Long-Term Oncological Outcomes in Laparoscopic Versus Open Pancreaticoduodenectomy for Pancreatic Cancer: A Systematic Review and Meta-Analysis

Abstract

Background:

Short-term outcomes after laparoscopic pancreaticoduodenectomy (LPD) seem promising, but long-term outcomes of LPD for pancreatic cancer (PC) warrant further investigation.

Methods:

A systematic research of various databases was performed to identify studies analyzing long-term outcomes in LPD versus open pancreaticoduodenectomy (OPD) for PC. Survival parameters of overall survival (OS) and disease-free survival (DFS) were extracted. The search was last conducted before May 23, 2018.

Results:

A total of 10 studies involving 11,180 patients (1437 in LPD and 9743 in OPD) met the final inclusion criteria. Pooled analyses showed that LPD was associated with longer DFS compared with OPD (hazard ratio [HR]: 0.77, 95% confidence interval [CI]: 0.61 to 0.98, P = .033). No significant difference in OS was found between LPD and OPD (HR: 0.98, 95% CI: 0.90 to 1.07, P = .672). In addition, patients of LPD had much shorter time to receive postoperative adjuvant chemotherapy compared with OPD (weighted mean difference: −10.17, 95% CI: −17.90 to −2.45, P = .010).

Discussion:

With regard to long-term survival, LPD is comparable with OPD for PC. Furthermore, LPD is associated with longer DFS compared with OPD. Future well-designed, randomized controlled trials with longer follow-up are still essential to further demonstrate the advantages of LPD for PC.

Introduction

Pancreatic cancer (PC) is one of the most aggressive tumors in the digestive system, which is the fourth leading cause of cancer-related mortality in the United States.¹ Although several kinds of neoadjuvant and adjuvant therapies have been recommended, pancreaticoduodenectomy (PD) is the only potentially curative treatment approach for patients with pancreatic head or periampullary tumors. Because of the complex intra-abdominal dissection and the risk of perioperative morbidity, especially postoperative pancreatic fistula and delayed gastric emptying, PD is still challenging for surgeons ever since Whipple performed the first PD in 1935.² In 1994, Gagner and Pomp described the first laparoscopic PD (LPD) in the world.³ Since then, LPD has become increasingly popular.^4–6

In comparison with open PD (OPD), LPD has several advantages in short-term outcomes such as lower intraoperative blood loss and transfusion, fewer postoperative morbidities, faster recovery, and shortened hospital stay.^7–9 Although some single studies reported the long-term oncological outcomes of LPD for PC, the conclusions are still inconsistent.^10–19 Until now, no meta-analysis has been performed to analyze long-term oncological outcomes of LPD for PC. In this work, we conduct a comparison of long-term oncological outcomes between LPD and OPD. The results indicate that LPD is comparable with OPD for PC with respect to long-term survival. In addition, LPD is associated with longer disease-free survival (DFS) compared with OPD.

Methods

Study selection

Several databases, including Web of Science, PubMed, The Cochrane Library, and Ovid, were retrieved with the following combinations of search terms: “Whipple procedure,” “pancreat*,” “pancreatoduodenectomy,” “pancreaticoduodenectomy,” “PD,” “minimally invasive,” “laparoscopic assisted,” “laparoscopic,” “long-term,” and “survival.” The search was updated to May 23, 2018. References of the obtained studies were manually retrieved to gain additional studies.

Inclusion and exclusion criteria

Each acquired study in this meta-analysis must satisfy the following standards: (1) comparative studies of LPD versus OPD involving observational clinical studies or randomized controlled trials (RCTs); (2) the median follow-up time of the included studies should be more than 12 months, and the studies should have 60 months of follow-up since the first patient; (3) at least one prognostic data item, including overall survival (OS), DFS, or recurrence free survival, should be reported; and (4) if the same authors and/or institutions published more than one publication of the same databases, the study quality, publication year, and sample size were considered, and only one article with high quality was selected. Exclusion criteria were as follows: (1) editorials, expert opinions, abstracts, letters, case reports, and reviews without original data; (2) studies in non-English languages; and (3) studies without available data.

Data extraction and quality assessment

All candidate articles were independently assessed and extracted by 2 reviewers and the following data were used: first author; year of publication; study region; age of patients; months of follow-up; prognostic data; and a hazard ratio (HR) with 95% confidence interval (CI). Inconsistencies were resolved through discussion or an agreement was reached with a third reviewer. For survival data extraction, the HR with 95% CI was directly obtained from the acquired article or estimated using the data, relying on the method reported by Parmar et al.²⁰ If means and standard deviations of continuous variable data were not provided by the selected study in this meta-analysis, the means and standard deviations were imputed from medians and ranges relying on the method reported by Hozo et al.²¹

The quality of nonrandomized studies was independently assessed by 2 reviewers using the Newcastle–Ottawa Scale (NOS).²² The total scaled scores (0–9 scores) contained three parts, including selection (0–4 scores), outcome assessment (0–3 scores), and comparability (0–2 scores), and high-quality studies were those with scores higher than six.

Statistical analyses

Data analyses were performed using Stata SE 12.0 software. I² statistics and chi-square Q test were used to identify the heterogeneity of acquired data. A Ph (P value of heterogeneity) <.10 or I² > 50% indicated significant heterogeneity among the acquired data, and the random-effects model was adopted.^22,23 Otherwise, a fixed-effects model was used. To assess the effects of any single study, sensitivity analysis was conducted. Begg's funnel plot and Egger's test were used to assess the publication bias of the article. A P value less than .05 was considered statistically significant.

Results

Eligible studies

A total of 1027 studies were found in the primary literature search. Among the 1027 articles, 228 were duplicate studies. After carefully reviewing the title and abstract, there were 38 articles that compared LPD with OPD for PC. Of the above 38 studies, 28 studies were further excluded due to omission of long-term survival data. Finally, only 10 studies were included in the present meta-analysis, which were published from 2014 to 2018.^10–19 The flow diagram is listed in Figure 1.

FIG. 1.

Flow diagram depicting the process of identifying and selecting studies for inclusion.

There were 7 retrospective design studies and 3 prospective design studies, and 2 matched retrospective design studies were found among the 10 studies. These studies were derived from different countries, including America, China, Korea, and Italy. The NOS was used to assess the quality of nonrandomized studies, and all of the included studies were considered as high-quality studies (higher than six scores). In addition, main characteristics of all the included studies are summarized in Table 1.

Table 1.

Main Characteristics of All the Studies Included in the Meta-Analysis

Study cohort	Year	Group	No. (M/F)	Age (years)	Follow-up (months)	Outcome	HR	Type of study	NOS score
Chapman et al.¹⁰	2018	LPD	248 (132/116)	79.6 (3.5)^a	17.1	OS	R(M)	Retrospective	8
Chapman et al.¹⁰	2018	OPD	1520 (721/799)	79.5 (3.4)^a	14.0	OS	R(M)	Retrospective	8
Chapman et al.¹¹	2018	LPD	22 (9/13)	64.1 ± 12.6	26.0 (14.6–46.6)	OS/DFS	R(U)	Retrospective	8
Chapman et al.¹¹	2018	OPD	25 (12/13)	60.5 ± 15.6	24.0 (15.1–36.6)	OS/DFS	R(U)	Retrospective	8
Conrad et al.¹²	2017	LPD	40 (26/14)	68 (45–83)	35.9 (0.3–140.5)	OS/DFS	R(U)	Prospective	7
Conrad et al.¹²	2017	OPD	25 (18/7)	66 (43–76)	22.6 (0.6–154.5)	OS/DFS	R(U)	Prospective	7
Croome et al.¹³	2014	LPD	108 (51/57)	66.6 ± 9.6	16.5	OS/DFS	R(U)	Prospective	9
Croome et al.¹³	2014	OPD	214 (131/83)	65.4 ± 10.9	15.1	OS/DFS	R(U)	Prospective	9
Delitto et al.¹⁴	2016	LPD	52 (34/16)	65.3 (1.7)^a	≥12	OS	R(U)	Prospective	7
Delitto et al.¹⁴	2016	OPD	50 (28/22)	68.6 (1.4)^a	≥12	OS	R(U)	Prospective	7
Hakeem et al.¹⁵	2014	LPD	12 (8/4)	67.0 ± 10.2	46.8 (13.0–73.7)	OS/DFS	R(U)	Retrospective matched	8
Hakeem et al.¹⁵	2014	OPD	12 (8/4)	66.3 ± 10.3	56.0 (1.0–97.4)	OS/DFS	R(U)	Retrospective matched	8
Kantor et al.¹⁶	2017	LPD	828 (NR)	65.9 ± 10.7	18	OS	R(U)	Retrospective	8
Kantor et al.¹⁶	2017	OPD	7385 (NR)	65.7 ± 10.4	18	OS	R(U)	Retrospective	8
Meng et al.¹⁷	2018	LPD	58 (32/26)	59.95 ± 9.12	34 (8–60)	OS	R(U)	Retrospective matched	7
Meng et al.¹⁷	2018	OPD	58 (34/24)	60.33 ± 8.66	34 (8–60)	OS	R(U)	Retrospective matched	7
Song et al.¹⁸	2015	LPD	11 (NR)	68.1 ± 7.0	NR	OS	R(U)	Retrospective	6
Song et al.¹⁸	2015	OPD	261 (NR)	61.8 ± 10.5	NR	OS	R(U)	Retrospective	6
Stauffer et al.¹⁹	2017	LPD	58 (32/26)	69.9 (40.6–84.8)	19.6 ± 17.4	OS	R(U)	Retrospective	8
Stauffer et al.¹⁹	2017	OPD	193 (96/97)	68.9 (33.3–86.9)	24.5 ± 27.4	OS	R(U)	Retrospective	8

Statistically significant results are shown in bold.

Mean (SE).

“M” means the HR is calculated using multivariate analysis, “U” means the HR is calculated using univariate analysis.

DFS, disease-free survival; F, female; HR, hazard ratio, obtained by reporting in text (R); LPD, laparoscopic pancreaticoduodenectomy; M, male; NOS score, Newcastle–Ottawa Quality Assessment Scale; OPD, open pancreaticoduodenectomy; NR, not reported; OS, overall survival.

Patient and pathological characteristics

The studies included a total of 11,180 patients, among which 1437 (12.9%) patients were grouped in LPD and 9743 (87.1%) were grouped in OPD. One study¹⁸ showed that patients in the LPD group were older than that of OPD group, but no significant difference was found between LPD and OPD for age in the pooled data (weighted mean difference [WMD] = 0.48, 95% CI: −0.20 to 1.16, P = .167; Table 2). With regard to gender, no significant difference was found (odds ratio [OR] = 1.05, 95% CI: 0.86 to 1.27, P = .636; Table 2). The follow-up time of the included articles ranged from 0.3 to 154.5 months. The study by Chapman et al.¹⁰ showed that LPD had a significantly longer follow-up time compared with OPD, but no significant difference was found in the other studies (Table 1).

Table 2.

Results of the Meta-Analysis

					Heterogeneity
Outcomes of interest	No. of studies	No. of patients	HR, OR, or WMD (95% CI)	P	I² (%)	Ph	Model
Characteristics of patient
Age	8	1137/8173	0.48 (−0.20 to 1.16)	.167	29.1	.196	Fixed effects
Gender	8	596/2097	1.05 (0.86 to 1.27)	.636	36.7	.136	Fixed effects
Pathological characteristics
TNM stage (I, II versus III, IV)	6	1075/8092	1.21 (0.72 to 2.03)	.476	0.0	.627	Fixed effects
Tumor size	7	309/788	−0.12 (−0.43 to 0.20)	.467	79.3	<.001	Random effects
Histological grade (W, M versus P)	4	148/456	0.99 (0.58 to 1.69)	.969	17.3	.304	Fixed effects
Vascular resection	6	290/585	0.95 (0.66 to 1.37)	.796	18.9	.294	Fixed effects
R0 resection	10	1437/9743	1.21 (1.05 to 1.40)	.008	0.0	.678	Fixed effects
Number of lymph nodes resected	8	1137/8173	1.17 (0.26 to 2.08)	.012	70.9	.001	Random effects
Node status	10	1437/9743	0.90 (0.80 to 1.02)	.097	7.6	.372	Fixed effects
Chemotherapy parameters
Neoadjuvant chemo	5	1276/9194	1.06 (0.86 to 1.25)	.715	0.0	.508	Fixed effects
Adjuvant chemo	6	1273/9340	1.05 (0.93 to 1.19)	.421	0.0	.555	Fixed effects
Time to adjuvant chemo	5	1262/9328	−10.17 (−17.90 to 2.45)	.010	94.8	<.001	Random effects
Long-term survival outcomes
OS	9	1425/9731	0.98 (0.90 to 1.07)	.672	0.0	.444	Fixed effects
DFS	4	182/276	0.77 (0.61 to 0.98)	.033	0.0	.922	Fixed effects

Statistically significant results are shown in bold.

CI, confidence interval; DFS, disease-free survival; HR, hazard ratio; M, moderate differentiation; OR, odds ratio; OS, overall survival; P, poor differentiation; Ph, P value of heterogeneity; TNM, tumor–node–metastasis; W, well differentiated; WMD, weighted mean difference.

As shown in Table 3, six studies only have pancreatic ductal adenocarcinoma (PDAC) patients, and the rest four reported periampullary malignancies (PMs). Six studies^13,15–19 provided the tumor–node–metastasis (TNM) stage of tumors, and a meta-analysis comparing I, II versus III, IV between LPD and OPD was performed. The results showed that no significant difference was found (OR = 1.21, 95% CI: 0.72 to 2.03, P = .476; Table 2 and Table 3). In addition, no significant difference was found in pathological characteristics of tumor sizes (WMD = −0.12, 95% CI: −0.43 to 0.20, P = .467), histological grade (W, M versus P) (OR = 0.99, 95% CI: 0.58 to 1.69, P = .969), vascular resection (OR = 0.95, 95% CI: 0.66 to 1.37, P = .796), and the number of positive lymph nodes (OR = 0.90, 95% CI: 0.80 to 1.02, P = .097) among the included patients (Table 2 and Table 3).

Table 3.

Pathological Characteristics of Patients Included in the Meta-Analysis

Study cohort	Year	Group	Pathology	TNM stage (I/II/III/IV)	Tumor size	Histological grade (W/M/P)	Vascular resection (Y/N)	Margin (N/P)	Number of lymph nodes resected	Node status (N/P)
Chapman et al.¹⁰	2018	LPD	PDAC	1/29/202/2/9/5^b	25/169/49/5^b	NR	NR	192/53	74/171/3^b	86/160
Chapman et al.¹⁰	2018	OPD	PDAC	2/140/1180/46/29/123^b	165/1002/329/24^b	NR	NR	1109/396	455/1031/34^b	501/988
Chapman et al.¹¹	2018	LPD	PM	0/2/9/6/3^a	1.9 (1–2.6)	3/9/8	10/10	20/0	19 (16–21.5)	9/11
Chapman et al.¹¹	2018	OPD	PM	1/4/3/2/6^a	1.4 (0.6–2.4)	4/7/5	8/8	16/0	15 (14–21.5)	7/9
Conrad et al.¹²	2017	LPD	PDAC	NR	2.5 (0.3–8.0)	NR	3/37	35/5	18 (6–53)	16/24
Conrad et al.¹²	2017	OPD	PDAC	NR	3.0 (1.3–6.0)	NR	16/84	21/4	17 (9–62)	9/16
Croome et al.¹³	2014	LPD	PDAC	8/15/85/0	3.3 ± 1.0	0/16/90	22/86	84/24	21.4 ± 8.1	29/79
Croome et al.¹³	2014	OPD	PDAC	13/25/176/0	3.3 ± 1.3	3/33/167	51/163	164/50	20.1 ± 7.5	60/154
Delitto et al.¹⁴	2016	LPD	PM	NR	2.5 (0.1)^a	NR	0/52	47/5	23.0 (1.2)^a	18/34
Delitto et al.¹⁴	2016	OPD	PM	NR	3.1 (0.2)^a	NR	0/50	37/13	20.8 (1.2)^a	10/40
Hakeem et al.¹⁵	2014	LPD	PM	4/6/2/0	1.98 ± 1.03	12/0	8/4	9/3	20.7 ± 6.3	6/6
Hakeem et al.¹⁵	2014	OPD	PM	1/11/0/0	1.92 ± 0.73	8/4	4/8	8/4	18.5 ± 10.3	2/10
Kantor et al.¹⁶	2017	LPD	PDAC	89/739/0/0	149/500/168	NR	NR	651/177	18.1 ± 9.5	263/566
Kantor et al.¹⁶	2017	OPD	PDAC	677/6708/0/0	1248/4618/1431	NR	NR	5623/1762	17.1 ± 9.6	2225/5127
Meng et al.¹⁷	2018	LPD	PM	30/26/2/0	1.85 (1.50–2.60)	6/42^a	NR	58/0	16 (15–18)	48/10
Meng et al.¹⁷	2018	OPD	PM	27/28/3/0	2.00 (1.95–3.00)	6/42^a	NR	55/3	15 (15–17)	46/12
Song et al.¹⁸	2015	LPD	PDAC	7/4/0/0	1.9 ± 1.1	5/5/0	NR	11/0	9.9 ± 3.9	6/5
Song et al.¹⁸	2015	OPD	PDAC	134/99/0/0	2.1 ± 1.0	81/124/20	NR	218/4	12.2 ± 8.7	138/81
Stauffer et al.¹⁹	2017	LPD	PDAC	7/50/0/1	2.5 (0.3–10.0)	NR	20/38	49/9	27 (9–70)	17/41
Stauffer et al.¹⁹	2017	OPD	PDAC	16/168/4/2	3.5 (0.3–14.0)	NR	60/133	154/39	17 (1–63)	63/126

Statistically significant results are shown in bold.

Mean (SE).

The results are shown in different ways.

LPD, laparoscopic pancreaticoduodenectomy; M, moderate differentiation; N, no or negative; NR, not reported; OPD, open pancreaticoduodenectomy; P, poor differentiation; P, positive; PDAC, pancreatic ductal adenocarcinoma; PM, periampullary malignancy; TNM, tumor–node–metastasis; W, well differentiated; Y, yes.

However, LPD was associated with a higher R0 resection rate compared with OPD (OR = 1.21, 95% CI: 1.05 to 1.40, P = .008; Table 2 and Table 3). Furthermore, LPD was also associated with a larger number of harvested lymph nodes in comparison with that of OPD (WMD = 1.17, 95% CI: 0.26 to 2.08, P = .012; Table 2 and Table 3), with significant heterogeneity (I² = 70.9%; Ph = .001). Then, sensitivity analysis was conducted to assess the effect of any single study. We found two studies reported by Stauffer et al. and Song et al.^18,19 with great effects on the result of conversion rate. After excluding studies by Stauffer et al. and Song et al., no obvious heterogeneity was found (I² = 47.0%; Ph = .093).

Chemotherapy parameters

Five studies^{10,12–14,16} of the included articles reported neoadjuvant chemotherapy before surgery. A meta-analysis was performed and results revealed that no significant difference was observed between LPD and OPD (OR = 1.06, 95% CI: 0.86 to 1.25, P = .715; Table 2 and Table 3). In addition, six studies^{10,11,13,15,16,19} provided adjuvant chemotherapy (ACT) after surgery, and results of the meta-analysis showed that no obvious difference was found between LPD and OPD (OR = 1.05, 95% CI: 0.93 to 1.19, P = .421; Table 2 and Table 3).

However, a significantly shorter time to ACT after surgery was found in the group of LPD when compared with the group of OPD (WMD = −10.17, 95% CI: −17.90 to 2.45, P = .010; Fig. 2), with significant heterogeneity (I² = 94.8%; Ph < .001). Then, sensitivity analysis was conducted. The result revealed one study reported by Croome et al.¹³ with great effects on the result. After excluding the study by Croome et al., no obvious heterogeneity was found (I² = 0.0%; Ph = .455). LPD also had a significantly shorter time to ACT than OPD (WMD = −1.45, 95% CI: −2.91 to 0.00, P = .050).

FIG. 2.

Forest plot illustrating time to ACT after surgery. ACT, adjuvant chemotherapy.

Long-term survival outcomes

There were five studies^{10,13,14,16,19} that provided data of OS with HRs and 95% CI, and Song et al.¹⁸ provided the raw data. The remaining studies at least provided Kaplan–Meier survival curves. Then, the HR with 95% CI was directly obtained from the acquired articles or the data of HRs with 95% CI acquired from the Kaplan–Meier survival curves were estimated relying on the method reported by Parmar et al.²⁰ However, the Kaplan–Meier survival curve by Hakeem et al.¹⁵ was not of sufficient quality to allow for generating data of OS. Then, a meta-analysis of data of the remaining nine studies was performed and results showed that no significant difference was found between LPD and OPD (HR = 0.98, 95% CI: 0.90 to 1.07, P = .672; Fig. 3A). Results of the subgroup analysis also showed no significant difference between LPD and OPD in the subgroup of pathology of PDAC and PM.

FIG. 3.

Forest plots illustrating long-term survival outcomes: (A) OS HR; (B) DFS HR. DFS, disease-free survival; HR, hazard ratio; OS, overall survival.

Four studies^11–13,15 provided the Kaplan–Meier survival curves of DFS. Data of HRs with 95% CI also acquired from the Kaplan–Meier survival curves were estimated, and results showed that LPD was associated with better DFS compared with OPD (HR = 0.77, 95% CI: 0.60 to 1.00, P = .043; Fig. 3B). Results of the subgroup analysis also showed that LPD was associated with longer DFS in the subgroup of the pathology of PDAC (HR = 0.76, 95% CI: 0.59 to 0.98, P = .033; Fig. 3B); however, no significant difference was found between LPD and OPD in the subgroup of the pathology of PM (HR = 1.10, 95% CI: 0.39 to 3.14, P = .852; Fig. 3B). In addition, the data of long-term survival were summarized in Table 4.

Table 4.

Data of Long-Term Survival in the Meta-Analysis

Study cohort	Year	Group	1 year/10 months ^a	2 years/20 months ^a	3 years/30 months ^a	4 years/40 months ^a	5 years/60 months ^a
OS
Chapman et al.¹⁰	2018	LPD	118 (64.5%)	57 (31.1%)	16 (8.7%)	4 (2.2%)	0 (0.0%)
Chapman et al.¹⁰	2018	OPD	628 (55.4%)	290 (25.6%)	104 (9.2%)	24 (2.1%)	0 (0.0%)
Chapman et al.¹¹	2018	LPD	18 (90.0%)	11 (55.0%)	6 (30.0%)	4 (20.0%)	-
Chapman et al.¹¹	2018	OPD	13 (81.3%)	9 (56.3%)	6 (37.5%)	5 (31.3%)	-
Conrad et al.¹²	2017	LPD	32 (80.5%)	25 (62.0%)	20 (49.2%)	17 (43.3%)	16 (39.7%)
Conrad et al.¹²	2017	OPD	19 (77.8%)	13 (50.7%)	12 (46.4%)	9 (34.3%)	8 (30.0%)
Croome et al.¹³	2014	LPD	NR	46 (42.6%)	NR	10 (9.3%)	0 (0.0%)
Croome et al.¹³	2014	OPD	NR	74 (34.6%)	NR	22 (10.3%)	1 (0.5%)
Hakeem et al.¹⁵	2014	LPD	12 (100.0%)	NR	9 (75.0%)	NR	9 (75.0%)
Hakeem et al.¹⁵	2014	OPD	9 (75.0%)	NR	7 (58.0%)	NR	6 (50.0%)
Song et al.¹⁸	2015	LPD	6 (54.5%)	2 (18.2%)	1 (9.1%)	1 (9.1%)	1 (9.1%)
Song et al.¹⁸	2015	OPD	191 (73.2%)	89 (34.1%)	49 (18.8%)	28 (10.7%)	13 (5.0%)
Stauffer et al.¹⁹	2017	LPD	34 (58.6%)	15 (25.9%)	10 (17.2%)	6 (10.3%)	2 (3.4%)
Stauffer et al.¹⁹	2017	OPD	115 (59.9%)	60 (31.3%)	34 (17.7%)	21 (10.9%)	17 (8.9%)
DFS
Chapman et al.¹¹	2018	LPD	17 (85.0%)	7 (35.0%)	4 (20.0%)	4 (20.0%)	-
Chapman et al.¹¹	2018	OPD	12 (75.0%)	9 (56.3%)	4 (25.0%)	3 (18.8%)	-
Conrad et al.¹²	2017	LPD	28 (70.9%)	17 (43.7%)	13 (33.3%)	12 (31.5%)	9 (21.9%)
Conrad et al.¹²	2017	OPD	16 (62.3%)	9 (37.9%)	9 (37.9%)	8 (30.0%)	6 (25.7%)
Croome et al.¹³	2014	LPD	NR	35 (32.4%)	NR	7 (6.5%)	0 (0.0%)
Croome et al.¹³	2014	OPD	NR	48 (22.4%)	NR	14 (6.5%)	1 (0.5%)
Hakeem et al.¹⁵	2014	LPD	12 (100.0%)	NR	11 (92.0%)	NR	10 (83.0%)
Hakeem et al.¹⁵	2014	OPD	12 (100.0%)	NR	10 (83.0%)	NR	10 (83.0%)

Chapman et al.¹¹ and Croome et al.¹³ provided data of long-term survival with months; the others provided data with years.

DFS, disease-free survival; LPD, laparoscopic pancreaticoduodenectomy; NR, not reported; OPD, open pancreaticoduodenectomy; OS, overall survival.

Publication bias

Begg's funnel plot and Egger's test were used to assess the publication bias of the articles. As shown in the funnel plot of OS (Fig. 4), no evidence of significant publication bias was found.

FIG. 4.

Funnel plots of OS. OS, overall survival.

Discussion

With continuous technical advances and improvement in laparoscopic training, laparoscopic pancreatoduodenectomy (LPD) has been adopted by expert surgeons in selected patients.^{3–19,24–26} Compared with open pancreatoduodenectomy (OPD), the short-term surgical outcomes of LPD presented comparable results. Some meta-analyses even showed that minimally invasive pancreatoduodenectomy was associated with lower estimated blood loss, more harvested lymph nodes, more R0 resections, fewer postoperative complications of delayed gastric emptying, and shorter time of hospital stay.^7,9 In addition, one recent randomized clinical trial (RCT) of LPD versus OPD for periampullary tumors revealed that LPD was beneficial in terms of intraoperative blood loss, blood transfusion, surgical site infection, and postoperative duration of hospital stay.²⁷

Due to the limited number of patients included in the single studies and no meta-analysis performed on long-term survival with LPD for PC until now, long-term survival with LPD for PC, which is a major factor to assess the benefits of LPD, is still uncertain. To provide a reliable conclusion that LPD is comparable with OPD with respect to long-term oncological outcomes, this meta-analysis was performed.

The aim of the present meta-analysis is to investigate OS and DFS with LPD versus OPD for PC. Studies without the two factors are excluded in this meta-analysis. The median follow-up time of the included studies should be longer than 12 months. Furthermore, to find whether LPD was better for long-term survival and DFS compared with OPD, we pooled all the data of the included studies to perform a meta-analysis (Table 2). Results revealed that although no significant difference is found between LPD and OPD in OS, LPD is associated with longer DFS compared with OPD. This is the first study to confirm that LPD has comparable long-term survival effectiveness for PC in comparison with OPD.

Long-term survival is one of the important factors to assess the benefits of LPD. Apart from surgical factors such as lymph node resections (LNR), negative margin rate, and postoperative complications, several other factors also affect long-term survival for patients treated by surgery, including demographics of patients, characteristics of the tumor, and adjuvant therapy. For example, Seppänen et al. analyzed 306 PDAC patients treated by surgery, and the overall 1-, 3-, 5-, and 10-year survival rates were 74%, 36%, 22%, and 14%, respectively.²⁸ Univariate analysis showed that tumor size, histological differentiation grade, LNR, perivascular invasion, and adjuvant therapy significantly affected OS of patients. In the present meta-analysis, all the data were pooled from the included studies in order to find whether LPD could improve long-term survival (Table 2). Although LPD is associated with more harvested lymph nodes and R0 resections in this study, no significant difference is found between LPD and OPD in OS.

Recently, several researches have revealed that R0 resection determined by the 1-mm rule (R0^{1-mm rule}) was an independent determinant for postoperative survival. Delpero et al. found that R0^{1-mm rule} resection for PDAC was an independent determinant for postoperative survival.²⁹ In addition, Nitta et al. found that R0^{1-mm rule} resection correlated with better OS than R1^{1-mm rule} resection, but R0 resection determined by TNM classification (no malignancy infiltrating at any of the resection margins) had no impact on OS when analyzing the impact of margin status on PDAC recurrence and survival following PD. They concluded that R0^{1-mm rule} resection might be an independent variable for predicting long-term survival.³⁰ In the present study, however, most included articles define R0 resection determined by TNM classification. R0 resection, similar to blood loss, transfusion, lymph node harvest, complication rate, and adjuvant treatment, is just one factor affecting long-term survival of PC patients treated by laparoscopic surgery. No significant difference is found between LPD and OPD in OS in the present meta-analysis. In the future, carefully case-matched or RCT studies relying on R0^{1-mm rule} resection are required to analyze long-term survival with LPD versus OPD for the treatment of PDAC.

Interestingly, our study reveals that LPD is associated with better DFS compared with OPD. Potential reasons may be reduced physiologic and immunologic effects of LPD and the enhanced ability of patients treated with LPD to receive timely and complete postoperative adjuvant therapy.^13,31 In patients with colon cancer, Biagi et al. revealed that the increasing time to ACT correlated with a decrease in survival.³² Similarly, the increasing time interval to ACT is significantly associated with decreased survival in patients with ovarian cancer.³³ Recently, the European Study Group for Pancreatic Cancer-3 (ESPAC-3) found that completion of all six cycles of ACT was predictive of OS after surgery for PDAC. They also observed that if patients could not complete six cycles of therapy, chemotherapy started within 12 weeks was an important factor for long-term survival.³⁴ In the present study, LPD had a significantly shorter time to ATC after surgery compared with OPD, which was associated with fewer complications and faster recovery in patients of LPD. Therefore, it can be assumed that LPD can significantly shorten the time to ATC, which may prolong DFS of patients treated with surgery. However, all the data of DFS were estimated from Kaplan–Meier survival curves in our meta-analysis, and the follow-up time of the included studies was not equal. Reporting bias may exist in the data of DFS. The data of DFS just provided some reference values in the present meta-analysis. Therefore, carefully case-matched or RCT studies are required to prove the above hypothesis in the future.

A number of limitations in the present meta-analysis of nonrandomized studies existed, which should be acknowledged and carefully considered when interpreting the results. First, all studies included in the meta-analysis are non-RCTs (NRCTs) in design, which may be liable to cause selection bias and exaggerate the effect of the approaches. Pooling data from NRCTs is a debated topic in the field of meta-analysis. Nevertheless, several studies proved that results derived from studies of well-designed NRCTs may be reliable.^35,36 In this meta-analysis, all of the 10 included studies are well-designed NRCTs, according to the NOS. Second, characteristics of the populations as well as tumors, R0 resection, transaction devices in laparoscopic surgery, the learning curve of surgeons, and definitions of outcomes vary considerably among the included studies, which may cause heterogeneity and impact the reliability of our results. For example, Park et al. found that the 5-year DFS rate of stage III disease in the expert period (85%) was better than that of the learning curve period (53.3%; P = .046) in laparoscopic right hemicolectomy.³⁷ In laparoscopy-assisted distal gastrectomy for gastric cancer, however, Hu et al. revealed that long-term outcomes were not sacrificed during the learning stage.³⁸ Therefore, these factors should be included in the analysis of long-term survival in future, which certainly plays a role in the findings of this analysis. Finally, due to the limited or nonexistent descriptive statistics of the included studies, it is impossible to generate more in-depth comparative results, such as the longer follow-up and recurrence rate.

In conclusion, results of our meta-analysis indicate that LPD for PC is comparable with OPD with respect to long-term survival. In addition, LPD is associated with better DFS compared with OPD. However, evidence is still limited due to the lack of high-quality RCTs. Additional prospective and multicenter RCTs with longer follow-up are needed for further investigation in the future.

Footnotes

Acknowledgments

The authors thank all colleagues of the department of general surgery in The Second Affiliated Hospital of Nanchang University. The authors also thank the professors from The First Affiliated Hospital of Nanchang University for reading and commenting on the manuscript. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Authors' Contributions

J.C. conceived the study; L.P. collected data, performed the data analysis, and wrote the article; Z.Z. and Z.C. edited the article; and all authors read and approved the final manuscript.

Disclosure Statement

No competing financial interests exist.

References

Siegel

, Miller

, Jemal

. Cancer Statistics, 2017. CA Cancer J Clin, 2017; 67:7–30.

Whipple

, Parsons

, Mullins

. Treatment of carcinoma of the ampulla of Vater. Ann Surg, 1935; 102:763–779.

Gagner

, Pomp

. Laparoscopic pylorus-preserving pancreatoduodenectomy. Surg Endosc, 1994; 8:408–410.

Langan

, Graham

, Chin

, et al. Laparoscopic-assisted versus open pancreaticoduodenectomy: Early favorable physical quality-of-life measures. Surgery, 2014; 156:379–384.

Sharpe

, Talamonti

, Wang

, et al. Early national experience with laparoscopic pancreaticoduodenectomy for ductal adenocarcinoma: A comparison of laparoscopic pancreaticoduodenectomy and open pancreaticoduodenectomy from the National Cancer Data Base. J Am Coll Surg, 2015; 221:175–184.

Zureikat

, Breaux

, Steel

, et al. Can laparoscopic pancreaticoduodenectomy be safely implemented?. J Gastointest Surg, 2011; 15:1151–1157.

Correa-Gallego

, Dinkelspiel

, Sulimanoff

, Fisher

, Viñuela

, Kingham

, et al. Minimally-invasive vs open pancreaticoduodenectomy: Systematic review and meta-analysis. J Am Coll Surg, 2014; 218:129–139.

Qin

, Qiu

, Zhao

, Pan

, Zeng

. Does minimally-invasive pancreaticoduodenectomy have advantages over its open method? A meta-analysis of retrospective studies. PLoS One, 2014; 9:e104274.

de Rooij

, Lu

, Steen

, Gerhards

, Dijkgraaf

, Busch

, Lips

, Festen

, Besselink

; Dutch Pancreatic Cancer Group. Minimally invasive versus open pancreatoduodenectomy: Systematic review and meta-analysis of comparative cohort and registry studies. Ann Surg, 2016; 264:257–267.

10.

Chapman

, Gajdos

, Hosokawa

, Henderson

, Paniccia

, et al. Comparison of laparoscopic to open pancreaticoduodenectomy in elderly patients with pancreatic adenocarcinoma. Surg Endosc, 2018; 32:2239–2248.

11.

Chapman

, Gleisner

, Ibrahim-Zada

, Overbey

, Paniccia

, Meguid

, et al. Laparoscopic pancreaticoduodenectomy: Changing the management of ampullary neoplasms. Surg Endosc, 2018; 32:915–922.

12.

Conrad

, Basso

, Passot

, Zorzi

, Li

, et al. Comparable long-term oncologic outcomes of laparoscopic versus open pancreaticoduodenectomy for adenocarcinoma: A propensity score weighting analysis. Surg Endosc, 2017; 31:3970–3978.

13.

Croome

, Farnell

, Que

, Reid-Lombardo

, Truty

, Nagorney

, Kendrick

. Total laparoscopic pancreaticoduodenectomy for pancreatic ductal adenocarcinoma: Oncologic advantages over open approaches?. Ann Surg, 2014; 260:633–638; discussion 638–640.

14.

Delitto

, Luckhurst

, Black

, Beck

, George

Jr. , Sarosi

, et al. Oncologic and perioperative outcomes following selective application of laparoscopic pancreaticoduodenectomy for periampullary malignancies. J Gastrointest Surg, 2016; 20:1343–1349.

15.

Hakeem

, Verbeke

, Cairns

, Aldouri

, Smith

, Menon

. A matched-pair analysis of laparoscopic versus open pancreaticoduodenectomy: Oncological outcomes using leeds pathology protocol. Hepatobiliary Pancreat Dis Int, 2014; 13:435–441.

16.

Kantor

, Talamonti

, Sharpe

, Lutfi

, Winchester

, Roggin

, et al. Laparoscopic pancreaticoduodenectomy for adenocarcinoma provides short-term oncologic outcomes and long-term overall survival rates similar to those for open pancreaticoduodenectomy. Am J Surg, 2017; 213:512–515.

17.

Meng

, Cai

, Li

, Cai

, Peng

. Comparison of laparoscopic and open pancreaticoduodenectomy for the treatment of nonpancreatic periampullary adenocarcinomas. Surg Laparosc Endosc Percutan Tech, 2018; 28:56–61.

18.

Song

, Kim

, Hwang

, Lee

, et al. Matched case-control analysis comparing laparoscopic and open pylorus-preserving pancreaticoduodenectomy in patients with periampullary tumors. Ann Surg, 2015; 262:146–155.

19.

Stauffer

, Coppola

, Villacreses

, Mody

, Johnson

, Li

, Asbun

. Laparoscopic versus open pancreaticoduodenectomy for pancreatic adenocarcinoma: Long-term results at a single institution. Surg Endosc, 2017; 31:2233–2241.

20.

Parmar

, Torri

, Stewart

. Extracting summary statistics to perform meta-analyses of the published literature for survival endpoints. Stat Med, 1998; 17:2815–2834.

21.

Hozo

, Djulbegovic

, Hozo

. Estimating the mean and variance from the median, range, and the size of a sample. BMC Med Res Methodol, 2005; 5:13.

22.

Peng

, Lin

, Li

, et al. Systematic review and meta-analysis of robotic versus open pancreaticoduodenectomy. Surg Endosc, 2017; 31:3085–3097.

23.

Peng

, Xiao

, Liu

, et al. Laparoscopic versus open left-sided hepatectomy for hepatolithiasis: A systematic review and meta-analysis. J Laparoendosc Adv Surg Tech A, 2017; 27:951–958.

24.

Kamyab

, Khatlani

, Jacobs

. Total laparoscopic pancreaticoduodenectomy: Early single institution experience. HPB, 2012; 14:38.

25.

Lei

, Wei

, Guo

, Wei

. Minimally invasive surgical approach compared with open pancreaticoduodenectomy: A systematic review and meta-analysis on the feasibility and safety. Surg Laparosc Endosc Percutan Tech, 2014; 24:296–305.

26.

Shin

, Kim

, Song

, Kim

, Hwang

, Lee

, et al. Totally laparoscopic or robot-assisted pancreaticoduodenectomy versus open surgery for periampullary neoplasms: Separate systematic reviews and meta-analyses. Surg Endosc, 2017; 31:3459–3474.

27.

Palanivelu

, Senthilnathan

, Sabnis

, Babu

, Srivatsan Gurumurthy

, Anand Vijai

, et al. Randomized clinical trial of laparoscopic versus open pancreatoduodenectomy for periampullary tumours. Br J Surg, 2017; 104:1443–1450.

28.

Seppänen

, Juuti

, Mustonen

, Haapamäki

, Nordling

, Carpelan-Holmström

, et al. The results of pancreatic resections and long-term survival for pancreatic ductal adenocarcinoma: A single-institution experience. Scand J Surg, 2017; 106:54–61.

29.

Delpero

, Jeune

, Bachellier

, Regenet

, Le Treut

, Paye

, et al. Prognostic value of resection margin involvement after pancreaticoduodenectomy for ductal adenocarcinoma: Updates from a french prospective multicenter study. Ann Surg, 2017; 266:787–796.

30.

Nitta

, Nakamura

, Mitsuhashi

, Asano

, Okamura

, Tsuchikawa

, et al. The impact of margin status determined by the one-millimeter rule on tumor recurrence and survival following pancreaticoduodenectomy for pancreatic ductal adenocarcinoma. Surg Today, 2017; 47:490–497.

31.

Burpee

, Kurian

, Murakame

, et al. Themetabolic and immune response to laparoscopic versus open liver resection. Surg Endosc, 2002; 16:899–904.

32.

Biagi

, Raphael

, Mackillop

, et al. Association between time to initiation of adjuvant chemotherapy and survival in colorectal cancer: A systematic review and meta-analysis. JAMA, 2011; 305:2335–2342.

33.

Hofstetter

, Concin

, Braicu

, et al. The time interval from surgery to start of chemotherapy significantly impacts prognosis in patients with advanced serous ovarian carcinoma—Analysis of patient data in the prospective OVCAD study. Gynecol Oncol, 2013; 131:15–20.

34.

Valle

, Palmer

, Jackson

, et al. Optimal duration and timing of adjuvant chemotherapy after definitive surgery for ductal adenocarcinoma of the pancreas: Ongoing lesion from the ESPAC-3 study. J Clin Oncol, 2014; 32:504–512.

35.

Bekheit

, Khafagy

, Bucur

, Katri

, Elgendi

, Abdel-Salam

, et al. Donor safety in live donor laparoscopic liver procurement: Systematic review and meta-analysis. Surg Endosc, 2015; 29:3047–3064.

36.

Yin

, Fan

, Ye

, Yin

, Wang

. Short- and long-term outcomes after laparoscopic and open hepatectomy for hepatocellular carcinoma: A global systematic review and meta-analysis. Ann Surg Oncol, 2013; 20:1203–1215.

37.

Park

, Kim

, Hur

, Min

, et al. Long-term oncologic outcomes of laparoscopic right hemicolectomy during the learning curve period: Comparative study with cases after the learning curve period. Surg Laparosc Endosc Percutan Tech, 2015; 25:52–58.

38.

, Ma

, Zang

, Xue

, Xu

, Wang

, et al. Learning curve and long-term outcomes of laparoscopy-assisted distal gastrectomy for gastric cancer. J Laparoendosc Adv Surg Tech A, 2014; 24:487–492.