparoscopic 62 4 28 4 1 52
(10) (1) (8) (3) (1) (126)
20 (57) 2 (2)
P value .972 .378 .706 .810 .929 .137 .470 .861
The inclusive cohort includes patients from 1995 through 2010, while the restricted cohort includes patients from 2002 through 2010. Continuous variables are presented as mean values (standard deviation) or median values (interquartile range) where appropriate, and categorical variables as count (percentage). BMI, Body mass index; CCI, Charlson Comorbidity Index; CEA, carcinoembryonic antigen; IQR, interquartile range; SD, standard deviation.
Table II. Summary of operative procedures performed in each cohort Inclusive cohort, n (%) Procedure Left lateral segmentectomy Wedge/bisegmentectomy Right posterior sectionectomy Right hepatectomy Extended right hepatectomy Left hepatectomy Extended left hepatectomy
Laparoscopic 10 5 1 14 1 4 0
(29) (14) (3) (40) (3) (11) (0)
Open 10 52 7 37 14 17 3
operative procedure performed, 19 (54%) of patients in the laparoscopic cohort underwent major hepatectomy compared with 71 (51%) in the open cohort (P = .705).Synchronous colectomy was undertaken in 9% of both groups (P = .976). Comparison of specific hepatic resections in the laparoscopic and open groups is given in Table II. It should be noted that left lateral segmentectomy was significantly more common in the laparoscopic group, while wedge resection/bisegmentectomy was significantly more common in the open group (Table II). Patients in the laparoscopic group had less operative blood loss than those undergoing open resection (202 ± 180 vs 392 ± 322 mL; P < .001). Transfusions were required in 5 (17%) patients undergoing laparoscopic resection compared with 30 (25%; P = .334) in the open group. Of those who were transfused, the median number of units transfused in the laparoscopic group was 3 (range, 1–18) compared with 2 (range, 1–38) in the open group (P = .811). The complication rate was less in the laparoscopic group versus the open group (23% vs 50%; P = .004). Specific classes of complication
(7) (37) (5) (26) (10) (12) (2)
Restricted cohort, n (%) P value <.001 .001 .966 .114 .310 .966 .988
Laparoscopic 10 5 1 14 1 4 0
(29) (14) (3) (40) (3) (11) (0)
Open 9 48 4 42 15 16 4
(7) (35) (3) (30) (11) (12) (3)
P value .001 .019 .987 .280 .199 .996 .584
are listed in Table III. The median highest grade of complication was similar in both the laparoscopic and open resection groups (2.5 vs 2.0; P = .969). Duration of stay was an average of 4.8 vs 8.3 days (P < .001). Ninety-day mortality was 0% in the laparoscopic cohort and 1.4% (n = 2) in the open cohort (P = .984). In terms of oncologic outcome, R0 resection was achieved in 97% (n = 34) of laparoscopically resected patients compared with 86% (n = 121) of patients in the open cohort (P = .082). Perioperative chemotherapy was received in 89% of the laparoscopic group and in 84% of the open group (P = .502). OS at 1, 3, and 5 years, respectively, in the laparoscopic and open cohorts was 97%, 63%, and 36% vs 96%, 65%, and 42% (Fig 1, A; P = .818). DFS in the laparoscopic and open cohorts was 79%, 37%, and 15% vs 83%, 39%, and 22% at 1, 3, and 5 years, respectively (Fig 1, B; P = .346). There were no recurrences at laparoscopic port sites or at the laparotomy incision. Analysis of the restricted cohort. There were 35 patients undergoing laparoscopic resection and 138 matched patients in the open cohort. The
570 Cannon et al
Surgery October 2012
Table III. Frequency (number) of specific classes of complication in both the laparoscopic and open groups Inclusive
Restricted
Complication type Laparoscopic Open Laparoscopic Open Ileus Intra-abdominal infection Wound related Postoperative bleeding Bile leak Hepatic insufficiency Cardiac Pulmonary Miscellaneous Renal
2 1
3 4
2 1
3 4
0 1
9 4
0 1
9 4
0 0
5 2
0 0
5 2
1 1 0 2
7 10 18 5
1 1 0 2
8 10 17 6
groups were well matched on age, tumor size and number, body mass index, CCI,16 and Clinical Risk Score (Table III). Major hepatectomy was performed in 54% (n = 19) of patients undergoing laparoscopic resection compared with 55.8% (n = 77; P = .872) of patients undergoing open resection. A synchronous colectomy was performed in 9% (n = 3) of the laparoscopic cohort and 9% (n = 13; P = .976) of the open cohort. Liver parenchyma was abnormal in 28 (80%) patients undergoing laparoscopic resection compared with 92 (67%) patients undergoing open resection (P = .126). Comparison of specific hepatic resections in the laparoscopic and open groups is given in Table II. It should be noted that left lateral segmentectomy was more common in the laparoscopic group, while wedge resection/bisegmentectomy was more common in the open group (Table II). Patients in the laparoscopic group had less operative blood loss than those undergoing open resection (202 ± 180 vs 392 ± 324 mL; P < .001). Rates of blood transfusion were similar (17% vs 32%; P = .102). Of those who were transfused, the median number of units transfused in the laparoscopic group was 3 (range, 1–18) compared with 2 (range, 1–38) in the open group (P = .839). The complication rate in the laparoscopic group was less (23% vs 49%; P = .007). Specific classes of complication are listed in Table III. Of the patients who did experience a complication, the median highest grade of complication was similar in both the laparoscopic and open groups (2.5 vs 2.5; P = .861). Duration of stay was less in the laparoscopic cohort (4.8 vs 7.8 days; P < .001).
Fig 1. Comparison of (A) OS (in months) and (B) DFS in the matched open and laparoscopic groups of the inclusive cohort.
In terms of oncologic outcomes, patients undergoing laparoscopic resection were more likely to have a margin negative resection than patients in the open cohort (97% vs 81%; P = .020).Perioperative chemotherapy was received in 89% of the laparoscopic group and in 84% of the open group (P = .531). OS at 1, 3, and 5 years in the laparoscopic and open cohorts was 97%, 63%, and 36% vs 95%, 60%, and 37%, respectively (Fig 2, A; P = .911). DFS at 1, 3, and 5 years in the laparoscopic and open cohorts was 79%, 37%, and 15% vs 78%, 35%, and 22%, respectively (Fig 2, B; P = .715). Ninety-day mortality was 0% in the laparoscopic group and 0.7% (n = 1) in the open group (P = .957). DISCUSSION Results with open resection of hepatic colorectal metastases have been excellent, with several centers reporting 5-year survival in excess of 50%.18,19 The group at Memorial Sloan Kettering have reported 10-year cancer specific survival to
Surgery Volume 152, Number 4
Fig 2. Comparison of (A) OS (in months) and (B) DFS in the matched open and laparoscopic groups of the restricted cohort.
be 23%15 in a series of >1,000 patients. Given these excellent results, the bar has been set relatively high for the application of laparoscopic techniques to this disease. Specific concerns about the oncologic adequacy of laparoscopy in general include port site metastases, the trophic effect of pneumoperitoneum on malignant cells, and inability to inspect the peritoneal cavity adequately, and lack of tactile sensation when inspecting the liver.9,20-24 Experience has shown these fears to be largely unfounded. In the review by Nguyen et al3 of the world literature on laparoscopic liver resection, there were no port site recurrences reported after laparoscopic resection of hepatic colorectal metastases. In the current study, there were no port site recurrences. Randomized trials of laparoscopic colon resection for cancer also failed to demonstrate increased rates of wound recurrence with laparoscopy.25 As such, fears over port site recurrence seem to be unsupported by the available evidence.
Cannon et al 571
Rates of margin-negative resection in this series were also similar or better in the laparoscopic versus open cohorts, demonstrating that adequate delineation of intrahepatic tumor anatomy can be obtained during laparoscopic resection. The importance of facility with laparoscopic ultrasonography is a key factor in obtaining such results. The ability to perform intraoperative ultrasonography has been noted by consensus of an international group of hepatic surgery experts as one of the prerequisite skill sets necessary before embarking on a program of laparoscopic liver resection.8 The authors have no reason to believe that there are any factors intrinsic to laparoscopic resection that should lead to a greater negative margin resection rate when compared with open resection. Instead, the significant increase in margin-negative resection in the laparoscopic versus open group in the restricted cohort is more likely a function of residual bias rather than any superiority of laparoscopy in this aspect. Finally, OS and DFS were equivalent in the laparoscopic and open cohorts of this study. When combined with previously published findings by Abu Hilal et al,12 Mala et al,13 and Castaing et al,14 these results add further evidence to suggest that laparoscopic resection for hepatic colorectal metastasis is likely to be oncologically equivalent to open surgery. In addition to oncologic equivalency, we have also demonstrated a number of benefits to laparoscopic hepatectomy. Patients undergoing a minimally invasive procedure experienced less operative blood loss, and fewer postoperative complications at less than half the rate of their counterparts in the open cohort. Furthermore, duration of stay was less in the laparoscopic cohorts. Another potential benefit of laparoscopic liver resection is for patients who present with synchronous disease. Simultaneous resection of both the primary tumor and hepatic metastases allows for less overall hospitalization by avoiding a second operation26 and eliminates any delay in hepatectomy while patients receive adjuvant chemotherapy. The ability to perform laparoscopic hepatectomy extends the benefits of minimally invasive surgery to this patient population. These results confirm our prior study evaluating laparoscopic hepatic lobectomy versus open hepatic lobectomy in which there were fewer adverse events.11 Differences between our prior study and the current study are that only lobar resections were included in the earlier analysis, and the indication for operation was a mixture of various benign and malignant indications. The current study adds additional information to the previously obtained
572 Cannon et al
results by extending the results to patients operated on solely for colorectal metastasis. There is some overlap in the 2 studies, in that any patients in the current study undergoing lobar resection would have been included in the previous study as well. Whereas the first study was a comparison based on the operation performed, the current study represents a comparison based on indication for operation. The most important aspect to the safe development of a laparoscopic liver resection program, in our opinion, is proper patient selection. What constitutes proper patient selection varies depending on the skill and experience of the individual surgeon. For those just beginning a laparoscopic liver program, we recommend limiting early experience to benign lesions located in the periphery of the liver. After obtaining more experience, more extensive resections such as left lateral segmentectomy can be approached. The last area for a laparoscopic liver surgeon to approach should be malignant neoplasms and those lesions requiring formal hepatic lobectomy or even trisegmentectomy. At present, our center has no absolute inclusion criteria for laparoscopic versus open hepatic resection, although we do exclude lesions requiring vascular/biliary reconstruction and caudate resections. In general, our preference is to reserve open resection for centrally located lesions, such as those at the base of segments 4 or 5 at the hilum or in proximity to the hepatic veins. There is some variability between the individual surgeons within our group, however, as to which lesions are appropriate for laparoscopic resection and which would be better approached via open laparotomy. Given the demonstrated patient benefits of laparoscopy across the field of surgery, it is unlikely that a randomized, controlled trial of laparoscopic versus open hepatic resection will ever be undertaken. There are also substantial practical obstacles to performing a randomized trial. For example, to achieve 80% power to detect a 1% difference in mortality between a laparoscopic and open operation would require 2,319 patients per arm. Such a study would likely require a decade to accrue enough patients, if it were even possible. Another major hurdle to performing a randomized trial is that patients may be unwilling to be randomized to an open operation when the existing data suggest overwhelmingly that laparoscopic resection is safe. Thus, comparisons will be limited to observational studies that are potentially confounded by selection bias. A strength of the current study is the utilization of propensity score based matching,
Surgery October 2012
which simulates randomization in eliminating the confounding effects of variables used to create the model. In utilizing this design, the treatment effect of laparoscopy versus open resection in this study is more accurately estimated.27 There remains the potential, however, that residual confounding by unmeasured variables continues to exist. Another limitation of this study is its relatively small sample size, with only 35 patients undergoing laparoscopic resection. Despite these drawbacks, our work suggests that laparoscopic resection is an effective and beneficial alternative to open resection for appropriately selected patients with hepatic colorectal metastases. Further study is needed as experience increases to confirm these findings. REFERENCES 1. Lefor AT, Flowers JL. Laparoscopic wedge biopsy of the liver. J Am Coll Surg 1994;178:307-8. 2. Cherqui D, Husson E, Hammoud R, et al. Laparoscopic liver resections: a feasibility study in 30 patients. Ann Surg 2000;232:753-62. 3. Nguyen KT, Gamblin TC, Geller DA. World review of laparoscopic liver resection---2,804 patients. Ann Surg 2009; 250:831-41. 4. Koffron AJ, Auffenberg G, Kung R, Abecassis M. Evaluation of 300 minimally invasive liver resections at a single institution: less is more. Ann Surg 2007;246:385-92. 5. Cannon RM, Brock GN, Marvin MR, Buell JF. Laparoscopic liver resection: an examination of our first 300 patients. J Am Coll Surg 2011;213:501-7. 6. Nguyen KT, Marsh JW, Tsung A, Steel JJ, Gamblin TC, Geller DA. Comparative benefits of laparoscopic vs open hepatic resection: a critical appraisal. Arch Surg 2011;146: 348-56. 7. Ardito F, Tayar C, Laurent A, Karoui M, Loriau J, Cherqui D. Laparoscopic liver resection for benign disease. Arch Surg 2007;142:1188-93. 8. Buell JF, Cherqui D, Geller DA, et al. The international position on laparoscopic liver surgery: the Louisville Statement, 2008. Ann Surg 2009;250:825-30. 9. Gagner M, Rogula T, Selzer D. Laparoscopic liver resection: benefits and controversies. Surg Clin North Am 2004;84:451-62. 10. Koopmann MC, Heise CP. Laparoscopic and minimally invasive resection of malignant colorectal disease. Surg Clin North Am 2008;88:1047-72. 11. Martin RC, Scoggins CR, McMasters KM. Laparoscopic hepatic lobectomy: advantages of a minimally invasive approach. J Am Coll Surg 2010;210:627-34, 634-6. 12. Abu HM, Underwood T, Zuccaro M, Primrose J, Pearce N. Short- and medium-term results of totally laparoscopic resection for colorectal liver metastases. Br J Surg 2010;97:927-33. 13. Mala T, Edwin B, Gladhaug I, et al. A comparative study of the short-term outcome following open and laparoscopic liver resection of colorectal metastases. Surg Endosc 2002; 16:1059-63. 14. Castaing D, Vibert E, Ricca L, Azoulay D, Adam R, Gayet B. Oncologic results of laparoscopic versus open hepatectomy for colorectal liver metastases in two specialized centers. Ann Surg 2009;250:849-55. 15. Fong Y, Fortner J, Sun RL, Brennan MF, Blumgart LH. Clinical score for predicting recurrence after hepatic resection
Surgery Volume 152, Number 4
16.
17.
18.
19.
20.
21.
22.
23.
24. 25.
26.
27.
for metastatic colorectal cancer: analysis of 1001 consecutive cases. Ann Surg 1999;230:309-18. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373-83. Martin RC, Jarnagin WR, Fong Y, Biernacki P, Blumgart LH, DeMatteo RP. The use of fresh frozen plasma after major hepatic resection for colorectal metastasis: is there a standard for transfusion? J Am Coll Surg 2003;196:402-9. House MG, Ito H, Gonen M, et al. Survival after hepatic resection for metastatic colorectal cancer: trends in outcomes for 1,600 patients during two decades at a single institution. J Am Coll Surg 2010;210:744-5. Choti MA, Sitzmann JV, Tiburi MF, et al. Trends in longterm survival following liver resection for hepatic colorectal metastases. Ann Surg 2002;235:759-66. Johnstone PA, Rohde DC, Swartz SE, Fetter JE, Wexner SD. Port site recurrences after laparoscopic and thoracoscopic procedures in malignancy. J Clin Oncol 1996;14: 1950-6. Paolucci V, Schaeff B, Schneider M, Gutt C. Tumor seeding following laparoscopy: international survey. World J Surg 1999;23:989-95. Ishida H, Murata N, Yamada H, et al. Effect of CO pneumoperitoneum on growth of liver micrometastases in a rabbit model. World J Surg 2000;24:1004-8. Gutt CN, Riemer V, Kim ZG, Erceg J, Lorenz M. Impact of laparoscopic surgery on experimental hepatic metastases. Br J Surg 2001;88:371-5. Hsu TC. Intra-abdominal lesions could be missed by inadequate laparoscopy. Am Surg 2008;74:824-6. Lacy AM, Garcia-Valdecasas JC, Delgado S, et al. Laparoscopy-assisted colectomy versus open colectomy for treatment of non-metastatic colon cancer: a randomised trial. Lancet 2002;359:2224-9. Martin RC, Augenstein V, Reuter NP, Scoggins CR, McMasters KM. Simultaneous versus staged resection for synchronous colorectal cancer liver metastases. J Am Coll Surg 2009;208:842-50. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:41-55.
DISCUSSION Dr Sharon Weber (Madison, WI): Thank you so much for an excellent presentation. Clearly, this is a topic of interest for all hepatobiliary surgeons, and the great group at Louisville is really the leaders in this field. Really, this paper reiterates findings of prior investigators and your own group with the perioperative outcomes, but really extended to these patients that have colorectal metastases, and it has similar benefits for the short term and then for the long term. I have a couple of questions for you about what you started to discuss about who are candidates for the approach. I think even as we get more aggressive in utilizing MIS approach for liver resection, there will always be candidates who really are not going to be candidates for the approach, those that have portal vein embolization, if they have extensive disease, if they have tumor close to major vessels.
Cannon et al 573
So I think part of this really affects the finding and explains the finding that your margin positivity was higher in the open groups. And as you said, I think that this really shows that there still is some selection bias. So the question is, do you think that we can actually compare these groups statistically? Or do we just say for those patients who are candidates for laparoscopic resection, they tend to have better outcomes than the ones that you mentioned? The second question is how you chose the factors that went into your propensity scoring. And I think some would consider other factors equally relevant, particularly the use of neoadjuvant therapy, bilateral disease, underlying liver fibrosis or cirrhosis. And I think the Clinical Risk Score, there’s a suggestion by the group from The Netherlands that maybe in the era of contemporary neoadjuvant therapy that the Clinical Risk Score has become less relevant, so I just wonder why you included that. Dr Robert Cannon (Louisville, KY): Yeah, I agree that the 2 groups will probably never be strictly comparable. There’s always going to be a role for open operations, especially for the lesions you mentioned, especially those that may require biliary reconstruction or resection. Those that are probably centrally located tumors, such as like at the base of segment 4 or 5, those are ones that will really be best for open. And there’s always going to be a role for both. So, yeah, that’s the main idea, is that in patients who are candidates for laparoscopy, those are the benefits here. And, really, the purpose of doing match studies is just to show that if have a patient that could maybe go either way, that perhaps the laparoscopic procedure does offer these benefits. But truly, they’re not apples to apples. It is apples to oranges. And we really just want to say is that laparoscopy, I think, is safe and feasible in the right hands with wellselected patients and it does offer these benefits. In terms of what we included in the risk score, mainly a lot of that had to do with data that was available. I just didn’t necessarily feel that we had strong enough data as far as the type of regimens and the like that we could put together, and also just the ability to actually obtain a match cohort. We really would have liked to have matched on what segment within the liver is the lesion, and the like, but I think it would be more difficult to actually get a matched cohort for comparison there, when, really, you’re never going to have perfectly comparable groups anyway. Dr Mark Carlson (Omaha, NE): Two quick questions. You’ve already addressed this some, but how do you feel about these data? Do you think they’re strong enough to say that, for some cases, these resections should be done laparoscopically, or is it just an option at this point? The second question is, what about trainees? How do you train somebody to do this procedure? Do you do a lap first, open first, both at the same time? How would you train someone? Dr Robert Cannon: I’ll start with the training first, because I think this is an area where, really, we’re seeing a
574 Cannon et al
paradigm shift. The traditional model is you would have to have open expertise first, and then you would go back and get the additional laparoscopic training for this. And this is mainly because people who invented this procedure initially came up in the open surgery era and then came back and developed a laparoscopic procedure later. Now that we are seeing advanced laparoscopy more integrated into just a regular general surgery training, I think we’re probably going to see surgeons trained to do this both open and laparoscopically probably more at the same time. I don’t think we’re quite ready to say that laparoscopy is the standard and should be done. Now, I think if you are at a center where you have that expertise and you’re comfortable with it, I think your patients are going to want you to do it. But I don’t think we’re at the point where you can really say you are violating a standard of care to do an open operation at this point. It’s still very viable. But these are these benefits out there and I think it is a very good option, especially if you develop that expertise. Dr Scott Gruber (Detroit, MI): I didn’t see any comparison of the total cost of the hospital stay or of the operative time. And maybe that may be apples and oranges because of the number of left lateral segmentectomies that were in the laparoscopic group. But still, I think it would be nice to say something about that. Dr Robert Cannon: I think that is a great question to ask, especially because there’s a concern you’re going to
Surgery October 2012
have greater instrument cost, perhaps, with a laparoscopic case. And whether or not that’s balanced by the length of stay, that’s a different question. We didn’t analyze it for this paper; however, we will be presenting that data at the SSO later this month, so if would you like to come and see us then, we will be happy to address it. Dr Gerald Fried (Montreal, Quebec, Canada): During the period involved in the study, you extracted your matched open. Did you control for the year in which the procedure was done? Because I would assume that you probably did increasing proportion of your cases over time laparoscopically. Of course, at the same time, there’s other innovations, new technology, et cetera, hemostatic agents. So I wonder if you put that in your model. Dr Robert Cannon: We did not put the actual year in the model. In the manuscript, we actually looked at 2 different open cohorts. We actually took an older cohort from 1995 on as a separate comparison, just because we think selection bias may be less there, because before laparoscopy came around, obviously, everyone got open. So those might have been more anatomically suitable or similar lesions to the laparoscopic group. But then the way we controlled for year was essentially just limiting to 2004 on. So we assumed that, as they were going on, they were going to have similar adjuvant therapy regimens and the like, so we didn’t actually control for year itself, just in the limitation to 2004 on is how we controlled for that.
Background. This study was undertaken to assess the safety and efficacy of laparoscopic versus open resection of hepatic colorectal metastases (CRM). Methods. We reviewed retrospectively of all patients undergoing initial resection of CRM at a single institution between 1995 and 2010. The study cohort consisted of all patients undergoing laparoscopic resection and a cohort of patients undergoing open resection matched on a 4:1 basis by propensity scoring. Variables analyzed included patient and tumor characteristics, short-term outcomes, and OS and disease-free (DFS) survivals. Results. The 35 patients in the laparoscopic cohort and 140 patients in the open cohort were equivalent in terms of age, Charlson Comorbidity Index, tumor characteristics, and Clinical Risk Score. Similar proportions of patients in the laparoscopic and open groups underwent synchronous colectomy (9% in both; P = .976) and major hepatectomy (54% vs 51%; P = .705). Blood loss (202 vs 385 mL; P < .001), complications (23% vs 50%; P = .004), and duration of stay (4.8 vs 8.3 days; P < .001) were less in the laparoscopic cohort. Five-year OS (36% vs 42%; P = .818) and DFS (15% vs 22%; P = .346) were also similar in the laparoscopic and open groups. Conclusion. Laparoscopic resection of hepatic CRM seems to be a beneficial alternative to open surgery in appropriately selected patients. (Surgery 2012;152:567-74.) From the University of Louisville, Department of Surgery, Division of Surgical Oncology, Louisville, KY
SINCE THE FIRST REPORTS of limited laparoscopic liver resection in the first half of the 1990s, the application of minimally invasive techniques to hepatic surgery has been relatively slow.1 Acceptance of the technique in specialized centers has been gaining ground steadily since Cherqui’s initial report of 30 laparoscopic hepatectomies.2 At the time of publication in 2000, Cherqui’s series was the largest in the literature. A review by Nguyen et al in 2009 demonstrated 2,804 cases reported in the world literature.3 Some experienced centers are now performing the majority of liver resections in a minimally invasive fashion,4 and several reports of experience with >300 patients have now been published.4-6 For example, in a report from Koffron et al4 of 300 laparoscopic liver resections, it is noted that the percentage of hepatic resections in their center increased from 10% in 2002 to 80% in 2007. With respect to the indication for hepatectomy, laparoscopic resection for benign lesions gained Accepted for publication July 10, 2012. Reprint requests: Robert C. G. Martin II, MD, PhD, University of Louisville, Department of Surgery, Division of Surgical Oncology, 315 E. Broadway - #312, Louisville, KY 40202. E-mail: [email protected]. 0039-6060/$ - see front matter Ó 2012 Mosby, Inc. All rights reserved. http://dx.doi.org/10.1016/j.surg.2012.07.013
acceptance relatively early on.7,8 Enthusiasm for laparoscopic resection of malignant lesions seems to have been slower to develop, owing to concerns about the oncologic adequacy of laparoscopic hepatectomy.9 In colorectal cancer, laparoscopic resection of the primary tumor has been demonstrated by randomized, controlled trials to be oncologically equivalent to open surgery, with the benefit of shorter postoperative duration of stay.10 We demonstrated recently the significant benefits of laparoscopic hepatic lobectomy in comparison with open hepatic lobectomy with benefits of decreases in adverse events, incision-related adverse events, and in hospital stay11 in a study with a mix of benign and malignant indications for operation. There have now been a handful of retrospective reports demonstrating the equivalence of laparoscopic and open hepatic resection for colorectal metastases as well.12-14 This study was undertaken to evaluate the safety and efficacy of laparoscopic resection of hepatic colorectal metastases compared with a matched group of patients undergoing open resection. PATIENTS AND METHODS With Institutional Review Board approval, a retrospective review of all patients undergoing laparoscopic first resection of hepatic colorectal metastases by the division of surgical oncology at SURGERY 567
568 Cannon et al
the University of Louisville from 1995 to 2010 was undertaken. These patients were compared with a cohort of patients undergoing open resection during the same period who were matched on a 4:1 basis. Matching was by propensity scoring, with scores based on patient age, size and number of lesions, performance of major hepatectomy (resection of $3 Couinaud segments) or synchronous colectomy, and Clinical Risk Score.15 These variables were chosen empirically based on factors we believed to be important contributors to selection bias and differences in oncologic outcomes. Matching on the Clinical Risk Score in particular was thought to be important given its validation and extensive use in predicting recurrence after resection for colorectal metastasis. The other variables are thought to contribute to operative difficulty or risk of complications/mortality. More granular measures such as specific lesion location were not included, because this approach would make achieving adequate matching difficult. Briefly, a propensity score represents the probability that a patient underwent laparoscopic resection based on a logistic regression model including the factors outlined as covariates. Open controls with the closest propensity score (and a maximum difference in propensity score of no greater than 0.1) to each laparoscopic case were chosen as the comparison group. The purpose of propensity score matching is to achieve balance (equal distributions) on the covariates included in the model between the laparoscopic and open cases. Two propensity-matched groups were constructed. In the ‘‘inclusive’’ cohort, controls undergoing open resection were selected from the group of all patients undergoing open resection from 1995 through 2010. A ‘‘restricted’’ cohort was also created in which the controls undergoing open resection were selected from a group limited to those undergoing open resection from 2004 (when the first laparoscopic resection was performed) through 2010. The rationale for analyzing the inclusive and restricted cohorts separately is to limit time selection bias. Use of controls from the earlier time period in the inclusive cohort has the limitation that there have been improvements in patient care and adjuvant therapy over the past 2 decades. Thus, utilization of controls undergoing open operation before the introduction of laparoscopy may have introduced bias owing to the improved adjuvant therapy in the laparoscopic cohort, which represents a more recent group of patients. In contrast, using controls who underwent operation before the introduction of laparoscopy minimizes
Surgery October 2012
selection bias, in that even the ‘‘easy’’ cases underwent open resection before the introduction of laparoscopy. In the restricted cohort, the patients undergoing open and laparoscopic resection are contemporaneous, thus minimizing the effect of changes in adjuvant therapy. In contrast, the restricted cohort is subject to greater selection bias because the more difficult cases receive preferentially an open operation. Comparisons were made between the matched groups in terms of patient demographics, tumor characteristics, operative factors, short-term outcomes, and overall (OS) and disease-free (DFS) survival. Resection margins were defined as either microscopically positive (R1) or negative (R0). Baseline comorbidities were assessed using the Charlson Comorbidity Index (CCI).16 Complications were graded according our standard scale which has been published previously.17 Grade 1 and 2 complications required oral or intravenous medical management, respectively. Grade 3 complications required intervention, either with interventional radiology or reoperation, or an escalation of care (reintubation, transfer to intensive care, etc). Grade 4 complications caused chronic disability, additional organ resection, or enteric diversion. Grade 5 complications were fatal. Continuous variables were summarized as mean values (standard deviation) or median values (interquartile range) and analyzed using the Student t-test or the Wilcoxon Mann–Whitney test, where appropriate. Categorical variables were summarized as count (percentage) and compared using the chi-square or Fisher exact test, where appropriate. OS was calculated from the time of resection to death or last follow-up according to Kaplan–Meier; DFS was the time from resection to recurrence, death, or last follow-up. Differences in survival curves were compared using the logrank test or Cox proportional hazards regression. All statistical analyses were performed using SAS version 9.3 (SAS Institute, Cary, NC). P < .05 was considered significant. Continuous variable are presented as mean values ± standard deviations. RESULTS Analysis of the inclusive cohort. There were 35 patients in the laparoscopic and 140 in the open cohort. The groups were well matched on age, tumor size and number, body mass index, and CCI16 (Table I). Patients in the laparoscopic resection cohort were more likely to have abnormal liver parenchyma (chemotherapy-associated steatohepatitis) compared with patients undergoing open resection (80% vs 58%; P = .016). In terms of
Cannon et al 569
Surgery Volume 152, Number 4
Table I. Baseline patient and tumor characteristics in the matched laparoscopic and open cohorts Inclusive cohort Open Age (yrs), mean (SD) CCI, median (IQR) BMI (kg/m2), mean (SD) Size of largest tumor (cm), mean (SD) Number of lesions, median (IQR) CEA at time of liver resection (ng/mL), mean (SD) Node-positive primary tumor, n (%) Clinical Risk Score, median (IQR)
62 4 27 5 1 77
Laparoscopic
(11) (1) (6) (3) (1) (163)
62 4 28 4 1 52
83 (59) 2 (1)
Restricted cohort P value
(10) (1) (8) (3) (1) (126)
20 (57) 2 (2)
Open
.972 .855 .626 .609 .381 .399
62 4 28 4 1 92
(11) (1) (6) (2) (2) (181)
.818 .372
88 (64) 2 (2)
Laparoscopic 62 4 28 4 1 52
(10) (1) (8) (3) (1) (126)
20 (57) 2 (2)
P value .972 .378 .706 .810 .929 .137 .470 .861
The inclusive cohort includes patients from 1995 through 2010, while the restricted cohort includes patients from 2002 through 2010. Continuous variables are presented as mean values (standard deviation) or median values (interquartile range) where appropriate, and categorical variables as count (percentage). BMI, Body mass index; CCI, Charlson Comorbidity Index; CEA, carcinoembryonic antigen; IQR, interquartile range; SD, standard deviation.
Table II. Summary of operative procedures performed in each cohort Inclusive cohort, n (%) Procedure Left lateral segmentectomy Wedge/bisegmentectomy Right posterior sectionectomy Right hepatectomy Extended right hepatectomy Left hepatectomy Extended left hepatectomy
Laparoscopic 10 5 1 14 1 4 0
(29) (14) (3) (40) (3) (11) (0)
Open 10 52 7 37 14 17 3
operative procedure performed, 19 (54%) of patients in the laparoscopic cohort underwent major hepatectomy compared with 71 (51%) in the open cohort (P = .705).Synchronous colectomy was undertaken in 9% of both groups (P = .976). Comparison of specific hepatic resections in the laparoscopic and open groups is given in Table II. It should be noted that left lateral segmentectomy was significantly more common in the laparoscopic group, while wedge resection/bisegmentectomy was significantly more common in the open group (Table II). Patients in the laparoscopic group had less operative blood loss than those undergoing open resection (202 ± 180 vs 392 ± 322 mL; P < .001). Transfusions were required in 5 (17%) patients undergoing laparoscopic resection compared with 30 (25%; P = .334) in the open group. Of those who were transfused, the median number of units transfused in the laparoscopic group was 3 (range, 1–18) compared with 2 (range, 1–38) in the open group (P = .811). The complication rate was less in the laparoscopic group versus the open group (23% vs 50%; P = .004). Specific classes of complication
(7) (37) (5) (26) (10) (12) (2)
Restricted cohort, n (%) P value <.001 .001 .966 .114 .310 .966 .988
Laparoscopic 10 5 1 14 1 4 0
(29) (14) (3) (40) (3) (11) (0)
Open 9 48 4 42 15 16 4
(7) (35) (3) (30) (11) (12) (3)
P value .001 .019 .987 .280 .199 .996 .584
are listed in Table III. The median highest grade of complication was similar in both the laparoscopic and open resection groups (2.5 vs 2.0; P = .969). Duration of stay was an average of 4.8 vs 8.3 days (P < .001). Ninety-day mortality was 0% in the laparoscopic cohort and 1.4% (n = 2) in the open cohort (P = .984). In terms of oncologic outcome, R0 resection was achieved in 97% (n = 34) of laparoscopically resected patients compared with 86% (n = 121) of patients in the open cohort (P = .082). Perioperative chemotherapy was received in 89% of the laparoscopic group and in 84% of the open group (P = .502). OS at 1, 3, and 5 years, respectively, in the laparoscopic and open cohorts was 97%, 63%, and 36% vs 96%, 65%, and 42% (Fig 1, A; P = .818). DFS in the laparoscopic and open cohorts was 79%, 37%, and 15% vs 83%, 39%, and 22% at 1, 3, and 5 years, respectively (Fig 1, B; P = .346). There were no recurrences at laparoscopic port sites or at the laparotomy incision. Analysis of the restricted cohort. There were 35 patients undergoing laparoscopic resection and 138 matched patients in the open cohort. The
570 Cannon et al
Surgery October 2012
Table III. Frequency (number) of specific classes of complication in both the laparoscopic and open groups Inclusive
Restricted
Complication type Laparoscopic Open Laparoscopic Open Ileus Intra-abdominal infection Wound related Postoperative bleeding Bile leak Hepatic insufficiency Cardiac Pulmonary Miscellaneous Renal
2 1
3 4
2 1
3 4
0 1
9 4
0 1
9 4
0 0
5 2
0 0
5 2
1 1 0 2
7 10 18 5
1 1 0 2
8 10 17 6
groups were well matched on age, tumor size and number, body mass index, CCI,16 and Clinical Risk Score (Table III). Major hepatectomy was performed in 54% (n = 19) of patients undergoing laparoscopic resection compared with 55.8% (n = 77; P = .872) of patients undergoing open resection. A synchronous colectomy was performed in 9% (n = 3) of the laparoscopic cohort and 9% (n = 13; P = .976) of the open cohort. Liver parenchyma was abnormal in 28 (80%) patients undergoing laparoscopic resection compared with 92 (67%) patients undergoing open resection (P = .126). Comparison of specific hepatic resections in the laparoscopic and open groups is given in Table II. It should be noted that left lateral segmentectomy was more common in the laparoscopic group, while wedge resection/bisegmentectomy was more common in the open group (Table II). Patients in the laparoscopic group had less operative blood loss than those undergoing open resection (202 ± 180 vs 392 ± 324 mL; P < .001). Rates of blood transfusion were similar (17% vs 32%; P = .102). Of those who were transfused, the median number of units transfused in the laparoscopic group was 3 (range, 1–18) compared with 2 (range, 1–38) in the open group (P = .839). The complication rate in the laparoscopic group was less (23% vs 49%; P = .007). Specific classes of complication are listed in Table III. Of the patients who did experience a complication, the median highest grade of complication was similar in both the laparoscopic and open groups (2.5 vs 2.5; P = .861). Duration of stay was less in the laparoscopic cohort (4.8 vs 7.8 days; P < .001).
Fig 1. Comparison of (A) OS (in months) and (B) DFS in the matched open and laparoscopic groups of the inclusive cohort.
In terms of oncologic outcomes, patients undergoing laparoscopic resection were more likely to have a margin negative resection than patients in the open cohort (97% vs 81%; P = .020).Perioperative chemotherapy was received in 89% of the laparoscopic group and in 84% of the open group (P = .531). OS at 1, 3, and 5 years in the laparoscopic and open cohorts was 97%, 63%, and 36% vs 95%, 60%, and 37%, respectively (Fig 2, A; P = .911). DFS at 1, 3, and 5 years in the laparoscopic and open cohorts was 79%, 37%, and 15% vs 78%, 35%, and 22%, respectively (Fig 2, B; P = .715). Ninety-day mortality was 0% in the laparoscopic group and 0.7% (n = 1) in the open group (P = .957). DISCUSSION Results with open resection of hepatic colorectal metastases have been excellent, with several centers reporting 5-year survival in excess of 50%.18,19 The group at Memorial Sloan Kettering have reported 10-year cancer specific survival to
Surgery Volume 152, Number 4
Fig 2. Comparison of (A) OS (in months) and (B) DFS in the matched open and laparoscopic groups of the restricted cohort.
be 23%15 in a series of >1,000 patients. Given these excellent results, the bar has been set relatively high for the application of laparoscopic techniques to this disease. Specific concerns about the oncologic adequacy of laparoscopy in general include port site metastases, the trophic effect of pneumoperitoneum on malignant cells, and inability to inspect the peritoneal cavity adequately, and lack of tactile sensation when inspecting the liver.9,20-24 Experience has shown these fears to be largely unfounded. In the review by Nguyen et al3 of the world literature on laparoscopic liver resection, there were no port site recurrences reported after laparoscopic resection of hepatic colorectal metastases. In the current study, there were no port site recurrences. Randomized trials of laparoscopic colon resection for cancer also failed to demonstrate increased rates of wound recurrence with laparoscopy.25 As such, fears over port site recurrence seem to be unsupported by the available evidence.
Cannon et al 571
Rates of margin-negative resection in this series were also similar or better in the laparoscopic versus open cohorts, demonstrating that adequate delineation of intrahepatic tumor anatomy can be obtained during laparoscopic resection. The importance of facility with laparoscopic ultrasonography is a key factor in obtaining such results. The ability to perform intraoperative ultrasonography has been noted by consensus of an international group of hepatic surgery experts as one of the prerequisite skill sets necessary before embarking on a program of laparoscopic liver resection.8 The authors have no reason to believe that there are any factors intrinsic to laparoscopic resection that should lead to a greater negative margin resection rate when compared with open resection. Instead, the significant increase in margin-negative resection in the laparoscopic versus open group in the restricted cohort is more likely a function of residual bias rather than any superiority of laparoscopy in this aspect. Finally, OS and DFS were equivalent in the laparoscopic and open cohorts of this study. When combined with previously published findings by Abu Hilal et al,12 Mala et al,13 and Castaing et al,14 these results add further evidence to suggest that laparoscopic resection for hepatic colorectal metastasis is likely to be oncologically equivalent to open surgery. In addition to oncologic equivalency, we have also demonstrated a number of benefits to laparoscopic hepatectomy. Patients undergoing a minimally invasive procedure experienced less operative blood loss, and fewer postoperative complications at less than half the rate of their counterparts in the open cohort. Furthermore, duration of stay was less in the laparoscopic cohorts. Another potential benefit of laparoscopic liver resection is for patients who present with synchronous disease. Simultaneous resection of both the primary tumor and hepatic metastases allows for less overall hospitalization by avoiding a second operation26 and eliminates any delay in hepatectomy while patients receive adjuvant chemotherapy. The ability to perform laparoscopic hepatectomy extends the benefits of minimally invasive surgery to this patient population. These results confirm our prior study evaluating laparoscopic hepatic lobectomy versus open hepatic lobectomy in which there were fewer adverse events.11 Differences between our prior study and the current study are that only lobar resections were included in the earlier analysis, and the indication for operation was a mixture of various benign and malignant indications. The current study adds additional information to the previously obtained
572 Cannon et al
results by extending the results to patients operated on solely for colorectal metastasis. There is some overlap in the 2 studies, in that any patients in the current study undergoing lobar resection would have been included in the previous study as well. Whereas the first study was a comparison based on the operation performed, the current study represents a comparison based on indication for operation. The most important aspect to the safe development of a laparoscopic liver resection program, in our opinion, is proper patient selection. What constitutes proper patient selection varies depending on the skill and experience of the individual surgeon. For those just beginning a laparoscopic liver program, we recommend limiting early experience to benign lesions located in the periphery of the liver. After obtaining more experience, more extensive resections such as left lateral segmentectomy can be approached. The last area for a laparoscopic liver surgeon to approach should be malignant neoplasms and those lesions requiring formal hepatic lobectomy or even trisegmentectomy. At present, our center has no absolute inclusion criteria for laparoscopic versus open hepatic resection, although we do exclude lesions requiring vascular/biliary reconstruction and caudate resections. In general, our preference is to reserve open resection for centrally located lesions, such as those at the base of segments 4 or 5 at the hilum or in proximity to the hepatic veins. There is some variability between the individual surgeons within our group, however, as to which lesions are appropriate for laparoscopic resection and which would be better approached via open laparotomy. Given the demonstrated patient benefits of laparoscopy across the field of surgery, it is unlikely that a randomized, controlled trial of laparoscopic versus open hepatic resection will ever be undertaken. There are also substantial practical obstacles to performing a randomized trial. For example, to achieve 80% power to detect a 1% difference in mortality between a laparoscopic and open operation would require 2,319 patients per arm. Such a study would likely require a decade to accrue enough patients, if it were even possible. Another major hurdle to performing a randomized trial is that patients may be unwilling to be randomized to an open operation when the existing data suggest overwhelmingly that laparoscopic resection is safe. Thus, comparisons will be limited to observational studies that are potentially confounded by selection bias. A strength of the current study is the utilization of propensity score based matching,
Surgery October 2012
which simulates randomization in eliminating the confounding effects of variables used to create the model. In utilizing this design, the treatment effect of laparoscopy versus open resection in this study is more accurately estimated.27 There remains the potential, however, that residual confounding by unmeasured variables continues to exist. Another limitation of this study is its relatively small sample size, with only 35 patients undergoing laparoscopic resection. Despite these drawbacks, our work suggests that laparoscopic resection is an effective and beneficial alternative to open resection for appropriately selected patients with hepatic colorectal metastases. Further study is needed as experience increases to confirm these findings. REFERENCES 1. Lefor AT, Flowers JL. Laparoscopic wedge biopsy of the liver. J Am Coll Surg 1994;178:307-8. 2. Cherqui D, Husson E, Hammoud R, et al. Laparoscopic liver resections: a feasibility study in 30 patients. Ann Surg 2000;232:753-62. 3. Nguyen KT, Gamblin TC, Geller DA. World review of laparoscopic liver resection---2,804 patients. Ann Surg 2009; 250:831-41. 4. Koffron AJ, Auffenberg G, Kung R, Abecassis M. Evaluation of 300 minimally invasive liver resections at a single institution: less is more. Ann Surg 2007;246:385-92. 5. Cannon RM, Brock GN, Marvin MR, Buell JF. Laparoscopic liver resection: an examination of our first 300 patients. J Am Coll Surg 2011;213:501-7. 6. Nguyen KT, Marsh JW, Tsung A, Steel JJ, Gamblin TC, Geller DA. Comparative benefits of laparoscopic vs open hepatic resection: a critical appraisal. Arch Surg 2011;146: 348-56. 7. Ardito F, Tayar C, Laurent A, Karoui M, Loriau J, Cherqui D. Laparoscopic liver resection for benign disease. Arch Surg 2007;142:1188-93. 8. Buell JF, Cherqui D, Geller DA, et al. The international position on laparoscopic liver surgery: the Louisville Statement, 2008. Ann Surg 2009;250:825-30. 9. Gagner M, Rogula T, Selzer D. Laparoscopic liver resection: benefits and controversies. Surg Clin North Am 2004;84:451-62. 10. Koopmann MC, Heise CP. Laparoscopic and minimally invasive resection of malignant colorectal disease. Surg Clin North Am 2008;88:1047-72. 11. Martin RC, Scoggins CR, McMasters KM. Laparoscopic hepatic lobectomy: advantages of a minimally invasive approach. J Am Coll Surg 2010;210:627-34, 634-6. 12. Abu HM, Underwood T, Zuccaro M, Primrose J, Pearce N. Short- and medium-term results of totally laparoscopic resection for colorectal liver metastases. Br J Surg 2010;97:927-33. 13. Mala T, Edwin B, Gladhaug I, et al. A comparative study of the short-term outcome following open and laparoscopic liver resection of colorectal metastases. Surg Endosc 2002; 16:1059-63. 14. Castaing D, Vibert E, Ricca L, Azoulay D, Adam R, Gayet B. Oncologic results of laparoscopic versus open hepatectomy for colorectal liver metastases in two specialized centers. Ann Surg 2009;250:849-55. 15. Fong Y, Fortner J, Sun RL, Brennan MF, Blumgart LH. Clinical score for predicting recurrence after hepatic resection
Surgery Volume 152, Number 4
16.
17.
18.
19.
20.
21.
22.
23.
24. 25.
26.
27.
for metastatic colorectal cancer: analysis of 1001 consecutive cases. Ann Surg 1999;230:309-18. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373-83. Martin RC, Jarnagin WR, Fong Y, Biernacki P, Blumgart LH, DeMatteo RP. The use of fresh frozen plasma after major hepatic resection for colorectal metastasis: is there a standard for transfusion? J Am Coll Surg 2003;196:402-9. House MG, Ito H, Gonen M, et al. Survival after hepatic resection for metastatic colorectal cancer: trends in outcomes for 1,600 patients during two decades at a single institution. J Am Coll Surg 2010;210:744-5. Choti MA, Sitzmann JV, Tiburi MF, et al. Trends in longterm survival following liver resection for hepatic colorectal metastases. Ann Surg 2002;235:759-66. Johnstone PA, Rohde DC, Swartz SE, Fetter JE, Wexner SD. Port site recurrences after laparoscopic and thoracoscopic procedures in malignancy. J Clin Oncol 1996;14: 1950-6. Paolucci V, Schaeff B, Schneider M, Gutt C. Tumor seeding following laparoscopy: international survey. World J Surg 1999;23:989-95. Ishida H, Murata N, Yamada H, et al. Effect of CO pneumoperitoneum on growth of liver micrometastases in a rabbit model. World J Surg 2000;24:1004-8. Gutt CN, Riemer V, Kim ZG, Erceg J, Lorenz M. Impact of laparoscopic surgery on experimental hepatic metastases. Br J Surg 2001;88:371-5. Hsu TC. Intra-abdominal lesions could be missed by inadequate laparoscopy. Am Surg 2008;74:824-6. Lacy AM, Garcia-Valdecasas JC, Delgado S, et al. Laparoscopy-assisted colectomy versus open colectomy for treatment of non-metastatic colon cancer: a randomised trial. Lancet 2002;359:2224-9. Martin RC, Augenstein V, Reuter NP, Scoggins CR, McMasters KM. Simultaneous versus staged resection for synchronous colorectal cancer liver metastases. J Am Coll Surg 2009;208:842-50. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:41-55.
DISCUSSION Dr Sharon Weber (Madison, WI): Thank you so much for an excellent presentation. Clearly, this is a topic of interest for all hepatobiliary surgeons, and the great group at Louisville is really the leaders in this field. Really, this paper reiterates findings of prior investigators and your own group with the perioperative outcomes, but really extended to these patients that have colorectal metastases, and it has similar benefits for the short term and then for the long term. I have a couple of questions for you about what you started to discuss about who are candidates for the approach. I think even as we get more aggressive in utilizing MIS approach for liver resection, there will always be candidates who really are not going to be candidates for the approach, those that have portal vein embolization, if they have extensive disease, if they have tumor close to major vessels.
Cannon et al 573
So I think part of this really affects the finding and explains the finding that your margin positivity was higher in the open groups. And as you said, I think that this really shows that there still is some selection bias. So the question is, do you think that we can actually compare these groups statistically? Or do we just say for those patients who are candidates for laparoscopic resection, they tend to have better outcomes than the ones that you mentioned? The second question is how you chose the factors that went into your propensity scoring. And I think some would consider other factors equally relevant, particularly the use of neoadjuvant therapy, bilateral disease, underlying liver fibrosis or cirrhosis. And I think the Clinical Risk Score, there’s a suggestion by the group from The Netherlands that maybe in the era of contemporary neoadjuvant therapy that the Clinical Risk Score has become less relevant, so I just wonder why you included that. Dr Robert Cannon (Louisville, KY): Yeah, I agree that the 2 groups will probably never be strictly comparable. There’s always going to be a role for open operations, especially for the lesions you mentioned, especially those that may require biliary reconstruction or resection. Those that are probably centrally located tumors, such as like at the base of segment 4 or 5, those are ones that will really be best for open. And there’s always going to be a role for both. So, yeah, that’s the main idea, is that in patients who are candidates for laparoscopy, those are the benefits here. And, really, the purpose of doing match studies is just to show that if have a patient that could maybe go either way, that perhaps the laparoscopic procedure does offer these benefits. But truly, they’re not apples to apples. It is apples to oranges. And we really just want to say is that laparoscopy, I think, is safe and feasible in the right hands with wellselected patients and it does offer these benefits. In terms of what we included in the risk score, mainly a lot of that had to do with data that was available. I just didn’t necessarily feel that we had strong enough data as far as the type of regimens and the like that we could put together, and also just the ability to actually obtain a match cohort. We really would have liked to have matched on what segment within the liver is the lesion, and the like, but I think it would be more difficult to actually get a matched cohort for comparison there, when, really, you’re never going to have perfectly comparable groups anyway. Dr Mark Carlson (Omaha, NE): Two quick questions. You’ve already addressed this some, but how do you feel about these data? Do you think they’re strong enough to say that, for some cases, these resections should be done laparoscopically, or is it just an option at this point? The second question is, what about trainees? How do you train somebody to do this procedure? Do you do a lap first, open first, both at the same time? How would you train someone? Dr Robert Cannon: I’ll start with the training first, because I think this is an area where, really, we’re seeing a
574 Cannon et al
paradigm shift. The traditional model is you would have to have open expertise first, and then you would go back and get the additional laparoscopic training for this. And this is mainly because people who invented this procedure initially came up in the open surgery era and then came back and developed a laparoscopic procedure later. Now that we are seeing advanced laparoscopy more integrated into just a regular general surgery training, I think we’re probably going to see surgeons trained to do this both open and laparoscopically probably more at the same time. I don’t think we’re quite ready to say that laparoscopy is the standard and should be done. Now, I think if you are at a center where you have that expertise and you’re comfortable with it, I think your patients are going to want you to do it. But I don’t think we’re at the point where you can really say you are violating a standard of care to do an open operation at this point. It’s still very viable. But these are these benefits out there and I think it is a very good option, especially if you develop that expertise. Dr Scott Gruber (Detroit, MI): I didn’t see any comparison of the total cost of the hospital stay or of the operative time. And maybe that may be apples and oranges because of the number of left lateral segmentectomies that were in the laparoscopic group. But still, I think it would be nice to say something about that. Dr Robert Cannon: I think that is a great question to ask, especially because there’s a concern you’re going to
Surgery October 2012
have greater instrument cost, perhaps, with a laparoscopic case. And whether or not that’s balanced by the length of stay, that’s a different question. We didn’t analyze it for this paper; however, we will be presenting that data at the SSO later this month, so if would you like to come and see us then, we will be happy to address it. Dr Gerald Fried (Montreal, Quebec, Canada): During the period involved in the study, you extracted your matched open. Did you control for the year in which the procedure was done? Because I would assume that you probably did increasing proportion of your cases over time laparoscopically. Of course, at the same time, there’s other innovations, new technology, et cetera, hemostatic agents. So I wonder if you put that in your model. Dr Robert Cannon: We did not put the actual year in the model. In the manuscript, we actually looked at 2 different open cohorts. We actually took an older cohort from 1995 on as a separate comparison, just because we think selection bias may be less there, because before laparoscopy came around, obviously, everyone got open. So those might have been more anatomically suitable or similar lesions to the laparoscopic group. But then the way we controlled for year was essentially just limiting to 2004 on. So we assumed that, as they were going on, they were going to have similar adjuvant therapy regimens and the like, so we didn’t actually control for year itself, just in the limitation to 2004 on is how we controlled for that.