Ann Surg Oncol DOI 10.1245/s10434-014-3639-0

ORIGINAL ARTICLE – PANCREATIC TUMORS

Systematic Review and Meta-Analysis Comparing the Surgical Outcomes of Invasive Intraductal Papillary Mucinous Neoplasms and Conventional Pancreatic Ductal Adenocarcinoma Ye-Xin Koh, MBBS, MRCS1, Aik-Yong Chok, MBBS, MRCS1, Hui-Li Zheng, MSc2, Chuen-Seng Tan, BSc, MSc, PhD2, and Brian K. P. Goh, MBBS, MMed, MSc, FRCS1,3 1

Department of Hepatopancreatobiliary and Transplant Surgery, Division of Surgery, Singapore General Hospital, Singapore, Singapore; 2Saw Swee Hock School of Public Health, National University of Singapore, Singapore, Singapore; 3 Office of Clinical Sciences, Duke-NUS Graduate Medical School, Singapore, Singapore

ABSTRACT Objective. The aim of this study was to summarize the current literature comparing the surgical outcomes of invasive intraductal papillary mucinous neoplasms (IPMNINV) and conventional pancreatic ductal adenocarcinomas (PDAC) in order to determine the differences in disease characteristics and prognosis. Methods. Systematic review of the literature yielded 12 comparative studies reporting the clinicopathological characteristics and overall survival (OS) of 1,450 patients with IPMNINV with 19,304 patients with conventional PDAC. Results. IPMNINV had a significantly lower likelihood of tumors extending beyond the pancreas [27.6 vs. 94.3 %; T4 vs. T1: odds ratio (OR) 0.111, 95 % confidence intervals (CI) 0.057–0.214], nodal metastasis (45.4 vs. 62.9 %: OR 0.507, 95 % CI 0.347–0.741), positive margin (14.2 vs. 28.3 %; OR 0.438, 95 % CI 0.322–0.596), perineural invasion (49.2 vs. 76.5 %; OR 0.304, 95 % CI 0.106–0.877) and vascular invasion (25.2 vs. 45.7 % OR 0.417, 95 % CI 0.177–0.980) when compared with PDAC. The 5-year OS of IPMNINV was significantly better than PDAC [31.4 vs. 12.4 %: hazard ratio (HR) 0.659, 95 % CI 0.574–0.756]. The tubular subtype had a poorer 5-year OS and demonstrated significantly more aggressive features such as nodal metastases, vascular invasion, and perineural invasion compared with the colloid subtype.

Conclusion. IPMNINV were significantly more likely to present at an earlier stage and were less likely to demonstrate nodal involvement, perineural invasion and vascular invasion. When controlled for stage, IPMNINV had an improved OS when compared with PDAC in the early stages.

Ó Society of Surgical Oncology 2014

METHODS

Intraductal papillary mucinous neoplasm (IPMN) was first recognized as a distinct pathological entity and distinguished from mucinous cystic neoplasms by the WHO in 1996.1–3 According to the WHO classification, IPMNs are graded based on the most aggressive histological epithelial change observed, ranging from low-grade dysplasia (adenoma) to invasive carcinoma.1,4 Survival following resection of non-invasive IPMNs is excellent and superior to invasive IPMNs (IPMNINV).5,6 IPMNINV have been reported to account for up to 40–60 % of all resected IPMNs.5,7–10 Presently, there is controversy regarding the outcome of surgically-resected IPMNINV when compared with conventional pancreatic ductal adenocarcinoma (PDAC). Some investigators have reported that IPMNINV have a less aggressive behavior and better prognosis compared with PDAC, whereas others have reported no difference in outcomes.5,6,8,11–17 The primary aim of this study was to summarize the current literature comparing the surgical outcomes of IPMNINV and conventional PDAC in order to determine the difference in disease characteristics and prognosis between the two entities.

First Received: 2 January 2014 B. K. P. Goh, MBBS, MMed, MSc, FRCS e-mail: [email protected]

A systematic review of the published literature was performed using the PubMed and Cochrane databases upto

Y.-X. Koh et al.

30 July 2013 to identify studies that compared IPMNINV and conventional PDAC. The medical subject heading search terms used were ‘pancreas’, ‘mucinous’, ‘neoplasm’, ‘adenocarcinoma’, and ‘surgical procedures’. The keywords used were ‘invasive’, ‘intraductal’, ‘ductal’, ‘papillary’, ‘tumors’, and ‘comparative study’. The acronyms used were ‘IPMN’ and ‘IPMT’. The described terms were also searched as text words and their combinations are described: ‘invasive intraductal papillary neoplasm/ tumor compared with PDAC’, and ‘outcomes of surgical resection of IPMN/IPMT compared with PDAC’. Key references of the shortlisted studies were also searched manually. The search was conducted independently by two authors (YXK and AYC) and the search results obtained by both authors were discussed with the senior author (BKG). The final list of studies to be shortlisted was decided by consensus between all three authors. This study was conducted in accordance to the PRISMA guidelines.18 Data Extraction All shortlisted studies were assessed independently by two authors (YXK and AYC) according to a modified Newcastle-Ottawa scale, to match the requirements of the current study. Three main factors were assessed: (1) selection of the patients; (2) comparability of the study groups; and (3) outcome assessment. The scoring scale ranged from 0 to 9, and studies with a score of 6 or greater were considered to be of high quality and were included in this study. The following data were extracted from the included studies: first author, year of data collection, year of publication, country of origin, characteristics of study population, number of patients in invasive IPMN and PDAC groups, clinicopathological characteristics, matching criteria and overall survival (OS).

used: (1) studies that did not report the comparative outcomes of interest or if the information was impossible to derive; (2) studies that focused on imaging, DNA, biochemical or immunohistochemical aspects of IPMN; (3) studies that focused on ultrasonographic and cyst fluid analysis of IPMN; (4) studies that focused on epidemiology and incidence of extrapancreatic malignancies in IPMN; (5) studies that only focused on comparison of the histological subtypes of IPMN; (6) studies that only compared concomitant PDAC not deemed to arise from a background of IPMN with conventional PDAC; and (7) studies written in languages other than English. Definitions, Parameters and Outcomes of Interest IPMNINV were compared with PDAC based on clinicopathological features such as location, type of surgical resection, size, nodal metastases, margin status, perineural invasion, vascular invasion, tumor grade, and tumor stage. The OS and stage-matched survival were also compared. IPMNINV was defined as the presence of invasive component in IPMN or ductal adenocarcinoma arising from a background of IPMN according to the WHO criteria.1 Conventional PDAC were defined as ductal adenocarcinomas not related to IPMNs. Accepted staging methods were the American Joint Committee on Cancer (AJCC), the International Union Against Cancer (UICC), and the Japan Pancreas Society (JPS) staging methods. The differences between the staging methods were accounted for and comparisons were only made for the comparable stages. The T staging for the JPS, AJCC, and UICC staging methods were comparable; however, TNM staging was similar in the AJCC and UICC classification, but different in the JPS TNM staging. Hence JPS staging was not used as a classification in the comparison of TNM stage.19–21

Statistical Analysis Inclusion Criteria The inclusion criteria were (1) comparison of invasive IPMN with PDAC, (2) evaluation of at least one of the clinicopathological or survival characteristics mentioned in the Definitions, Parameters and Outcomes of Interest section below, and (3) when there were studies reported by the same institution (and/or) authors, the study with the larger study population or the one with higher quality as graded by our modified Newcastle-Ottawa scale was included. Exclusion Criteria All studies that did not meet the inclusion criteria were excluded. In addition, the following exclusion criteria were

The statistical methods and techniques are summarized in Appendix 1.22–29 RESULTS A total of 477 potential studies were identified. Overall, 278 duplicate studies were excluded and 199 studies underwent abstract review. Eighteen studies were selected for full-text review.5–8,12–17,23–26,30–33 Of these, 12 studies met the inclusion criteria and were included in our metaanalysis (See Figure 3 in Appendix 2).8,12–17,23–26,30 Three studies5,7,17 were from the same institution and the two earlier studies by Sohn et al..5,7 were excluded. Four studies were excluded from analysis as they scored less

1987–1999

1992–2004

1992–2005

1999–2006

1990–2007

1971–2005

1995–2006

1989–2009

1983–2007

1987–2009

1990–2008

1995–2009

Maire et al.12

Shimada et al.23

Schnelldorfer et al.8

Woo et al.24

Murakami et al.25

Wasif et al.13

Poultsides et al.17

Waters et al.15

Yopp et al.16

Yamaguchi et al.14

Mino-Kenudson et al.26

Kang et al.30 Korea

USA

Japan

USA

USA

USA

USA

Japan

Korea

USA

Japan

France

Country

59 (MD or MT: 32; BD: 27)

61 (MD or MT: 48; BD: 13)

122 (MD or MT: 61; BD: 61)

59 (not specified)

113 (MD or MT: 88; BD: 25)

MT: 94; BD: 38)

132 (MD or

729 (not specified)

16 (MD: 15; BD: 1)

19 (not specified)

63 (not specified)

18 (MD or MT: 14; BD: 4)

51 (MD or MT: 47; BD: 4)

Invasive IPMN

219

570

7,605

59

845

1,128

8,082

106

174

63

274

51

Conventional PDAC

–

66.3

63.5

–

65.4

–

65.1

–

61.4

–

62.4

63.8 (median)

Mean age (years)

–

305 (48.3)

4,751 (61.5)

59 (50.0)

563 (58.8)

660 (52.4)

4,453 (50.5)

71 (58.2)

232 (63.7)

–

171 (62.4)

70 (68.6)

Male [n (%)]

42.3 (median)

–

17.4

–

25.1

–

–

–

9.6

38.4

–

32.7

Mean follow-up time (months)

No matching, consecutive cases used/staging not specified

No matching, consecutive cases used/UICC 6th ed. staging

No matching, consecutive cases used/JPS 2nd ed. staging

Matching done by nomogram from MSKCC/no comparisons by staging

No matching, consecutive cases used/AJCC 7th ed. staging

No matching, consecutive cases used/AJCC staging, edition not specified

No matching, SEER database (based on ICD coding) used/ AJCC 6th ed. staging

No matching, consecutive cases used/UICC 6th ed. staging

No matching, consecutive cases used/UICC 6th ed. staging

Matched for age and TNM, UICC 6th ed. staging/ clinical data only for IPMN

No matching, consecutive cases used/UICC 5th ed. staging

Matched age and TNM/UICC 4th ed. staging/clinical data compares non-invasive vs. invasive IPMN

Exclusion criteria/comments

6*

9*

9*

9*

9*

9*

8*

7*

8*

7*

7*

7*

Study quality

*Study quality based on our modified Newcastle-Ottawa scale (only studies scoring 6 or greater were included)

IPMN intraductal papillary mucinous neoplasms, PDAC pancreatic ductal adenocarcinomas, MD main duct only, MT mixed main and branch duct, BD branch duct only, UICC International Union Against Cancer, SEER Surveillance, Epidemiology and End Results, ICD International Classification of Diseases, AJCC American Joint Committee on Cancer, JPS Japan Pancreas Society, MSKCC Memorial Sloan Kettering Cancer Center

Period of data acquisition

Study, references

TABLE 1 Characteristics of the included studies comparing invasive IPMN and conventional PDAC

Systematic Review and Meta-Analysis

Y.-X. Koh et al. TABLE 2 Results of meta-analysis comparing baseline clinical characteristics in invasive IPMN and conventional PDAC Clinical characteristic

No. of studies

No. of patients

Invasive IPMN: N (%)

Conventional PDAC: N (%)

OR (95 % CI)

Location

8

18,256

Head or uncinate: 840 (73.9)

Head or uncinate: 14,071 (82.2)

0.543 (0.360–0.820)

0.004*

Body or tail: 297 (26.1)

Body or tail: 3,048 (17.8)

Surgical type (distal vs. Whipple)

3

10,171

Whipple: 700 (82.7) Distal: 146 (17.3)

Whipple: 8,374 (89.8) Distal: 951 (10.2)

1.827 (1.510–2.211)

\0.001*

Surgical type (total vs. Whipple)

3

9,932

4.437 (1.265–15.564)

0.020*

Local spread (T2 vs. T1)

5

0.379 (0.098–1.460)

0.159

0.141 (0.044–0.454)

0.001*

0.111 (0.057–0.214)

\0.001*

0.277 (0.143–0.538)

\0.001*

0.126 (0.028–0.579)

0.008*

0.235 (0.052–1.063)

0.060

886

Local spread (T3 vs. T1)

5

3,812

Local spread (T4 vs. T1)

5

5,307

Tumor stage (TNM2 vs. TNM1)

5

Tumor stage (TNM3 vs. TNM1)

5

Tumor stage (TNM4 vs. TNM1)

4

8,516 2,077 2,121

Whipple: 700 (84.5)

Whipple: 8,374 (92.0)

Total: 128 (15.5)

Total: 730 (8.0)

T1: 71 (51.4)

T1: 296 (39.6)

T2: 67 (48.6)

T2: 452 (60.4)

T1: 71 (27.3)

T1: 296 (8.3)

T3: 189 (72.7)

T3: 3,256 (91.7)

T1: 71 (72.4),

T1: 296 (5.7)

T4: 27 (27.6)

T4: 4,913 (94.3)

TNM1: 221 (27.7)

TNM1: 1,100 (14.3)

TNM2: 577 (72.3)

TNM2: 6,618 (85.7)

TNM1: 221 (83.4)

TNM1: 1,100 (60.7)

TNM3: 44 (16.6)

TNM3: 712 (39.3)

TNM1: 187 (67.0)

TNM1: 1,083 (58.8)

TNM4: 92 (33.0)

TNM4: 759 (41.2)

p value

Reference group is conventional PDAC IPMN intraductal papillary mucinous neoplasms, PDAC pancreatic ductal adenocarcinomas, OR odds ratio, CI confidence interval * p value \ 0.05

than 6 according to our modified Newcastle-Ottawa scale.5,31–33 The 12 studies are summarized in Table 1 and the comparison of the baseline characteristics of IPMNINV and PDAC are summarized in Table 2. The summary of the pooled results from the meta-analysis comparing the clinical, pathological, and survival characteristics in IPMNINV and PDAC are presented in Tables 3 and 4. Detailed results for each outcome are presented in Figs. 1 and 2; Appendix 3. Clinical Parameters Eight studies classified IPMN as main duct (MD), mixed duct type (MT) or branched duct (BD).12,14,15,17,23,25,26,30 Of 572 invasive IPMN patients, 30.2 % were BD IPMNs. Tumor location, which was reported in eight studies,13,14,16,17,21–24 demonstrated that IPMNINV was significantly less likely than PDAC to be located at the head/uncinate area [73.9 vs. 82.2 %; odds ratio (OR) 0.543, 95 % confidence intervals (CI) 0.360–0.820]. Three studies reported the types of surgical resections.13,15,17 The likelihood of patients with IPMNINV undergoing distal

pancreatectomy was 1.827 (95 % CI 1.510–2.211). Compared with PDAC, patients with IPMNINV were also more likely to undergo total pancreatectomy rather than pancreatoduodenectomy (15.5 vs. 8.0 %: OR 4.437, 95 % CI 1.265–15.564) (Table 3, and Appendix 4b and c). Staging Analysis of five studies demonstrated that IPMNINV had significantly lower likelihood of T3 or T4 tumors (T3 vs. T1: OR 0.141, 95 % CI 0.044–0.454; T4 vs. T1: OR 0.111, 95 % CI 0.057–0.214) (Appendix 4d, e, and f).13,17,23,24,26 Overall, IPMNINV had a lower likelihood of being in an advanced stage when compared with PDAC (TNM stage 2, 3, or 4: OR 0.277, 0.126, 0.235, respectively) (Appendix 4g, h, and i). The pooled results from eight studies suggested that the mean tumor size of IPMN was not significantly different from PDAC,13–16,23–26 but IPMNINV had a significantly lower likelihood of nodal metastasis (45.4 vs. 62.9 %: OR 0.507, 95 % CI 0.347–0.741).13–17,22–25 IPMNINV also had a lower likelihood of harboring positive margins (14.2 vs. 28.3 %: OR 0.438, 95 % CI 0.322–0.596),15–17,25,26

Systematic Review and Meta-Analysis TABLE 3 Results of meta-analysis comparing pathological and survival characteristics in invasive IPMN and conventional PDAC Pathological characteristic

No. of studies

No. of patients

Invasive IPMN: N (%)

Conventional PDAC: N (%)

WESa/ORb/HRc (95 % CI) p value

Tumor size (mm)

8

18,852

–

–

-0.619 (-1.383 to 0.145)a

Nodal metastasis

9

19,362

Positive: 574 (45.4)

Positive: 11,377 (62.9)

0.112

b

0.507 (0.347–0.741)

\0.001*

0.438 (0.322–0.596)b

\0.001*

0.304 (0.106–0.877)b

0.028*

0.417 (0.177–0.980)b

0.045*

0.545 (0.047–6.375)b

0.628

1.068 (0.444–2.569)b

0.884

0.659 (0.574–0.756)c

\0.001*

Negative: 691 (54.6) Negative: 6,720 (37.1) Margin status

5

3,088

Positive: 54 (14.2)

Positive: 767 (28.3)

Negative: 327 (85.8) Negative: 1,940 (71.7) Perineural invasion

6

2,477

Positive: 150 (49.2)

Positive: 1,650 (76.5)

Vascular invasion

6

2,420

Negative: 155 (50.8) Negative: 522 (24.0) Positive: 77 (25.2) Positive: 967 (45.7) Negative: 228 (74.8) Negative: 1,148 (54.3)

Lymphatic invasion

3

Tumor grade

896

5

10,925

Overall 5-year survival 12

20,754

Positive: 31 (31.6)

Positive: 386 (48.4)

Negative: 67 (68.4)

Negative: 412 (51.6)

High: 344 (34.5)

High: 3,908 (39.4)

Low: 601 (61.5)

Low: 6,020 (60.6)

Alive: 456 (31.4)

Alive: 2,402 (12.4)

Died: 994 (68.6)

Died: 16,902 (87.6)

Reference group is conventional PDAC IPMN intraductal papillary mucinous neoplasms, PDAC pancreatic ductal adenocarcinomas, WES weighted effect size, OR odds ratio, HR hazard ratio, CI confidence interval * p value \ 0.05 a

WES

b

OR

c

HR

TABLE 4 Stage-matched survival comparison between invasive IPMN and conventional PDAC Study

Stage

Maire et al.12

I

27

II

6

III Woo et al.24 Murakami et al.25

I II IA

Wasif et al.13

Poultsides et al.17

Waters et al.15

Invasive IPMN (n)

Conventional PDAC (n)

Survival rate for IPMN (%)

Survival rate for PDAC (%)

p value of the test of difference in survival

Conclusion

27

67

23

\0.001*

IPMN had better survival

6

–

–

[0.05

No significant difference

16

16

–

–

[0.05

No significant difference

I/II/III

21

264

–

–

\0.001*

IPMN had better survival

IV

11

68

–

–

0.725

No significant difference

5

11

–

–

0.346

No significant difference

11

60

–

–

0.049*

IPMN had better survival

45

323

–

–

0.01*

IPMN had better survival

IB

101

628

–

–

\0.001*

IPMN had better survival

IIA

174

1,876

–

–

0.018*

IPMN had better survival

IIB

257

3,632

–

–

0.073

No significant difference

I

36

34

–

–

0.006*

IPMN had better survival

II

28

77

–

–

0.325

No significant difference

III

60

633

–

–

0.248

No significant difference

IV IA

5 21

19 6

– 75

– 49

0.259 0.03*

No significant difference IPMN had better survival

IB

13

11

43

25

0.05*

IPMN had better survival

IIA

23

24

24

22

[0.05

No significant difference

IIB

43

58

16

8

[0.05

No significant difference

IPMN intraductal papillary mucinous neoplasms, PDAC pancreatic ductal adenocarcinomas * p value \ 0.05

Y.-X. Koh et al.

(a)

(b)

(c)

FIG. 1 Forest plots illustrating the results of a meta-analysis comparing pathological characteristics in invasive IPMN and conventional PDAC. Pooled ORs or WMDs with 95 % CIs were calculated using the fixed effects or random effects* model where

appropriate. a Nodal metastasis; b perineural invasion; c vascular invasion. IPMN intraductal papillary mucinous neoplasms, PDAC pancreatic ductal adenocarcinomas, ORs odds ratios, WMDs weighted mean differences, CIs confidence intervals, df degrees of freedom

Systematic Review and Meta-Analysis

FIG. 2 Forest plots illustrating the results of a meta-analysis comparing overall 5-year survival in invasive IPMN and conventional PDAC. Pooled HRs with 95 % CIs were calculated using the random

effects model. IPMN intraductal papillary mucinous neoplasms, PDAC pancreatic ductal adenocarcinomas, HRs hazard ratios, CIs confidence intervals, df degrees of freedom

demonstrating perineural invasion (49.2 vs. 76.5 %: OR 0.304, 95 % CI 0.106–0.877) and vascular invasion (25.2 vs. 45.7 %: OR 0.417, 95 % CI 0.177–0.980) compared with PDAC.16,17,23–26 There was no statistically significant difference in the frequency of lymphatic invasion or highgrade tumors between the two entities 13,16,17,25,26 (Fig. 1, and Appendix 3b, c, and d).

histological grade, and tumor location were similar between the two groups, tubular adenocarcinoma was associated with a significantly higher rate of nodal metastasis, vascular invasion, and perineural invasion. OS in tubular adenocarcinoma was poorer than colloid carcinoma (Table 5). Sensitivity Analyses

Survival Based on the pooled results of 12 studies, the 5-year OS of IPMNINV was significantly better than PDAC [31.4 vs. 12.4 %: hazard ratio (HR) 0.659, 95 % CI 0.574–0.756] (Fig. 2). In terms of stage-matched survival, the results were summarized in Table 4. In general, IPMNINV showed significantly better OS compared with PDAC when the tumors were in the early stages. However, for patients with advanced-stage tumors (stage III or IV), there did not seem to be a difference between IPMNINV and PDAC (Fig. 2; Table 4). Tubular Versus Colloid Invasive Intraductal Papillary Mucinous Neoplasms Four studies compared the invasive subtypes of IPMN, tubular adenocarcinoma and colloid carcinoma.15–17,26 Although tumor size, T-stage distribution,

Sensitivity analyses performed demonstrated that the findings of all the available studies were robust (Appendices 5–7).

DISCUSSION The 5-year OS of invasive IPMNs after resection has been reported to be up to 40 % in some series. This compares favorably with the surgical outcomes of PDAC, which has a dismal 5-year OS rate of about 10–20 %, even in large, specialized centers.5,9,11,15–17,34–36 However, conflicting results have been reported by other studies which have demonstrated that both pathological entities were associated with a similarly poor OS.6,8 The results of this meta-analysis demonstrate that at the time of surgical resection, IPMNINV was significantly less likely to be in the advanced stage, harboring nodal involvement, demonstrating perineural invasion, and

Y.-X. Koh et al. TABLE 5 Overview of various clinical, pathological, and survival characteristics comparison between tubular and colloid carcinoma subtypes of invasive IPMN Outcome

Studies

p value for each study

Tumor size

Poultsides et al.17

[0.05

Local spread (T1 vs. T2)

Yopp et al.16

0.854

Mino-Kenudson et al.26 Poultsides et al.17

0.712 0.458

Waters et al.15 Tumor grade

Location (head vs. other) Nodal metastasis

0.341

Poultsides et al.17

0.049

Yopp et al.16

0.725

Mino-Kenudson et al.26

1.000

16

0.69

Mino-Kenudson et al.26

0.039*

Poultsides et al.17

0.002*

Waters et al.15

0.07*

Yopp et al.16 Mino-Kenudson et al.

0.526

Poultsides et al.17

0.001*

Yopp et al.16

0.506

Mino-Kenudson et al.26 Perineural invasion

17

Poultsides et al. Yopp et al.

16

0.898

0.032 (-0.463, 0.528)

No significant difference

0.376

1.591 (0.569, 4.453)

No significant difference

0.095

0.506 (0.228, 1.125)

No significant difference

0.156

0.539 (0.230, 1.266)

No significant difference

2.809 (1.840, 4.286)

Tubular adenocarcinoma had higher rate of positive nodal metastasis

0.001*

4.855 (1.963, 12.006)

Tubular adenocarcinoma had higher rate of positive vascular invasion

0.010*

2.299 (1.219, 4.336)

Tubular adenocarcinoma had higher rate of positive perineural invasion

\0.001*

0.426 (0.288, 0.630)

Colloid carcinoma had better survival

\ 0.001*

0.144 0.045* 1.000

Mino-Kenudson et al.26 Overall 5-year survival

Conclusion

0.14 26

Vascular invasion

OR/HR (95 % CI)

\0.001*

Mino-Kenudson et al.26

Yopp et al.

Overall p value

17

0.038*

Poultsides et al.

0.015*

Yopp et al.16

0.0001*

Mino-Kenudson et al.26

0.056

IPMN intraductal papillary mucinous neoplasms, OR odds ratio, HR hazard ratio, CI confidence interval * p value \ 0.05

showing vascular invasion compared with PDAC. Hence, not surprisingly, IPMNINV was associated with a superior OS compared with PDAC. There are several possible reasons for these observations. First, IPMNINV is more likely to be detected earlier as its precursor lesion is frequently detectable radiologically as a cystic lesion of the pancreas, unlike that for PDAC. Pancreatic intraepithelial neoplasia, the precursor for PDAC, is a histological diagnosis and is almost always never demonstrated on cross-sectional imaging.37 Current international consensus guidelines for IPMN are likely to enable clinicians to detect and resect IPMNs at the premalignant or early stages of invasive disease.34,38,39 Second, IPMN is associated with the production of mucin giving rise to

symptoms early on in the disease and resulting in early detection. This may also account for our findings that there was an increased proportion of IPMNINV in the distal pancreas compared with PDAC as PDAC in the distal pancreas are notorious for their delayed presentation due to the absence of symptoms early on.32 Lastly, these findings may be accounted for by the postulation that IPMNINV are biologically less aggressive than PDAC. This hypothesis is partially supported by our observation that when both groups were in the early stage, invasive IPMN had a significantly better OS than conventional PDAC. However, this survival advantage was lost when both groups presented at a more advanced stage (Table 4). Unfortunately, we were unable to perform a

Systematic Review and Meta-Analysis

pooled analysis of staged-matched survival in this study as there was inadequate statistical information. Presently, IPMNINV is defined by the WHO as the presence of invasive component in IPMN or PDAC arising from a background of IPMN.1 The WHO criteria was met by most of the studies in the present review.8,12,14–17,23,24,26,30 Only the study by Wasif et al.,13 which extracted data from the Surveillance, Epidemiology and End Results (SEER) database based on International Classification of Diseases (ICD) coding, did not provide a definition of malignant IPMN. It is not known at present if IPMN with an invasive component is a distinct entity and should be distinguished from IPMN with concomitant PDAC. Most studies in this review did not attempt to distinguish between IPMNINV and IPMN with concomitant PDAC (PDAC not arising from the background of IPMN).8,12,15–17,23–25 Only the study by Yamaguchi et al. 14 attempted to differentiate the two entities. However, in the study, 15 % of the resected cases of intraductal papillary mucinous carcinoma could not be distinguished and assigned to either group. This demonstrated the inherent difficulty in distinguishing between the two groups, especially when PDAC arose in the vicinity of IPMN.14 Nonetheless, both IPMNINV and concomitant PDAC were similarly associated with favorable characteristics in terms of biological aggressiveness and survival compared with conventional PDAC.14 Genomic analysis may prove to be useful in differentiation of the two entities as it has been reported by some investigators to be helpful in distinguishing IPMN from PDAC.40 In addition to disease stage and grade, the prognosis of IPMN has been reported to be influenced by the percentage of invasive component within the tumor.30 Kang et al. classified IPMNINV according to the percentage of invasive component, and demonstrated that clinicopathological factors such as advanced T stage, nodal metastases, perineural invasion, vascular invasion, and recurrence rates increased significantly with increased percentage of invasive component of the IPMN. The 5-year survival of minimally IPMNINV (\5 % invasive component) was 80 versus 29.3 % for PDAC-associated IPMNs ([50 %).30 Several studies have classified IPMNINV into colloid and tubular subtypes.15–17,26 These studies have demonstrated that tubular subtypes were associated with a significantly worse prognosis compared with colloid subtypes, with an OS similar to that of PDAC.15–17,26 The present metaanalysis, which included four studies, showed that tubular adenocarcinomas were more likely to harbor nodal metastases and vascular invasion than colloid carcinomas.

Although direct TNM-stage comparisons were not performed in these studies, our findings suggest that there was a greater proportion of tubular adenocarcinoma at a higher TNM stage than colloid carcinomas. In this review, 69.8 % of invasive IPMNs originated from MD and/or MT lesions. This is consistent with the literature whereby MD or MT lesions were associated with a higher incidence of IPMNINV compared with BD IPMNs (40 vs. 11 %).2,10,34,41 In this study, more than 80 % of IPMNINV originated from the head/uncinate process of the pancreas, and a significant proportion (15 %) underwent total pancreatectomy. This was likely due to the multifocal nature of BD and MT IPMNs and the inherent difficulty in achieving clear margins with limited resections.8,42 The main limitation of this systematic review was the retrospective nature of the included case–control studies. Hence, important confounding factors such as the use of adjuvant chemotherapy were not available. The majority of the studies did not describe the chemotherapy administered to these patients, which could be significant in light of a recent randomized controlled trial where the use of adjuvant gemcitabine after resection of pancreatic cancer significantly delayed recurrent tumors and improved survival.43,44 Furthermore, most of the studies did not perform a staged-matched comparison between IPMN and PDAC, which would be critical in determining if IPMNINV was biologically less aggressive than PDAC. However, in the absence of any prospective studies, this review represents a strong attempt, with the inclusion of high-quality studies, at providing the best level of evidence in comparing the outcomes of invasive IPMN with conventional PDAC. It highlights the limitations of currently available evidence and the need for further studies to address this controversy. CONCLUSIONS The results of this meta-analysis demonstrated that IPMNINV were significantly more likely to present at an earlier stage and were less likely to demonstrate nodal involvement, perineural invasion and vascular invasion. Furthermore, in the early stages, IPMNINV had an improved OS when compared with conventional PDAC, even when controlled for stage. This suggests that invasive IPMN may be associated with a more favorable disease biology. Larger series performing staged-matched comparisons between IPMNINV versus PDAC, and further genetic studies, are needed to determine if there is a true difference in their biology.

Y.-X. Koh et al.

APPENDIX 1. STATISTICAL ANALYSES To facilitate the pooling of results across studies in the meta-analysis, the number of individuals with or without the event of interest in the IPMN and PDAC groups was used for dichotomous outcomes. The Mantel–Haenszel (MH) method was used to pool the OR across studies. A Woolf–Haldane continuity correction of 0.5 was used when the number of events for one of the groups was zero.22 Studies in which there was no event in an outcome of interest for both groups were excluded. Four studies reported tumor size as a continuous outcome,13,15,23,24 while four others reported it as a dichotomized outcome.14,16,25,26 ORs were transformed into effect sizes allowing the pooling of results as effect measures for continuous outcomes by using the inversevariance (IV) approach.27 Analysis of the 5-year OS was performed by computing the p value for the test of survival difference between the two groups, the number of events in each group, and the number of patients in each group. These statistics were used to compute the HR and its corresponding variance for each study. The HRs and variances from the 11 studies were, in turn, pooled together via the IV approach.28 Heterogeneity between the studies was evaluated using the Chi squared test of heterogeneity. If there was limited evidence supporting the assumption of homogeneity of studies for an outcome (i.e. the p value of the heterogeneity Chi squared test was \0.1), a random effects model was used;29 otherwise, a fixed effects model was used. Two separate sets of sensitivity analyses were performed: (1) excluding each study individually; and (2) excluding studies with a sample size fewer than 20 patients in each group, from the analysis of each outcome. Pooled results from these subgroups were computed and compared with

the pooled results from the set of studies without these exclusion criteria. The meta-analysis was conducted using STATA SE 10 (StataCorp. LP, College Station, TX, USA). APPENDIX 2 See Fig. 3.

Articles identified by literature search 477 results Articles duplicated 278 results Articles deleting duplicates 199 results

Articles eliminated from title and abstract review 181 results Articles selected for full text review 18 results

Articles included in this report 12 results

Articles eliminated from full text 6 results

FIG. 3 Results of a systematic search of the literature

Systematic Review and Meta-Analysis

APPENDIX 3 See Fig. 4.

(a)

(b)

(c)

FIG. 4 Forest plots illustrating the results of a meta-analysis comparing pathological characteristics in invasive IPMN and conventional PDAC. Pooled ORs or weighted mean differences (WMDs) with 95 %

CIs were calculated using the fixed effects or random effects* model where appropriate. a Tumor size; b margin status; c lymphatic invasion; d tumor grade

Y.-X. Koh et al.

(d)

FIG. 4 continued

APPENDIX 4 See Fig. 5. (a)

(b)

FIG. 5 Forest plots illustrating the results of a meta-analysis comparing clinical characteristics in invasive IPMN and conventional PDAC. Pooled ORs with 95 % CIs are calculated using the fixed effects or random effects* model where appropriate. a Location;

b surgical type: distal vs. Whipple, and c total vs. Whipple; d local spread: T2 vs. T1, e T3 vs. T1, and f T4 vs. T1; g tumor stage: TNM2 vs. TNM1, hTNM3 vs. TNM1, and iTNM4 vs. TNM1

Systematic Review and Meta-Analysis

(c)

(d)

(e)

FIG. 5 continued

Y.-X. Koh et al.

(f)

(g)

(h)

(i)

FIG. 5 continued

Systematic Review and Meta-Analysis

APPENDIX 5

TABLE 6 continued

Sensitivity analyses were carried out by excluding each study individually from the analysis of each outcome (Appendix 6). Among outcomes with a p value less than 0.05 in Table 3, location, surgical type, local spread, tumor stage, nodal metastasis, and margin status had consistent findings in the sensitivity analyses. Although surgical type, perineural invasion, and vascular invasion had p-values exceeding 0.05 in the sensitivity analyses, the ORs were in the same consistent direction as those in Table 3. Among outcomes with a p value more than 0.05 in Table 3, local spread (T2 vs. T1), tumor stage (TNM stage 4 vs. TNM stage 1), lymphatic invasion, and tumor size had p-values lower than 0.05 in the sensitivity analyses. The ORs and effect sizes were consistent with those in Table 3. Additional sensitivity analyses were performed by including those studies with more than 20 patients in each group (Appendix 7). The exclusion of studies with small sample sizes reduces heterogeneity across the studies and did not change the conclusion for all outcomes. This suggests that the findings based on all available studies shown were robust.

Clinical characteristic Local spread (T2 vs. T1)

0.159

0.426

0.266

Woo et al.24

0.385

0.231

Poultsides et al.

0.376

0.345

Yamaguchi et al.14

0.219 \0.001

Mino-Kenudson et al.26

0.574 0.141

0.001

Shimada et al.23

0.150

0.006

Woo et al.24 Poultsides et al.17

0.189 0.154

0.009 0.020

Yamaguchi et al.14

0.095 \0.001

Mino-Kenudson et al.26

0.140

23

Clinical characteristic

Studies excluded

Location

Surgical type (Whipple vs. distal)

p value

0.019

Shimada et al.

0.121 \0.001

Woo et al.24

0.119 \0.001

Poultsides et al.17

0.067 \0.001

Yamaguchi et al.14

0.118 \0.001

Mino-Kenudson et al.26

0.123 \0.001 0.277 \0.001

Woo et al.24

0.454 \0.001

Murakami et al.25

0.256 \0.001

13

0.216

0.001

Waters et al.15

0.238

0.002

Mino-Kenudson et al.26

0.263

0.003

0.126

0.008

Woo et al.24

0.420 \0.001

Murakami et al.25 Wasif et al.13

0.136 0.023 0.060 \0.001

Waters et al.15

0.119

0.017

Mino-Kenudson et al.26

0.113

0.027

0.235

0.060

0.357

0.218

0.319

0.153

0.543

0.004

Shimada et al.23

0.568

0.012

Woo et al.24

0.520

0.005

Murakami et al.25

0.504

0.002

Wasif et al.13

0.546

0.037

Poultsides et al.17

0.626

0.012

Murakami et al.

Yopp et al.16

0.528

0.006

Wasif et al.13

0.113 \0.001

Yamaguchi et al.14 Mino-Kenudson et al.26

0.480 0.601

0.001 0.024

Mino-Kenudson et al.26

0.222

1.827 \0.001 Wasif et al.13

2.032 \0.001 17

Poultsides et al. 15

Waters et al. Surgical type (Whipple vs. total)

OR

Tumor stage (TNM3 vs. TNM1)

0.447

0.111 \0.001

Local spread (T4 vs. T1)

Wasif et al.

TABLE 6 Results of sensitivity analysis by excluding each study individually from all available studies for each outcome (where the reference group is conventional PDAC)

p value

0.379

Local spread (T3 vs. T1)

See Table 6.

OR

Shimada et al.23 17

Tumor stage (TNM2 vs. TNM1)

APPENDIX 6

Studies excluded

1.751 \0.001

Tumor stage (TNM4 vs. TNM1)

Pathological characteristic

Woo et al.24 25

Studies excluded

Nodal metastasis Shimada et al.23 24

0.095

OR

p value

0.507

\0.001

0.430

\0.001

1.854 \0.001

Woo et al.

0.541

0.001

4.437

Murakami et al.25

0.527

0.002

0.020

Wasif et al.13

7.578 \0.001

Wasif et al.13

0.492

0.006

Poultsides et al.17

3.882

0.170

2.934

0.123

Poultsides et al.17 Waters et al.15

0.554 0.504

0.004 0.003

15

Waters et al.

Y.-X. Koh et al. TABLE 6 continued Pathological characteristic

TABLE 6 continued Studies excluded

OR

Survival characteristic

Studies excluded

HR

p value

Schnelldorfer et al.8

0.648

\0.001

Yopp et al.

0.473 \0.001

Woo et al.24

0.663

\0.001

Yamaguchi et al.14

0.518

Murakami et al.25

0.671

\0.001

16

26

0.005

13

0.647

\0.001

0.438 \0.001

Poultsides et al.17

0.648

\0.001

Murakami et al.25 Poultsides et al.17

0.450 \0.001 0.448 \0.001

Waters et al.15

0.687

\0.001

Yamaguchi et al.14

0.645

\0.001

Waters et al.15

0.463 \0.001

Mino-Kenudson et al.26

0.673

\0.001

0.438 \0.001

Kang et al.30

0.667

\0.001

Mino-Kenudson et al.

0.551

Margin status

16

Yopp et al.

26

0.304

0.028

Shimada et al.23

0.192

0.001

Woo et al.24

0.429

0.130

Murakami et al.25

0.287

0.043

17

Poultsides et al.

0.268

0.075

Yopp et al.16

0.308

0.075

Mino-Kenudson et al.26

0.410

0.099

0.417

0.045

Vascular invasion Shimada et al.23

0.303 \0.001

Woo et al.24

0.450

0.082

Murakami et al.25 Poultsides et al.17

0.492 0.362

0.106 0.136

Yopp et al.16

0.438

0.136

Mino-Kenudson et al.26

0.482

0.186

0.545

0.628

Shimada et al.23

0.193

0.005

Woo et al.24

1.310

0.872

Mino-Kenudson et al.26

0.755

0.907

Lymphatic invasion

Tumor grade

1.068

0.884

Murakami et al.25

0.767

0.558

Wasif et al.13

1.318

0.660

Poultsides et al.17

0.862

0.754

Yopp et al.16

1.060

0.911

Mino-Kenudson et al.26

1.543

0.402

-0.619

0.112

Shimada et al. Woo et al.24

-0.619 -0.620

0.118 0.118

Murakami et al.25

-0.533

0.285

Wasif et al.13

-0.620

0.118

Waters et al.15

-0.620

0.118

Yopp et al.

-0.623

0.226

Yamaguchi et al.14

-1.054 \0.001

Tumor size 23

16

Mino-Kenudson et al.26 -0.360 Studies excluded

0.281

HR

p value

0.659

\0.001

Maire et al.12

0.662

\0.001

Shimada et al.23

0.663

\0.001

Overall survival

Wasif et al.

0.004

0.394 \0.001

Mino-Kenudson et al. Perineural invasion

Survival characteristic

p value

OR odds ratio

Systematic Review and Meta-Analysis

APPENDIX 7 See Table 7. TABLE 7 Results of sensitivity analysis by excluding studies with a sample size fewer than 20 patients in either, or both, IPMN and PDAC for each outcome (where the reference group is conventional PDAC) Clinical characteristic

No. of No. of Invasive IPMN: studies patients N (%)

Conventional PDAC: N (%)

OR (95 % CI)

Location

7

Head or uncinate: 833 (74.3)

Head or uncinate: 13,900 (82.4)

0.524 (0.332–0.827)

0.006*

Body or tail: 288 (25.7)

Body or tail: 2,972 (17.6)

Surgical type (distal vs. Whipple)

Same as Appendix 2

Surgical type (total vs. Whipple)

Same as Appendix 2

Local spread (T2 vs. T1)

3

0.451 (0.076–2.679)

0.381

0.224 (0.051–0.995)

0.049*

Local spread (T3 vs. T1)

3

Local spread (T4 vs. T1)

3

Tumor stage (TNM2 vs. TNM1) 4 Tumor stage (TNM3 vs. TNM1) 3 Tumor stage (TNM4 vs. TNM1) 3

Pathological characteristic

No. of studies

17,993

848 3,530 5,130 8,429 2,048 2,083

No. of patients

T1: 63 (51.2)

T1: 289 (39.9)

T2: 60 (48.8)

T2: 436 (60.1)

T1: 63 (26.7)

T1: 289 (8.8)

T3: 173 (73.3)

T3: 3,005 (91.2)

T1: 63 (75.0)

T1: 289 (5.7)

T4: 21 (25.0)

T4: 4,757 (94.3)

TNM1: 216 (27.6)

TNM1: 1,089 (14.2)

TNM2: 566 (72.4)

TNM2: 6,558 (85.8)

TNM1: 216 (83.1)

TNM1: 1,089 (60.9)

TNM3: 44 (16.9)

TNM3: 699 (39.1)

TNM1: 182 (66.4)

TNM1: 1,072 (59.3)

TNM4: 92 (33.6)

TNM4: 737 (40.7)

Invasive IPMN: N (%)

p value

0.132 (0.064–0.272) \0.001* 0.256 (0.119–0.548) \0.001* 0.136 (0.024–0.762)

0.023*

0.319 (0.066–1.532)

0.153

Conventional PDAC: N (%)

WES/OR/ HR (95 % CI) -0.533 (-1.545–0.478)

0.301

0.463 (0.344–0.624)

\0.001*

0.450 (0.328–0.618)

\0.001*

Tumor size (mm)

5

18,852

–

–

Nodal metastasis

6

18,773

Positive: 1,212 (46.0)

Positive: 11,166 (63.6)

Negative: 654 (54.0)

Negative: 6,395 (36.4)

Margin status

4

2,966

Positive: 365 (14.0)

Positive: 719 (27.6)

Negative: 314 (86.0)

Negative: 1,882 (72.4)

p value

Perineural invasion

3

1,888

Positive: 132 (52.4) Negative: 120 (47.6)

Positive: 1,371 (83.8) Negative: 265 (16.2)

0.232 (0.071–0.753)

0.015*

Vascular invasion

3

1,832

Positive: 64 (25.4)

Positive: 793 (50.2)

0.337 (0.204–0.558)

\0.001*

Negative: 188 (74.6)

Negative: 787 (49.8)

Lymphatic invasion

Only one study was available

Tumor grade

4

0.767 (0.316–1.864)

0.558

0.677 (0.588–0.779)

\0.001*

Overall 5-year survival

10

10,803 20,341

High: 333 (33.9)

High: 3,878 (39.5)

Low: 648 (66.1)

Low: 5,944 (60.5)

Alive: 442 (31.2)

Alive: 2,328 (12.3)

Died: 974 (68.8)

Died: 16,597 (87.7)

Reference group is conventional PDAC IPMN intraductal papillary mucinous neoplasms, PDAC pancreatic ductal adenocarcinomas, WES weighted effect size, OR odds ratio, HR hazard ratio, CI confidence interval *p value \ 0.05

Y.-X. Koh et al.

REFERENCES 1. Kloppel G, Solcia E, Longnecker DS, Capella C, Sobin LH. Histological typing of exocrine pancreas. Berlin. Springer; 1996. 2. Goh BK, Tan YM, Cheow PC, et al. Cystic neoplasms of the pancreas with mucin-production. Eur J Surg Oncol. 2005;31:282–287. 3. Goh BK, Tan YM, Chung YF, et al. A review of mucinous cystic neoplasms of the pancreas defined by ovarian-type stroma: clinicopathologic features of 344 patients. World J Surg. 2006;30(12):2236–45. 4. Longnecker DS, Adler G, Hruban RH, Kloppel G. Intraductal papillary-mucinous neoplasms of the pancreas. In: Hamilton SR, Bltonen LA, editors. Pathology and genetics of tumours of the digestive system. Lyon: IARC Press; 2000:237e40. 5. Sohn TA, Yeo CJ, Cameron JL, et al. Intraductal papillary mucinous neoplasms of the pancreas: anupdated experience. Ann Surg. 2004;239(6):788–97. 6. Wada K, Kozarek RA, Traverso LW. Outcomes following resection of invasive and noninvasiveintraductal papillary mucinous neoplasms of the pancreas. Am J Surg. 2005;189(5):632–6. 7. Sohn TA, Yeo CJ, Cameron JL, Iacobuzio-Donahue CA, Hruban RH, Lillemoe KD. Intraductal papillary mucinous neoplasms of the pancreas: an increasingly recognized clinicopathologic entity. Ann Surg. 2001;234(3):313–21. 8. Schnelldorfer T, Sarr MG, Nagorney DM, et al. Experience with 208 resections for intraductal papillary mucinous neoplasm of the pancreas. Arch Surg. 2008;143(7):639–46. 9. Nara S, Shimada K, Sakamoto Y, Esaki M, Kosuge T, Hiraoka N. Clinical significance of frozen section analysis during resection of intraductal papillary mucinous neoplasm: should a positive pancreatic margin for adenoma or borderline lesion be resected additionally? J Am Coll Surg. 2009;209:614–621. 10. Crippa S, Ferna´ndez-del Castillo C, Salvia R, et al. Mucin-producing neoplasms of the pancreas: an analysis of distinguishing clinical and epidemiologic characteristics. Clin Gastroenterol Hepatol. 2010;8:213–219. 11. Salvia R, Fernandez-del Castillo C, Bassi C, et al. Main-duct intraductal papillary mucinous neoplasms of the pancreas: clinical predictors of malignancy and long-term survival following resection. Ann Surg. 2004;239(5):678–85. 12. Maire F, Hammel P, Terris B, et al. Prognosis of malignant intraductal papillary mucinous tumours of the pancreas after surgical resection. Comparison with pancreatic ductal adenocarcinoma. Gut. 2002;51(5):717–22. 13. Wasif N, Bentrem DJ, Farrell JJ, Ko CY, Hines OJ, Reber HA, et al. Invasive intraductal papillary mucinous neoplasm versus sporadic pancreatic adenocarcinoma: a stage-matched comparison of outcomes. Cancer. 2010;116(14):3369–77. 14. Yamaguchi K, Kanemitsu S, Hatori T, et al. Pancreatic ductal adenocarcinoma derived from IPMN and pancreatic ductal adenocarcinoma concomitant with IPMN. Pancreas. 2011;40(4):571–80. 15. Waters JA, Schnelldorfer T, Aguilar-Saavedra JR, et al. Survival after resection for invasive intraductal papillary mucinous neoplasm and for pancreatic adenocarcinoma: a multi-institutional comparison according to American Joint Committee on Cancer Stage. J Am Coll Surg. 2011;213(2):275–83. 16. Yopp AC, Katabi N, Janakos M, et al. Invasive carcinoma arising in intraductal papillary mucinous neoplasms of the pancreas: a matched control study with conventional pancreatic ductal adenocarcinoma. Ann Surg. 2011;253(5):968–74. 17. Poultsides GA, Reddy S, Cameron JL, et al. Histopathologic basis for the favorable survival after resection of intraductal papillary mucinous neoplasm-associated invasive adenocarcinoma of the pancreas. Ann Surg. 2010;251(3):470–6.

18. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration. BMJ. 2009;339:b2700. 19. Japan Pancreas Society. Classification of pancreatic carcinoma. 2nd English ed. Tokyo: Kanehara & Co, Ltd; 2003. 20. Sobin LH, Wittekind C; International Union Against Cancer. TNM classification of malignant tumours. 6th ed. New York: Wiley-Liss; 2002. 21. Fleming ID, Cooper JS, Henson D, et al. AJCC cancer staging manual. 5th ed. New York: Lippincott-Raven; 1997. 22. Schlesselman J, Stolley P. Case-control studies, design, conduct, analysis. New York: Oxford University Press; 1982. 23. Shimada K, Sakamoto Y, Sano T, Kosuge T, Hiraoka N. Invasive carcinoma originating in an intraductal papillary mucinous neoplasm of the pancreas: a clinicopathologic comparison with a common type of invasive ductal carcinoma. Pancreas. 2006;32(3):281–7. 24. Woo SM, Ryu JK, Lee SH, Yoo JW, Park JK, Kim YT, et al. Survival and prognosis of invasive intraductal papillary mucinous neoplasms of the pancreas: comparison with pancreatic ductal adenocarcinoma. Pancreas. 2008;36(1):50–5. 25. Murakami Y, Uemura K, Sudo T, Hayashidani Y, Hashimoto Y, Nakashima A, et al. Invasive intraductal papillary-mucinous neoplasm of the pancreas: comparison with pancreatic ductal adenocarcinoma. J Surg Oncol. 2009;100(1):13–8. 26. Mino-Kenudson M, Ferna´ndez-del Castillo C, Baba Y, et al. Prognosis of invasive intraductal papillary mucinous neoplasm depends on histological and precursor epithelial subtypes. Gut 2011;60(12):1712–20. 27. Chinn S. A simple method for converting an odds ratio to effect size for use in meta-analysis. Stat Med 2000;19:3127–3131. 28. Tierney JF, Stewart LA, Ghersi D, Burdett S, Sydes MR. Practical methods for incorporating summary time-to-event data into meta-analysis. Trials. 2007;8:16. 29. Dickersin K, Berlin JA. Meta-analysis: state-of-the-science. Epidermiol Rev. 1992;14:154–176. 30. Kang MJ, Lee KB, Jang JY, Kwon W, Park JW, Chang YR, et al. Disease spectrum of intraductal papillary mucinous neoplasm with an associated invasive carcinoma invasive IPMN versus pancreatic ductal adenocarcinoma-associated IPMN. Pancreas. 2013;42(8):1267–74. 31. Sadakari Y, Ohuchida K, Nakata K, et al. Invasive carcinoma derived from the nonintestinal type intraductal papillary mucinous neoplasm of the pancreas has a poorer prognosis than that derived from the intestinal type. Surgery. 2010;147(6):812–7. 32. Goh BK, Tan YM, Cheow PC, Chung YF, Chow PK, Wong WK, et al. Outcome of distal pancreatectomy for pancreatic adenocarcinoma. Dig Surg. 2008;25(1):32–8. 33. Le H, Ziogas A, Rhee JM, Lee JG, Lipkin SM, Zell JA. A population-based, descriptive analysis of malignant intraductal papillary mucinous neoplasms of the pancreas. Cancer Epidemiol Biomarkers Prev. 2008;17(10):2737–41. 34. Goh BK, Thng CH, Tan DM, et al. Evaluation of the Sendai and 2012 International Consensus Guidelines based on initial crosssectional imaging findings for the management of mucinous cystic lesions of the pancreas: a single institution experience with 114 surgically-treated patients. Am J Surg. Epub 17 Jan 2014. 35. Winter JM, Cameron JL, Campbell KA, et al. 1423 pancreaticoduodenectomies for pancreatic cancer: a single-institution experience. J Gastrointest Surg. 2006;10(9):1199–210. 36. Conlon KC, Klimstra DS, Brennan MF. Long-term survival after curative resection for pancreatic ductal adenocarcinoma. Clinicopathologic analysis of 5-year survivors. Ann Surg. 1996; 223(3):273–279.

Systematic Review and Meta-Analysis 37. Konstantinidis IT, Vinuela EF, Tang LH, et al. Incidentally discovered pancreatic intraepithelial neoplasia: what is its clinical significance? Ann Surg Oncol. 2013;20:3643–7. 38. Goh BK, Tan DM, Thng CH, et al. Are the Sendai and Fukuoka consensus guidelines for mucinous neoplasms of the pancreas useful in the initial triage of all suspected pancreatic cystic neoplasms? A single institution experience with 317 surgicallytreated patients. Ann Surg Oncol. Epub 7 Feb 2014. 39. Goh BK, Tan DM, Ho MF, Lim KH, Chung AY, Ooi LL. Utility of the Sendai consensus guidelines for branch-duct intraductal papillary mucinous neoplasms: a systematic review. J Gastrointestin Surg. In press. 40. Fritz S, Fernandez-del Castillo C, Mino-Kenudson M, et al. Global genomic analysis of intraductal papillary mucinous neoplasms of the pancreas reveals significant molecular differences compared to ductal adenocarcinoma. Ann Surg. 2009;249(3):440–7.

41. Tanaka M, Chari S, Adsay V, et al. International consensus guidelines for management of intraductal papillary mucinous neoplasms and mucinous cystic neoplasms of the pancreas. Pancreatology. 2006;6:17–32. 42. Salvia R, Partelli S, Crippa S, et al. Intraductal papillary mucinous neoplasms of the pancreas with multifocal involvement of branch ducts. Am J Surg. 2009;198(5):709–14. 43. Oettle H, Post S, Neuhaus P, et al. Adjuvant chemotherapy with gemcitabine vs observation in patients undergoing curative-intent resection of pancreatic cancer: a randomized controlled trial. JAMA 2007;297(3):267–77. 44. Oettle H, Neuhaus P, Hochhaus A, et al. Adjuvant chemotherapy with gemcitabine and long-term outcomes among patients with resected pancreatic cancer: the CONKO-001 randomized trial. JAMA. 2013;310(14):1473–1481.