Studies selected

The results of the systematic search of the literature in accordance with the PRISMA statement are reported in Figure 1. The search identified a total of 1820 potential articles: 818 from the Medline/PubMed database, 543 from the ISI Web of Science and 458 from Scopus. One additional study was included by manually searching the references. After selection process only 6 studies were eligible for quality assessment and quantitative synthesis. The characteristics of the studies selected are summarized in Supplementary Table 1.

Supplementary Figure 1 (A-H) : Panel A: Forest plot showing the risk difference for the distant metastatic rate between small (≤ 20 mm) and large (> 20 mm) non-functioning pancreatic endocrine tumors. Panel B: Forest plot showing the risk difference for the poorly-differentiated carcinoma (G3) rate between small (≤ 20 mm) and large (> 20 mm) non-functioning pancreatic endocrine tumors. Panel C: Forest plot showing the risk difference for the G2-3 neoplasm rate between small (≤ 20 mm) and large (> 20 mm) non-functioning pancreatic endocrine tumors. Panel D: Forest plot showing the risk difference for the metastatic stage (Stage III-IV according to ENETS-TNM) rate between small (≤ 20 mm) and large (> 20 mm) non-functioning pancreatic endocrine tumors. Panel E: Forest plot showing the risk difference for the vascular microscopic invasion rate between small (≤ 20 mm) and large (> 20 mm) non-functioning pancreatic endocrine tumors. Panel F: Forest plot showing the risk difference for the 5-year death (any reason) rate between small (≤ 20 mm) and large (> 20 mm) non-functioning pancreatic endocrine tumors. Panel G: Forest plot showing the risk difference for the 5-year death (disease related) rate between small (≤ 20 mm) and large (> 20 mm) non-functioning pancreatic endocrine tumors. Panel H: Forest plot showing the risk difference for the 5-year recurrence rate between small (≤ 20 mm) and large (> 20 mm) non-functioning pancreatic endocrine tumors. Legend: Small pNENs: non-functioning pancreatic endocrine tumors ≤ 20 mm; Large pNENs: non-functioning pancreatic endocrine tumors 20 mm; M-H: Mantel-Haenszel method; 95% CI: the 95% confidence interval; I2: between study heterogeneity according to the Higgins’s test. Blue square: risk difference of each study; Size of square: weight of each study in the analysis; Solid black line: the 95% confidence interval for each study; Black diamond: the pooled risk difference.

Supplementary Figure 2 Asymmetry due to the “small sample size” effect for distant metastases, tested using Egger analysis. Legend : Vertical axis: the measure of the effect divided by standard error (SND); Horizontal axis: the precision of each studies (1/standard error); Red line: regression line; Blue circle: each included study; Vertical red line: The intercept and its 90% confidence interval.

Supplementary Table 1. Characteristics of the studies included

Authors / Affiliation/Hospital / Year / WHO classification / Study Design / pNENs N (%) / Age / M/F
ratio / Type of Surgery / Risk parameters reported / MINORS score
<20 mm / 20mm / Typical / Atypical
La Rosa S et al.12 / Multicentric,
University of Pavia and Milano
Varese Hospital, Italy / 1996 / WHO 2000 / Retrospective consecutive / 10 (18.2) / 45 (81.8) / ^ / ^ / ^ / ^ / N status, M status, vascular invasion, Ki67,
OS, DSS, DFS / 13/24
Nomura et al.13 / University of Nagoya, Japan / 2009 / WHO 2000 / Retrospective
consecutive / 8 (53.3) / 7 (46.7) / 56.9 ± 13.7 / 1.14 / 12 (80%) / 3 (20%) / N status, M status, vascular invasion, Ki67,
OS, DSS, DFS / 11/24
Kim et al.6 / National University Hospital, Jeju
Samsung Medical center, Seul, Korea / 2012 / WHO 2000 / Retrospective consecutive / 51 (40.8) / 74 (59.2) / 54 ± 10.1 / 0.98 / 82 (65.6%) / 43 (34.4%) / N status, M status, Ki67, AJCC-TNM / 14/24
Kuo et al.4 / National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) database, US / 2013 / Arbitrary / Retrospective not consecutive / 263 (19.2) / 1108 (80.8) / 55.9 ± 0.6 / 1.05 / 1201 (87.6%) / 170 (12.4%) / N status, M status,
OS, DSS / 13/24
Kishi et al.14 / University of Tokio, Japan / 2014 / WHO 2010 / Retrospective consecutive / 27 (38.1) / 44 (61.9) / 56 ± 10.7 / 0.57 / 58 (81.7%) / 13 (18.3%) / N status, M status, vascular invasion, ENETS-TNM, DFS / 14/24
Lombardi et al.15 / University of Pisa, Italy / 2014 / WHO 2010 / Retrospective consecutive / 23 (38.3) / 37 (61.7) / 59.5 ± 14.8 / 1.31 / 59 (98.3%) / 1 (0.7%) / N status, M status, Ki67 ENETS- TNM / 13/24
Total / 382 (22.5) / 1315 (77.5) / 55.9 ± 2.4 / 1.05 / 1412 (86%) / 230 (14%) / 13/24*

Legend: pNENs: Non-Functioning Pancreatic Neuroendocrine Tumors; M: male; F: Female; ^: data not extractable; WHO: classification system according to the World Health Organization; N status: lymph nodal status; M status: distant metastases; OS: overall survival; DSS: disease-specific survival; DFS: disease-free survival; AJCC-TNM: staging according to the American Joint Committee on Cancer staging system 7th edition; ENETS-TNM: staging according to the European Neuroendocrine Tumor Society; *: median value; MINORS: Methodological Index for Non-Randomized Studies score.