Received: 18 October 2019 | Accepted: 27 October 2019

DOI: 10.1002/jso.25753

RESEARCH ARTICLE

Duodenal neuroendocrine tumors: Impact of tumor size

and total number of lymph nodes examined

Xu‐Feng Zhang MD, PhD1,2 | Xiao‐Ning Wu MD1 | Diamantis I. Tsilimigras MD2 |

George Poultsides MD 4
| Flavio Rocha MD 5
| Daniel E. Abbott MD 6
|
Ryan Fields MD 7
| Kamran Idrees MD 8
| Cliff Cho MD 9
| Shishir K. Maithel MD3 |
Timothy M. Pawlik MD, MPH, PhD 2
| other members of the US Neuroendocrine Tumor
Study Group

1
Department of Hepatobiliary, Surgery and
Institute of Advanced Surgical Technology and Abstract
Engineering, The First Affiliated Hospital of Background: The current study sought to investigate the impact of tumor size and
Xi’an Jiaotong University, Xi’an, China
2 total number of LN examined (TNLE) on the incidence of lymph node metastasis
Division of Surgical Oncology, The Ohio State
University Wexner Medical Center and James (LNM) among patients with duodenal neuroendocrine tumor (dNET).
Comprehensive Cancer Center, Columbus, OH
Methods: Patients who underwent curative resection for dNETs between 1997‐2016
3
Division of Surgical Oncology, Department of
Surgery, Winship Cancer Institute, Emory were identified from 8 high‐volume US centers. Risk factors associated with overall
University, Atlanta, GA survival and LNM were identified and the optimal cut‐off of TNLE relative to LNM
4
Department of Surgery, Stanford University,
was determined.
Palo Alto, CA
5
Department of Surgery, Virginia Mason
Results: Among 162 patients who underwent resection of dNETs, median patient age
Medical Center, Seattle, WA was 59 (interquartile range [IQR], 51‐68) years and median tumor size was 1.2 cm
6
Department of Surgery, University of (IQR, 0.7‐2.0 cm); a total of 101 (62.3%) patients underwent a concomitant LND at
Wisconsin School of Medicine and Public
Health, Madison, WI the time of surgery. Utilization of lymphadenectomy (LND) increased relative to
7
Department of Surgery, Washington tumor size (≤1 cm:52.2% vs 1‐2 cm:61.4% vs >2 cm:93.8%; P < .05). Similarly, the
University School of Medicine, St. Louis, WI
incidence of LNM increased with dNET size (≤1 cm: 40.0% vs 1‐2 cm:65.7% vs
8
Division of Surgical Oncology, Department of
Surgery, Vanderbilt University, Nashville, TN >2 cm:80.0%; P < .05). TNLE ≥ 8 had the highest discriminatory power relative to the
9
Division of Hepatopancreatobiliary and incidence of LNM (area under the curve = 0.676). On multivariable analysis, while
Advanced Gastrointestinal Surgery,
LNM was not associated with prognosis (hazard ratio [HR] = 0.9; 95% confidence
Department of Surgery, University of
Michigan, Ann Arbor, MI intervals [95%CI], 0.4‐2.3), G2/G3 tumor grade was (HR = 1.5; 95%CI, 1.0‐2.1).
Conclusions: While the incidence of LNM directly correlated with tumor size, patients
Correspondence
Timothy M. Pawlik, MD, MPH, PhD, FACS, with dNETs ≤ 1 cm had a 40% incidence of LNM. Regional lymphadenectomy of a
FRACS (Hon.), Professor and Chair,
least 8 LN was needed to stage patients accurately.
Department of Surgery, The Urban Meyer III
and Shelley Meyer Chair for Cancer Research,
Professor of Surgery, Oncology, Health KEYWORDS
Services Management and Policy, The Ohio Ampulla, neuroendocrine tumor, Duodenum, prognosis, surgery
State University, Wexner Medical Center, 395
W. 12th Ave., Suite.
Email: [email protected]

J Surg Oncol. 2019;1–9. wileyonlinelibrary.com/journal/jso © 2019 Wiley Periodicals, Inc. | 1

2 | ZHANG ET AL.

1 | INTRODUCTION scoring system16 and classified as grade G1 (Ki‐67 ≤2%), G2 (Ki‐67 3%‐
20%), and G3 (Ki‐67 >20%) according to the WHO 2010 classification.17
Duodenal neuroendocrine tumors (dNETs) represent up to 3% of all Overall survival (OS) was calculated from the date of surgery to the
primary duodenal malignancies and 2%‐3% of all gastrointestinal date of death or date of last follow‐up. Tumor recurrence was
neuroendocrine tumors (NETs).1-3 The incidence of dNETs has determined by suspicious imaging finding or biopsy‐proven tumor.
increased from 0.027/100 000 in 1983 to 1.1/100 000 in 2010, perhaps
reflecting the increased use of gastrointestinal endoscopy.4,5 According
2.2 | Statistical analysis
to the latest population study in the United States, dNETs present most
often in the 6th decade of life and males are slightly more predominant Categorical variables were reported as totals and percentages and
6
than females. In turn, resection of dNETs generally represents the most compared by χ2 test or Fisher exact test as appropriate. Continuous
common modality associated with the best chance at cure. variables were expressed as median with interquartile ranges (IQRs) and
Although dNETs are usually small on presentation (ie, 75% of compared using the Mann‐Whitney U test. Kaplan‐Meier survival curves
cases ≤20 mm), regional lymph node (LN) metastasis has been were plotted and compared using the logrank test. The receiver‐
reported in up to 40%‐80% of cases.7-10 Current treatment guide- operating characteristic curve analysis was used to investigate the
lines recommend treatment strategies for dNETs that are similar to discriminatory ability of TNLE relative to the number of LNM. Risk
gastric NETs for nonfunctional tumors and similar to pancreatic NETs factors associated with OS or LNM were identified by using Cox‐
for functional dNETs.11 For example, the European Neuroendocrine proportional hazard regression models or Logistic regression models,
Tumor Society (ENETS) consensus guidelines generally recommend respectively. Results were reported as hazard ratios (HRs) or odds ratio
surgical treatment for large (>20 mm) and/or metastatic dNETs, as (OR) and 95% confidence intervals (95%CI). A P‐value < 0.05 (two‐tailed)
well as periampullary dNETs.11 More often dNET cases are now was considered statistically significant for all analyses. Statistical analyses
diagnosed, however, incidentally when the tumors are small in size were performed using SPSS 22.0 (IBM, Chicago, IL).
(<10 mm), well‐differentiated, limited to the mucosa and submucosa,
and nonfunctioning. In turn, endoscopic local excision may be
3 | RESULTS
increasingly considered for these early‐stage tumors rather than
surgical resection.12 The impact of not staging the nodal basin in the
3.1 | Baseline characteristics
setting of endoscopic excision remains poorly defined. Unfortunately,
data on the clinical characteristics, tumor biology, treatment, and Among 162 patients with dNETs, median patient age was 59 (IQR, 51‐68)
prognosis of patients with dNETs have been limited largely to small, years and 53.7% (n = 87) of patients were male (Table 1). Roughly one‐in‐
single‐center case series.12-14 Therefore, the objective of the current five (n = 34/162, 21.0%) patients had a functional tumor; gastrinoma
study was to define the outcomes of patients who underwent (n = 31/34, 91.2%) was the predominant tumor type. The overwhelming
curative‐intent resection for periampullary and nonampullary dNETs majority of patients had no designated syndrome (n = 146, 90.1%),
using a large, multi‐institutional database. Specifically, the aim was to whereas a small subset of individuals had multiple endocrine neoplasia
characterize potential risk factors associated with the presence of type 1 (n = 8, 4.9%) or neurofibromatosis syndrome (n = 4, 2.5%).
lymph node metastasis (LNM), as well as identify the minimal number Abdominal pain (47.5%) was the most common symptom. The majority
of LN needed to stage patients optimally. (n = 127, 78.4%) of patients underwent an open surgical procedure.
Surgical procedures included transduodenal resection (n = 25, 15.4%),
partial duodenectomy (n = 20, 12.3%), segmental duodenectomy (n = 33,
2 | METHODS
20.4%), and pancreaticoduodenectomy (PD) (n = 52, 32.1%); 30 (18.5%)
patients underwent an endoscopic resection. Based on final pathological
2.1 | Study cohort and data collection
assessment, median tumor size was 1.2 cm (IQR, 0.7‐2.0 cm). A total of
Patients who underwent surgical resection for dNETs between 1997 101 (62.3%) patients underwent a concomitant lymphadenectomy (LND)
and 2016 were identified from the US Neuroendocrine Tumor Study at the time of surgery (transduodenal resection, n = 11; partial
Group (US‐NETSG).15 All patients were diagnosed with dNETs, which duodenectomy, n = 14; segmental duodenectomy, n = 26; PD, n = 50);
were confirmed by histological examination. Patients who presented median TNLE was 10 (IQR, 3‐16). Among patients who underwent LND
with distant metastasis or underwent cytoreductive/palliative (R2) (n = 101), at least one LNM was identified in 61 (60.4%) patients. The
resection were excluded. The study was approved by the Institutional overall incidence of procedure‐related complications was 48.1% (n = 78)
Review Boards at each participating institution. with 22.8% (n = 37) of patients experiencing a severe Clavien‐Dindo III‐V
A standardized datasheet was utilized to collect the demographic, complication.
clinical, and pathologic data at each institution. Largest tumor size,
primary tumor location, the total number of LNs examined (TNLE), the
3.2 | Periampullary versus nonampullary dNETs
number of LNM, perineural invasion, lymphovascular invasion, and
surgical margin status were determined based on the final pathological In assessing the cohort, 127 (78.4%) patients had a dNET located in
report. The tumors were staged according to the eighth TMN stage the duodenum away from ampulla of Vater, whereas 35 (21.6%)
ZHANG ET AL. | 3

T A B L E 1 Clinicopathologic characteristics and surgical procedures

Overall (n = 162) Periampullary dNETs (n = 35) Nonampullary dNETs (n = 127) P‐value
Age (y) 59 (51‐68) 56 (50‐70) 59 (52‐68) 0.559
Sex 0.340
Male 87 (53.7%) 16 (45.7%) 16 (45.7%)
Female 75 (46.3%) 19 (54.3%) 19 (54.3%)
Nonfunctional tumor 128 (79.0%) 30 (85.7%) 98 (77.2%) 0.352
Functional tumor 34 (21.0%) 5 (14.3%) 29 (22.8%) 0.326
Gastrinoma 31 (19.1%) 4 (11.4%) 27 (21.3%)
Somatostatinoma 2 (1.2%) 1 (2.9%) 1 (0.8%)
Glucagonoma 1 (0.6%) ‐ 1 (0.8%)
Genetic syndrome 0.059
None 146 (90.1%) 30 (85.7%) 116 (91.3%)
MEN 1 8 (4.9%) 1 (2.9%) 7 (5.5%)
Neurofibromatosis 4 (2.5%) 3 (8.6%) 1 (0.8%)
NA 4 (2.5%) 1 (2.9%) 6 (2.4%)
Symptomatic 121 (74.7%) 28 (80.0%) 93 (73.2%) 0.515
Abdominal pain 77 (47.5%) 24 (68.6%) 53 (41.7%) 0.007
Clinical jaundice 10 (6.2%) 8 (22.9%) 2 (1.6%) <0.001
Gastrointestinal bleeding 21 (13.0%) 2 (5.7%) 19 (15.0%) 0.254
Diarrhea 33 (20.4%) 7 (20.0%) 26 (20.5%) 1.000
Nausea/vomiting 37 (22.8%) 11 (31.4%) 26 (20.5%) 0.182
Preoperative FNA 47 (29.0%) 14 (40.0%) 33 (26.0%) 0.141
Primary location
D1 76 (46.9%) ‐ 76 (59.8%)
D2 26 (16.0%) ‐ 26 (20.5%)
D1+D2 7 (4.3%) ‐ 7 (5.5%)
D3/D4 4 (2.5%) ‐ 4 (3.1%)
Surgery technique 0.267
Endoscopic 30 (18.5%) 4 (11.4%) 26 (20.5%)
Open 127 (78.4%) 30 (85.7%) 97 (76.4%)
Laparoscopic 4 (2.5%) ‐ 4 (3.1%)
Type of resection <0.001
Endoscopic resection 30 (18.5%) 4 (11.4%) 26 (20.5%)
Transduodenal resection 25 (15.4%) 9 (25.7%) 16 (12.6%)
Partial duodenectomy 20 (12.3%) ‐ 20 (15.7%)
Segmental duodenectomy 33 (20.4%) ‐ 33 (26.0%)
Pancreaticoduodenectomy 52 (32.1%) 21 (60.0%) 31 (24.4%)
Operation time (min) 200 (138‐306) 250 (154‐375) 196 (133‐302) 0.082
Blood loss (mL) 125 (25‐300) 200 (50‐500) 100 (20‐300) 0.074
Surgical margin 1.000
R0 131 (80.9%) 28 (80.0%) 103 (81.1%)
R1 23 (14.2%) 5 (14.3%) 18 (14.2%)
Largest tumor size (cm) 1.2 (0.7‐2.0) 1.8 (1.2‐2.5) 1.1 (0.7‐1.7) 0.019
Tumor number 0.044
Single 141 (87.0%) 34 (97.1%) 107 (84.3%)
Multiple 21 (13.0%) 1 (2.9%) 20 (15.7%)
Lymphadenectomy 101 (62.3%) 24 (68.6%) 77 (60.6%) 0.424
No. of lymph node examined 10 (3‐16) 15 (11‐21) 7 (2‐15) 0.005
Lymph nodes status 0.241
Negative 40 (39.6%) 7 (29.2%) 33 (42.9%)
Positive 61 (60.4%) 17 (70.8%) 44 (57.1%)
No. of positive lymph nodes 2 (1‐4) 4 (2‐6) 2 (1‐3) 0.285
(Continues)
4 | ZHANG ET AL.

TABLE 1 (Continued)

Overall (n = 162) Periampullary dNETs (n = 35) Nonampullary dNETs (n = 127) P‐value

WHO classification 0.104
G1 106 (65.4%) 22 (62.9%) 84 (66.1%)
G2 51 (31.5%) 10 (28.6%) 22 (32.3%)
G3 5 (3.1%) 3 (8.6%) 2 (1.6%)
AJCC T stage 0.018
T1 56 (34.6%) 5 (14.3%) 51 (40.2%)
T2 61 (37.7%) 17 (48.6%) 44 (34.6%)
T3 21 (13.0%) 8 (22.9%) 13 (10.2%)
T4 6 (3.7%) 2 (5.8%) 4 (3.1%)
Lymph‐vascular invasion 31 (19.1%) 11 (31.4%) 20 (15.7%) 0.184
Perineural invasion 12 (7.4%) 3 (8.6%) 9 (7.1%) 1.000
Postoperative morbidity 78 (48.1%) 18 (51.4%) 60 (47.2%) 0.705
Severe complication (III‐V) 37 (22.8%) 10 (28.6%) 27 (21.3%) 0.432
Length of stay (d) 7 (5‐11) 15 (7‐21) 7 (4‐11) 0.029
Abbreviations: AJCC, American Joint Committee on Cancer; dNETs, duodenal neuroendocrine tumors; MEN 1, multiple endocrine neoplasia type 1.

F I G U R E 1 A, overall survival of the

whole cohort (n = 162); B, overall survival
of patients with periampullary (n = 35) or
nonampullary (n = 127) duodenal
neuroendocrine tumors
ZHANG ET AL. | 5

patients had a periampullary dNET. Clinicopathologic characteris- T A B L E 2 Factors associated with overall survival after curative
tics and surgical procedures associated with nonperi‐ampullary and resection for duodenal neuroendocrine tumors (dNETs)
periampullary pNET tumors were largely comparable (Table 1). Univariate analysis Multivariable analysis
Perhaps not surprisingly, patients with periampullary dNETs were HR (95%CI) P value HR (95%CI) P value
more likely to present with clinical jaundice (22.9% vs 1.6%;
Sex (male vs female) 1.3 (0.5‐3.4) 0.535
P < .001) and abdominal pain (68.6% vs 41.7%; P = .007) versus
Functional status 0.4 (0.1‐1.7) 0.202
patients with nonampullary dNETs. In addition, more patients who
Symptomatic 0.5 (0.2‐1.1) 0.083
had a periampullary dNET underwent a PD (n = 21, 60.0%)
Genetic syndrome 0.4 (0.1‐20.0) 0.309
compared with only 24.4% (n = 31) of patients with a nonampullary
dNET (P < .001). In turn, patients with a periampullary dNET had a AJCC T categories
T1 Ref. Ref.
higher TNLE compared with nonperi‐ampullary dNET (median 15 vs
T2 0.8 (0.3‐2.9) 0.835 0.4 (0.1‐1.7) 0.425
7, respectively; P = .005)(Table 1). In addition, patients with T3‐T4 1.9 (1.1‐6.6) 0.024 0.9 (0.2‐4.1) 0.924
periampullary dNETs were more likely to have a single tumor that
Multiple lesions 0.7 (0.2‐3.0) 0.623
was larger in size versus patients who had nonampullary tumors
Tumor size (cm)
(both P < .05). In contrast, receipt of LND, as well as incidence and ≤1 Ref.
number of LNM, were not different among patients with ampullary 1‐2 0.9 (0.3‐2.5) 0.831
versus nonampullary dNETs (Table 1). >2 0.9 (0.3‐2.8) 0.913
Tumor location 0.944
Duodenum Ref.
3.3 | Long‐term survival Ampulla of Vater 1.0 (0.7‐1.4)

With a median follow‐up of 27.2 (IQR, 8.9‐57.0) months, 1‐, 3‐, and 5‐ Surgery technique 0.732
Endoscopic Ref.
year OS was 94.2%, 86.9%, and 84.7%, respectively (Figure 1A). Of
Open/laparoscopic 1.3 (0.3‐5.7)
note, OS was similar among patients with periampullary versus
Surgical margin 0.537
nonampullary dNETs (5‐year OS, 70.8% vs 87.3%; P = .944) (Figure
R0 Ref.
1B). In the subset of patients who underwent PD, OS was also R1 0.6 (0.1‐2.7)
comparable among patients with periampullary (n = 21) and non- Lymphadenectomy 0.874
ampullary (n = 31) dNETs (5‐year OS, 81.0% vs 82.0%; P = 1.000). On No Ref.
multivariable analysis, only G2/G3 WHO classification was asso- Yes 0.9 (0.4‐2.3)
ciated with a higher risk of worse OS (Reference G1, HR, 1.5; 95%CI, Nodal metastasis 0.5 (0.2‐1.7) 0.260
1.0‐2.1; P = .032). In contrast, tumor location, size, and number, WHO classification
American Joint Committee on Cancer (AJCC) T stages, nodal status, G1 Ref. Ref.
as well as treatment procedures were not associated with long‐term G2/G3 1.7 (1.1‐3.2) 0.025 1.5 (1.0‐2.1) 0.032

survival of dNETs patients (Table 2). Lymph‐vascular 2.5 (0.5‐13.9) 0.282

invasion
Among the 18 (11.1%) patients who developed tumor recur-
rence during follow‐up, 9 (50%) developed hepatic recurrence Perineural invasion 2.6 (0.4‐15.6) 0.296

only, whereas 8 (44.4%) patients developed loco‐regional resec- Abbreviation: 95%CI, 95% confidence intervals; AJCC, American Joint
Committee on Cancer; HR, hazard ratio.
tion site and/or LN recurrence; one (5.6%) patient had both loco‐
regional and distant metastasis. Tumor recurrence was no
different among patients with periampullary versus nonampullary with tumors ≤ 1 cm, 1 to 2 cm and > 2 cm (LNM, 40.0% vs. 65.7% vs.
dNETs or among patients undergoing different surgical procedures 80.0%, P = 0.003) (Figure 2C).
(both P > 0.05). The likelihood to identify LNM was associated with TNLE.
Specifically, TNLE ≥ 8 was associated with a higher incidence of
identifying LNM (TNLE < 8: 20/45, 44.4% vs. TNLE ≥ 8: 41/56,
3.4 | Tumor size and nodal metastasis
73.2%; P = .004), as well as a higher number of LNM identified
As tumor size is a key indicator in defining the T category for dNETs in (median number of LNM, TNLE ≥ 8: 2 [IQR, 1‐5], vs TNLE < 8: 1
the AJCC staging manual,16 the impact of tumor size on the proportion of [IQR, 1‐2], P < 0.001). In addition, TNLE ≥ 8 had the highest
LND, as well as the incidence of nodal metastasis, was further analyzed. discriminatory power relative to the incidence of LNM (area
Of note, LND were performed among almost all (93.8%) patients with under the curve [AUC] 0.676, sensitivity 67.2%, specificity =
dNET >2 cm compared with 61.4% of patients with tumors of 1 to 2 cm 65.0%) (Figure 2D). On multivariable analysis, tumor size (1‐2 cm
and 52.2% of patients with tumors ≤1 cm (both P < 0.01) (Figure 2A). vs ≤1 cm (OR, 2.8; 95%CI, 1.0‐8.0; P = .048), >2 cm vs ≤1 cm (OR,
Patients with larger tumors were also more likely to have a higher TNLE 4.6; 95%CI, 1.4‐15.0; P = 0.012), as well as TNLE ≥ 8 vs <8 (OR, 3.6;
(median TNLE > 2 cm 14 vs 1‐2 cm 7 vs ≤1 cm 5; P < 0.001) (Figure 2b). 95%CI, 1.4‐9.2; P = .007) were associated with the likelihood of
In addition, the incidence of LNM incrementally increased among patients identifying LNM (Table 3).
6 | ZHANG ET AL.

F I G U R E 2 The proportion of lymphadenectomy (LND) (A) and number of LND (B) among patients with different tumor sizes. C, the
incidence of lymph node metastasis (LNM) among patients with different tumor sizes. D, Receiver operative characteristics (ROC) analysis
illustrating that total number of lymph nodes examined (TNLE) ≥8 had the highest discriminatory power relative to LNM

4 | D IS C U S S IO N one LNM. Perhaps not surprisingly, LND was utilized incrementally

more often among patients who had larger dNETs (≤1 cm, 52.2% vs
The AJCC TNM staging system for NET incorporates tumor size, 1‐2 cm, 61.4% vs >2 cm, 93.8%) and the incidence of LNM also
nodal status, and distant metastasis to stratify outcomes of patients increased (≤1 cm, 40.0% vs 1‐2 cm, 65.7% vs >2 cm, 80.0%). We
with dNETs. The impact of tumor size and nodal status on prognosis noted that the TNLE with the best discriminatory power to ensure
has, however, not been fully investigated, partially due to the rarity adequate staging of the nodal basin was a TNLE ≥ 8 (AUC, 0.676). In
of this disease. The current study was important as we utilized a large fact, on multivariable analysis, in addition to tumor size, TNLE ≥ 8 vs
multi‐institutional database to demonstrate that long‐term survival <8 was independently associated with the likelihood of identifying
of patients with dNETs was comparable among patients with peri‐ or LNM. Collectively, data from the current study serves to highlight
nonampullary dNETs. Rather than location or size, tumor grade G2/ that patients with dNET generally have a favorable prognosis,
G3 was the main risk factor associated with worse prognosis. Among however, the incidence of LNM may be high. Tumor size was
patients who underwent LND, three out of five patients had at least associated with LNM, especially among those patients with a dNET
ZHANG ET AL. | 7

T A B L E 3 Factors associated with lymph node metastasis (LNM) of comparable with previous data reported in the literature (66%‐
duodenal neuroendocrine tumors (dNETs) 93.8%).8,19,20 Interesting, the impact of LNM on long‐term outcome
Univariate analysis Multivariable analysis has been controversial. While some studies have suggested that LNM

OR (95%CI) P value OR (95%CI) P value was associated with a worse prognosis,7,16 data from the current
study, as well as several other previous reports, have not demon-
Sex (female vs male) 2.5 (1.1‐5.6) 0.030 2.6 (1.0‐6.4) 0.046
strated an association of LNM with long‐term prognosis.8,9,21 Given
Functional status
the low accuracy of preoperative imaging to detect LNM among
Symptomatic 1.1 (0.4‐3.1) 0.805
dNETs patients, most surgeons advocate for routine LND at the time
Genetic syndrome 2.1 (0.5‐8.2) 0.297
of surgical resection of dNETs.12,22 To this point, LNM was present in
Multiple lesions 1.3 (0.4‐4.2) 0.626 over one‐half of all patients with dNETs (60.4%). Perhaps of more
Tumor size (cm) interest, the data clearly demonstrated a strong correlation with
≤1 Ref. Ref.
dNET size and the likelihood of LNM, as patients with tumor >2 cm
1 to 2 2.9 (1.1‐7.6) 0.033 2.8 (1.0‐8.0) 0.048
(80.0%) or 1 to 2 cm (65.7%) in size had a much higher incidence of
>2 6.0 (2.0‐18.4) 0.002 4.6 (1.4‐15.0) 0.012
LNM than patients with tumors ≤1 cm (40.0%) (P = .003). Margonis
Tumor location 0.819
Duodenum Ref. et al10 had similarly reported a higher incidence of LNM among NET
Ampulla of Vater 0.8 (0.6‐1.1) tumors sized >1.5 cm,10 while Burke et al noted a difference in the
Total number of 0.004 0.007 incidence of LNM using tumor size using >2 cm as the cut‐off.23
LNs examined While 1 cm is currently used as the cut‐off size to differentiate T1
<8 Ref. Ref. from T2 in the AJCC staging manual,11,16 there was still a relatively
≥8 3.4 (1.5‐7.9) 3.6 (1.4‐9.2)
high incidence of LNM even among patients with small tumors
WHO classification (≤1 cm, 40.0%). Whereas surgical resection and lymphadenectomy
G1 Ref.
has traditionally been recommended for only tumors >1 cm, data in
G2/G3 1.8 (0.6‐5.7) 0.328
the current study suggest that patients with dNETs ≤1 cm should also
Abbreviations: 95%CI, 95% confidence intervals; OR, odds ratio.
be considered for LND rather than local excision alone.
The minimal number of LNs needed to examine to achieve accurate
>1 cm, and the ability to identify LNM depended on the adequacy of staging of the nodal basin for patients with dNETs has not been defined
the LND with TNLE ≥ 8 being the optimal TNLE to stage patients with in the latest NCCN guidelines.11,16 The topic of TNLE has been an area
dNETs. of interest for several other hepatopancreatic diseases. In fact, TNLE
Duodenal NETs have been anatomically classified most often as among patients with pancreatic and small bowel NETs relative to
periampullary versus nonampullary. Perhaps not surprisingly, periam- recurrence‐free survival and overall survival has been examined by our
pullary tumor location was associated with higher likelihood of group and others.15,24 Using a large multi‐institutional database with
abdominal pain (80% vs 73%) and clinical jaundice (22.9% vs 1.6%) external validation based on the surveillance, epidemiology, and end
versus nonampullary dNETs due to obstruction of bile and pancreatic results (SEER) registry, we demonstrated that TNLE ≥ 8 had the highest
ducts. Consistent with a previous population‐based study,18 the discriminatory power relative to recurrence‐free and overall survival
current study also noted that periampullary dNETs were larger at among patients with pNET who hadone to three LNM, and patients who
the time of presentation compared with nonampullary dNETs. Patients had ≥4 LNM in both a multi‐institutional data set and the SEER
with periampullary dNETs were also more likely to undergo PD database.15 In the current study, we similarly noted that TNLE ≥ 8 had
and had a higher number of LNs examined than patients with the strongest discriminatory power to ensure identification of possible
nonampullary dNETs. The incidence of LNM was, however, no LNM and optimally stage the nodal basin for patients with dNETs.
different among patients with periampullary versus nonampullary Specifically, TNLE ≥ 8 versus TNLE < 8 was associated with a higher
dNETs. Consistent with previous data, we also noted that tumor grade chance of identifying LNM (73.2% vs 44.4%), as well as a higher number
was associated with long‐term survival among patients with both of LNM identified. Taken together, LND as part of an operative
periampullary and nonampullary dNETs after curative resection.8 In procedure for patients with dNET should include examination of ≥8 LN
contrast, Randle et al18 failed to find an association of higher tumor to ensure adequate staging.
grade and worse survival among patients with dNETs—especially as Several limitations should be considered when interpreting data
related to patients who had periampullary versus nonampullary in the current study. While a multi‐institutional study increased the
dNETs. The reason for these disparate results are likely multifactorial sample size of the analytic cohort, patient selection, surgical
yet may be due to patients with periampullary dNET in the current technique, as well as utilization of LND and pathologic examination
study being more likely to undergo a more extensive resection, such as of LN, may have varied at different centers. All participatory centers
PD, compared with patients included in the study by Randle et al18 were, however, major hepatopancreatic institutions that followed
who more often underwent a local excision. standard state‐of‐the‐art care guidelines regarding the management
Prognosis following curative‐intent resection of dNET was of patients with NET. The current study also did not note a difference
generally very good with a 5‐year survival of 84.7%, which was of long‐term survival among patients with versus without LNM. The
8 | ZHANG ET AL.

reason for this may have been due to a lack of statistic power (Type II R E F E R E N CE S
error); however, other authors have noted a similar finding and
1. O'Toole D, Delle Fave G, Jensen RT. Gastric and duodenal
attributed the lack of prognostic impact to the indolent nature of
neuroendocrine tumours. Best Pract Res Clin Gastroenterol. 2012;26:
dNET and the very good overall prognosis.8,9,21 The current study 719‐735.
also only included patient who underwent curative‐intent resection 2. Jensen RT, Rindi G, Arnold R, et al. Well‐differentiated duodenal
and therefore patients receiving nonsurgical treatments were not tumor/carcinoma (excluding gastrinomas). Neuroendocrinology. 2006;
84:165‐172.
available for comparison.
3. Modlin IM, Lye KD, Kidd M. Carcinoid tumors of the stomach. Surg
In conclusion, patients with a dNET had an overwhelmingly Oncol. 2003;12:153‐172.
favorable prognosis after curative‐intent resection, despite the 4. Modlin IM, Lye KD, Kidd M. A 5‐decade analysis of 13,715 carcinoid
relatively high incidence of associated LNM. While the incidence of tumors. Cancer. 2003;97:934‐959.
5. Fitzgerald TL, Dennis SO, Kachare SD, Vohra NA, Zervos EE.
LNM directly correlated with tumor size, even patients with dNET
Increasing incidence of duodenal neuroendocrine tumors: incidental
≤1 cm had a 40% incidence of LNM. Regional lymphadenectomy of at discovery of indolent disease? Surgery. 2015;158:466‐471.
least eight LN was recommended to stage patients accurately. 6. Yao JC, Hassan M, Phan A, et al. One hundred years after "carcinoid":
Interestingly, the presence of LNM was not, however, associated with epidemiology of and prognostic factors for neuroendocrine tumors in
35,825 cases in the United States. J Clin Oncol. 2008;26:3063‐3072.
long‐term survival; rather, tumor grade was the factor that impacted
7. Masui T, Sato A, Nakano K, et al. Comparison of recurrence between
prognosis. Collectively, data from the current study should help to pancreatic and duodenal neuroendocrine neoplasms after curative
inform the care of patients with dNETs. resection: a single‐institution analysis. Ann Surg Oncol. 2018;25:528‐534.
8. Dogeas E, Cameron JL, Wolfgang CL, et al. Duodenal and ampullary
carcinoid tumors: size predicts necessity for lymphadenectomy.
A C K N O W L E D GM E N T J Gastrointest Surg. 2017;21:1262‐1269.
9. Untch BR, Bonner KP, Roggin KK, et al. Pathologic grade and tumor
Xu‐Feng Zhang and Xiao‐Ning Wu were supported by the Clinical size are associated with recurrence‐free survival in patients with
Research Award of the First Affiliated Hospital of the Xi’an Jiaotong duodenal neuroendocrine tumors. J Gastrointest Surg. 2014;18:457‐
University of China (No. XJTU1AF‐CRF‐2017‐004). 462. discussion 62‐3.
10. Margonis GA, Samaha M, Kim Y, et al. A Multi‐institutional analysis
of duodenal neuroendocrine tumors: tumor biology rather than extent
CO NFLICT OF I NTERE STS of resection dictates prognosis. J Gastrointest Surg. 2016;20:1098‐1105.
11. Delle Fave G, O'Toole D, Sundin A, et al. ENETS consensus guidelines
The authors declare that they have no conflict of interests. update for gastroduodenal neuroendocrine neoplasms. Neuroendocri-
nology. 2016;103:119‐124.
12. Rossi RE, Rausa E, Cavalcoli F, Conte D, Massironi S. Duodenal
US NEUROENDOCRINE TUMOR STUDY GROUP MEMBERS neuroendocrine neoplasms: a still poorly recognized clinical entity.
Scand J Gastroenterol. 2018;53:835‐842.
Alexandra G. Lopez‐Aguiar MD: Division of Surgical Oncology, 13. Massironi S, Campana D, Partelli S, et al. Heterogeneity of duodenal
Department of Surgery, Winship Cancer Institute, Emory University, neuroendocrine tumors: an italian multi‐center experience. Ann Surg
Oncol. 2018;25:3200‐3206.
Atlanta, GA; Eleftherios Makris MD: Department of Surgery, Stanford
14. Dasari BVM, Al‐Shakhshir S, Pawlik TM, et al. Outcomes of surgical
University, Palo Alto, CA; Zaheer Kanji MD: Department of Surgery, and endoscopic resection of duodenal neuroendocrine tumours
Virginia Mason Medical Center, Seattle, WA; Alexander Fisher MD: (NETs): a systematic review of the literature. J Gastrointest Surg.
Department of Surgery, University of Wisconsin School of Medicine and 2018;22:1652‐1658.
Public Health, Madison, WI; Bradley A. Krasnick MD: Department of 15. Zhang XF, Xue F, Dong DH, et al. New nodal staging for primary
pancreatic neuroendocrine tumors: a multi‐institutional and national
Surgery, Washington University School of Medicine, St. Louis, WI; Paula
data analysis. Ann Surg. 2019. [published online ahead of print July 26,
M. Smith MD: Division of Surgical Oncology, Department of Surgery, 2019].
Vanderbilt University, Nashville, TN; Megan Beems MD: Division of 16. Amin MB. American Joint Committee on Cancer. New York: Springer;
Hepatopancreatobiliary and Advanced Gastrointestinal Surgery, De- 2017.
17. Crippa S, Partelli S, Boninsegna L, Falconi M. Implications of the new
partment of Surgery, University of Michigan, Ann Arbor, MI.
histological classification (WHO 2010) for pancreatic neuroendocrine
neoplasms. Ann Oncol. 23, 2012:1928.
18. Randle RW, Ahmed S, Newman NA, Clark CJ. Clinical outcomes for
D A T A A V A I L A B I LI TY S TA T E ME N T neuroendocrine tumors of the duodenum and ampulla of Vater: a
population‐based study. J Gastrointest Surg. 2014;18:354‐362.
The data that support the findings of this study are available from the
19. Milanetto AC, Pasquali C, Da Broi M, Brambilla T, Capretti G, Zerbi A.
corresponding author upon reasonable request. Ampullary neuroendocrine neoplasms: surgical experience of a rare
and challenging entity. Langenbecks Arch Surg. 2018;403:581‐589.
20. Kachare SD, Liner KR, Vohra NA, Zervos EE, Fitzgerald TL. A
OR CID modified duodenal neuroendocrine tumor staging schema better
defines the risk of lymph node metastasis and disease‐free survival.
Xu‐Feng Zhang http://orcid.org/0000-0002-4483-7326 Am Surg. 2014;80:821‐826.
Diamantis I. Tsilimigras http://orcid.org/0000-0002-3676-9263 21. Burke CA, Beck GJ, Church JM, van Stolk RU. The natural history of
Timothy M. Pawlik http://orcid.org/0000-0002-7994-9870 untreated duodenal and ampullary adenomas in patients with familial
ZHANG ET AL. | 9

adenomatous polyposis followed in an endoscopic surveillance accurately staging small bowel neuroendocrine tumors. JAMA Surg.
program. Gastrointest Endosc. 1999;49:358‐364. 2018;154:134‐140.
22. Iwasaki T, Nara S, Kishi Y, Esaki M, Shimada K, Hiraoka N. Surgical
treatment of neuroendocrine tumors in the second portion of the
duodenum: a single center experience and systematic review of the
literature. Langenbecks Arch Surg. 2017;402:925‐933. How to cite this article: Zhang X‐F, Wu X‐N, Tsilimigras DI,
23. Burke AP, Sobin LH, Federspiel BH, Shekitka KM, Helwig EB. et al. Duodenal neuroendocrine tumors: Impact of tumor size
Carcinoid tumors of the duodenum. A clinicopathologic study of 99
and total number of lymph nodes examined. J Surg Oncol.
cases. Arch Pathol Lab Med. 1990;114:700‐704.
24. Zaidi MY, Lopez‐Aguiar AG, Dillhoff M, et al.. Prognostic role of
2019;1–9. https://doi.org/10.1002/jso.25753
lymph node positivity and number of lymph nodes needed for