Analysis of Postoperative Recurrence in Stage I–III Midgut Neuroendocrine Tumors

Analysis of Postoperative Recurrence in Stage I–III Midgut Neuroendocrine Tumors Abstract Background Surgery represents the only curative treatment for stage I–III midgut neuroendocrine tumors (NETs). At present, there are very limited data on the risk of postoperative recurrence. The optimal modality, duration and frequency of surveillance have not been well established. In this work, we investigated the long-term risk of recurrence, peak timing of recurrence, and potential predictors of relapse in patients with stage I–III midgut NETs. Methods We retrospectively evaluated 129 patients with stage I–III midgut NETs who were seen at the Moffitt Cancer Center between 2000 and 2010 following an R0/R1 resection. Disease-free survival (DFS) was estimated using the Kaplan-Meier method. Demographic, clinical, and pathological features were assessed as potential predictors of recurrence. All statistical tests were two-sided. Results After a median postoperative follow-up of 81 months (range = 1–295 months), recurrence was diagnosed in 40 out of 129 patients (31.0%, 95% confidence interval [CI] = 23.0% to 39.0%). Liver, mesentery, and pelvic lymph nodes were the main sites of relapse. The median DFS was 138 months (95% CI = 117 to 223 months). Resection of 17 or fewer lymph nodes predicted relapse (P = .01) and shorter DFS (P = .04). Among patients who relapsed, the cumulative risks of recurrence at one, five, and 10 years were 15.0% (95% CI = 3.9% to 26.1%), 50.0% (95% CI = 34.5% to 65.5%), and 85.0% (95% CI = 73.9% to 96.0%). No recurrence was observed among patients (n = 6) with stage I tumors, whereas similar rates of relapse were noted in patients with stage II or III NETs (n = 118). Conclusions An annual surveillance interval may allow early detection of recurrence. Given the apparent decline in recurrence after eight years from surgery, a decade-long duration of active surveillance may be proposed. Neuroendocrine tumors (NETs) of the small intestine and proximal colon, also known as midgut carcinoid tumors, represent the most common type of gastrointestinal NET (1). Their incidence has increased substantially in recent years (2). Midgut NETs tend to progress indolently and produce serotonin among other vasoactive substances. In patients with distant metastases, this hormonal output often results in the malignant carcinoid syndrome (3). Analyses of the Surveillance, Epidemiology, and End Results (SEER) database suggest that the majority of patients present initially with tumors that are either local or metastatic to locoregional lymph nodes (4). Symptoms related to intraluminal and mesenteric tumors include abdominal pain, bleeding, or nausea/vomiting. Mesenteric lymph node metastases often produce a dense fibrotic reaction that can tether surrounding bowel and mesenteric vessels, resulting in bowel obstruction or ischemia. Occasionally, tumors of the ileocecum are discovered incidentally on colonoscopy, while mesenteric masses may also be detected incidentally on imaging studies. Surgical resection, typically via right hemicolectomy or partial small bowel resection with lymphadenectomy, represents the standard of care for most patients with early-stage disease (3). While surgery is potentially curative, there are limited data on the risk of postoperative recurrence, and therefore very little evidence upon which to base guidelines for long-term surveillance. One reason for the dearth in literature on recurrence risk is the need for long-term follow-up: tumors can metastasize more than 10 years after primary tumor resection (5,6). In the past, the ability to predict recurrence was also limited by the absence of a commonly accepted staging classification. However, in 2007, the European Neuroendocrine Tumor Society (ENETS) developed a tumor/node/metastasis (TNM)-based staging classification for small bowel NETs of the jejunum and ileum (7). This classification distinguished between tumors confined to the submucosa and smaller than 1 cm (stage I), tumors invading the muscularis propria or subserosa or larger than 1 cm (stage II), tumors invading the visceral peritoneum (serosa) or other organs (stage IIIA), and tumors metastatic to locoregional lymph nodes (stage IIIB). This staging system was subsequently adopted by the American Joint Committee on Cancer (AJCC) for the 7th edition of its staging manual (8). Both ENETS and the AJCC have also adopted a grading classification proposed in 2010 by the World Health Organization (WHO) that distinguishes between low-grade tumors (0–1 mitoses per 10 high-powered fields, or Ki-67 index ≤ 2%), intermediate-grade tumors (mitotic rate 2–20 per 10 high-powered fields, or Ki-67 index 3%–20%), and high-grade tumors (mitotic rate >20 per 10 high-powered fields, or Ki-67 index > 20%) (9). Most small bowel NETs are low grade (10). In order to help counsel patients on risk of postoperative recurrence and to aid in development of surveillance guidelines, we searched a retrospective database for all patients with pathologically confirmed jejunal, ileal, or ileocecal NETs, identifying patients who presented without evidence of distant metastases (stage I–III). The primary goal of our study was to determine stage-based recurrence risk and to ascertain the peak timing of recurrence. In doing so, we hoped to answer the following questions: which patients benefit from postoperative surveillance?; what tests should be performed?; what is the appropriate duration of surveillance?; and what is the appropriate frequency? Methods Patients We searched a retrospective database of patients with midgut NETs seen at our institution (H. Lee Moffitt Cancer Center, Tampa, FL) between January 2000 and December 2010. Within this group, we identified patients who underwent R0/R1 surgical resection for tumors lacking both radiographic and intraoperative evidence of synchronous distant metastases (stage I–III). Patients who were lost to follow-up immediately after the surgery, patients with unresectable mesenteric lymph nodes, and patients with mixed adenoneuroendocrine carcinomas (including goblet cell carcinoids) were excluded. In order to limit referral bias, we also excluded patients who had undergone surgery and subsequently developed metastatic disease prior to referral to our institution. The following information was collected by review of patient medical records: demographics, date of initial diagnosis, date of surgery, presenting symptoms, surgery characteristics, initial tumor stage according to the ENETS/AJCC classification (7), location of primary tumor, presence of a functional hormonal syndrome at diagnosis, elevation of tumor markers such as chromogranin A (CgA) and 5-hydroxindoloacetic acid (5-HIAA), and follow-up data, including date and site of recurrence or date of last contact without recurrence. Pathological information including tumor grade by WHO 2010 criteria (9), tumor size and uni- or multifocality, extent of parietal invasion, clearance of margins and presence of perineural or lymphovascular invasion were obtained by review of surgical pathology reports. The NET diagnosis was considered incidental when patients were being worked up for signs and symptoms unrelated to their midgut NET. Institutional approval was obtained for this study. Statistical Analysis Descriptive statistics were used for patient demographics. The association between recurrence risk and patients’ clinicopathological features was evaluated by χ2 or Fisher’s test, as appropriate. Factors showing a P value of .2 or less at univariate analysis were introduced in a multivariable logistic regression model, in which variables were selected using backward stepwise elimination with statistical significance at a P value of .05 or less. Receiver operating characteristics (ROC) curve analysis was used to set the optimal cutoff point for possible predictors of relapse. DeLong’s test was used to compare the area under the ROC curves (AUC) (11). Disease-free survival (DFS) was calculated from date of surgery (R0/R1) until evidence of macroscopic recurrence by imaging or surgical exploration, or death. Time to progression (TTP) was calculated from date of surgery until evidence of recurrence. Overall survival (OS) was measured from date of initial diagnosis until death from any cause or last known follow-up. All time-to-event functions were estimated by the Kaplan-Meier method and compared by the log-rank test. Multivariable analysis was performed using Cox proportional hazards regression. The assumption of proportionality was verified by log-log plot. Only variables with a P value of .2 or less at univariate analysis were included in the Cox model. Exact 95% confidence intervals were calculated for each proportion of interest. All tests were two-sided, and statistical significance was declared at a P value of .05 or less. Statistical analysis was conducted using MedCal statistical software 12.7 (MedCalc Software bvba, Ostend, Belgium). A potential confounding factor in an institutional survival analysis is the immortal time bias, also known as left truncation bias. For the purpose of this study, immortal time bias refers to the span of time in the observation period of a cohort during which relapse could not have occurred (12). Therefore, here we defined immortal time bias as the interval of time between diagnosis and the original presentation to our institution. To mitigate this bias, we carried out separate analyses for patients who were seen at the Moffitt Cancer Center within 12 months of their diagnosis. Results Demographics and Tumor Characteristics Demographic variables and clinicopathological characteristics of 129 patients included in the study are listed in Table 1. The majority of patients (110/129, 85.3%) were referred to our institution within one year of their diagnosis. The median age at diagnosis was 57 years (range = 33–85 years). The diagnosis was incidental in 39.5% of patients. Presence of the carcinoid syndrome was documented in 17 patients (13.2%) at baseline. The majority of patients (86.0%) had ileal primaries, and multifocal tumors were detected in 30.2% of cases. Right hemicolectomy and partial small bowel resection were performed in 53.5% and 45.0% of the patient population, respectively. The median tumor size was 1.8 cm (range = 0.3–7 cm), and more than three-quarters of patients (76.8%) had lymph node metastases and consequently stage IIIB tumors according to ENETS/AJCC criteria. The median number of lymph nodes harvested was 12 (range = 0–52). Grossly involved mesenteric lymph nodes were present in 27.9% of patients. Proximal and distal bowel margins were always free of disease, whereas radial margins were microscopically negative in 82.2% of the cohort. The majority of tumors (82.9%) were G1, and lymphovascular and perineural invasion were described in 54.3% and 41.1% of cases, respectively. Before surgery, CgA and 5-HIAA were available only in 19 and 26 cases, respectively, and were elevated in 26.3% and 30.8% of patients. Table 1. Patient demographics and tumor characteristics Characteristics  No. of patients (%) (n = 129)  Age at diagnosis, y     Median  57   Range  33–85  Sex     Male  61 (47.3)   Female  68 (52.7)  Race     White  106 (82.2)   Black  13 (10.1)   Hispanic  3 (2.3)   Asian  1 (0.8)   Unknown  6 (4.6)  Incidental diagnosis     Yes  51 (39.5)   No  64 (49.6)   Unknown  14 (10.9)  Carcinoid syndrome     Yes  17 (13.2)   No  112 (86.8)  Tumor location     Jejunum  6 (4.7)   Ileum  111 (86.0)   Right colon and appendix  4 (3.1)   Unknown  8 (6.2)  Tumor multifocality     Yes  39 (30.2)   No  85 (65.9)   Unknown  5 (3.9)  Type of surgery     Right hemicolectomy  69 (53.5)   Partial small bowel resection  58 (45.0)   Unknown  2 (1.5)  Tumor size, cm     Median  1.8   Range  0.3–7  Degree of parietal invasion     Submucosa  6 (4.7)   Muscularis propria  32 (24.8)   Subserosa  19 (14.7)   Serosa/mesenteric fat  48 (37.2)   Unknown  24 (18.6)  Lymph node metastases     N0  19 (14.7)   N1  99 (76.8)   Nx  11 (8.5)  Lymph nodes harvested     ≤17  83 (64.3)   >17  33 (25.6)   Unknown  13 (10.1)  Macroscopic mesenteric lymph nodes     Yes  36 (27.9)   No  89 (69.0)   Unknown  4 (3.1)  Involvement of radial margins     Yes  15 (11.6)   No  106 (82.2)   Unknown  8 (6.2)  Tumor stage     I  6 (4.7)   IIA  12 (9.3)   IIB  2 (1.5)   IIIA  5 (3.9)   IIIB  99 (76.7)   Unknown  5 (3.9)  Tumor grade     G1  107 (82.9)   G2  20 (15.5)   Unknown  2 (1.6)  Lymphovascular invasion     Yes  70 (54.3)   No  13 (10.1)   Unknown  46 (35.6)  Perineural invasion     Yes  53 (41.1)   No  20 (15.5)   Unknown  56 (43.4)  Follow-up, mo     Median  81   Range  1–295  Recurrent disease     Yes  40 (31.0)   No  89 (69.0)  Sites of recurrence     Liver  21 (52.5)   Mesentery  11 (27.5)   Peritoneum  3 (7.5)   Pelvic lymph nodes  6 (15.0)   Others  9 (22.5)  Characteristics  No. of patients (%) (n = 129)  Age at diagnosis, y     Median  57   Range  33–85  Sex     Male  61 (47.3)   Female  68 (52.7)  Race     White  106 (82.2)   Black  13 (10.1)   Hispanic  3 (2.3)   Asian  1 (0.8)   Unknown  6 (4.6)  Incidental diagnosis     Yes  51 (39.5)   No  64 (49.6)   Unknown  14 (10.9)  Carcinoid syndrome     Yes  17 (13.2)   No  112 (86.8)  Tumor location     Jejunum  6 (4.7)   Ileum  111 (86.0)   Right colon and appendix  4 (3.1)   Unknown  8 (6.2)  Tumor multifocality     Yes  39 (30.2)   No  85 (65.9)   Unknown  5 (3.9)  Type of surgery     Right hemicolectomy  69 (53.5)   Partial small bowel resection  58 (45.0)   Unknown  2 (1.5)  Tumor size, cm     Median  1.8   Range  0.3–7  Degree of parietal invasion     Submucosa  6 (4.7)   Muscularis propria  32 (24.8)   Subserosa  19 (14.7)   Serosa/mesenteric fat  48 (37.2)   Unknown  24 (18.6)  Lymph node metastases     N0  19 (14.7)   N1  99 (76.8)   Nx  11 (8.5)  Lymph nodes harvested     ≤17  83 (64.3)   >17  33 (25.6)   Unknown  13 (10.1)  Macroscopic mesenteric lymph nodes     Yes  36 (27.9)   No  89 (69.0)   Unknown  4 (3.1)  Involvement of radial margins     Yes  15 (11.6)   No  106 (82.2)   Unknown  8 (6.2)  Tumor stage     I  6 (4.7)   IIA  12 (9.3)   IIB  2 (1.5)   IIIA  5 (3.9)   IIIB  99 (76.7)   Unknown  5 (3.9)  Tumor grade     G1  107 (82.9)   G2  20 (15.5)   Unknown  2 (1.6)  Lymphovascular invasion     Yes  70 (54.3)   No  13 (10.1)   Unknown  46 (35.6)  Perineural invasion     Yes  53 (41.1)   No  20 (15.5)   Unknown  56 (43.4)  Follow-up, mo     Median  81   Range  1–295  Recurrent disease     Yes  40 (31.0)   No  89 (69.0)  Sites of recurrence     Liver  21 (52.5)   Mesentery  11 (27.5)   Peritoneum  3 (7.5)   Pelvic lymph nodes  6 (15.0)   Others  9 (22.5)  Table 1. Patient demographics and tumor characteristics Characteristics  No. of patients (%) (n = 129)  Age at diagnosis, y     Median  57   Range  33–85  Sex     Male  61 (47.3)   Female  68 (52.7)  Race     White  106 (82.2)   Black  13 (10.1)   Hispanic  3 (2.3)   Asian  1 (0.8)   Unknown  6 (4.6)  Incidental diagnosis     Yes  51 (39.5)   No  64 (49.6)   Unknown  14 (10.9)  Carcinoid syndrome     Yes  17 (13.2)   No  112 (86.8)  Tumor location     Jejunum  6 (4.7)   Ileum  111 (86.0)   Right colon and appendix  4 (3.1)   Unknown  8 (6.2)  Tumor multifocality     Yes  39 (30.2)   No  85 (65.9)   Unknown  5 (3.9)  Type of surgery     Right hemicolectomy  69 (53.5)   Partial small bowel resection  58 (45.0)   Unknown  2 (1.5)  Tumor size, cm     Median  1.8   Range  0.3–7  Degree of parietal invasion     Submucosa  6 (4.7)   Muscularis propria  32 (24.8)   Subserosa  19 (14.7)   Serosa/mesenteric fat  48 (37.2)   Unknown  24 (18.6)  Lymph node metastases     N0  19 (14.7)   N1  99 (76.8)   Nx  11 (8.5)  Lymph nodes harvested     ≤17  83 (64.3)   >17  33 (25.6)   Unknown  13 (10.1)  Macroscopic mesenteric lymph nodes     Yes  36 (27.9)   No  89 (69.0)   Unknown  4 (3.1)  Involvement of radial margins     Yes  15 (11.6)   No  106 (82.2)   Unknown  8 (6.2)  Tumor stage     I  6 (4.7)   IIA  12 (9.3)   IIB  2 (1.5)   IIIA  5 (3.9)   IIIB  99 (76.7)   Unknown  5 (3.9)  Tumor grade     G1  107 (82.9)   G2  20 (15.5)   Unknown  2 (1.6)  Lymphovascular invasion     Yes  70 (54.3)   No  13 (10.1)   Unknown  46 (35.6)  Perineural invasion     Yes  53 (41.1)   No  20 (15.5)   Unknown  56 (43.4)  Follow-up, mo     Median  81   Range  1–295  Recurrent disease     Yes  40 (31.0)   No  89 (69.0)  Sites of recurrence     Liver  21 (52.5)   Mesentery  11 (27.5)   Peritoneum  3 (7.5)   Pelvic lymph nodes  6 (15.0)   Others  9 (22.5)  Characteristics  No. of patients (%) (n = 129)  Age at diagnosis, y     Median  57   Range  33–85  Sex     Male  61 (47.3)   Female  68 (52.7)  Race     White  106 (82.2)   Black  13 (10.1)   Hispanic  3 (2.3)   Asian  1 (0.8)   Unknown  6 (4.6)  Incidental diagnosis     Yes  51 (39.5)   No  64 (49.6)   Unknown  14 (10.9)  Carcinoid syndrome     Yes  17 (13.2)   No  112 (86.8)  Tumor location     Jejunum  6 (4.7)   Ileum  111 (86.0)   Right colon and appendix  4 (3.1)   Unknown  8 (6.2)  Tumor multifocality     Yes  39 (30.2)   No  85 (65.9)   Unknown  5 (3.9)  Type of surgery     Right hemicolectomy  69 (53.5)   Partial small bowel resection  58 (45.0)   Unknown  2 (1.5)  Tumor size, cm     Median  1.8   Range  0.3–7  Degree of parietal invasion     Submucosa  6 (4.7)   Muscularis propria  32 (24.8)   Subserosa  19 (14.7)   Serosa/mesenteric fat  48 (37.2)   Unknown  24 (18.6)  Lymph node metastases     N0  19 (14.7)   N1  99 (76.8)   Nx  11 (8.5)  Lymph nodes harvested     ≤17  83 (64.3)   >17  33 (25.6)   Unknown  13 (10.1)  Macroscopic mesenteric lymph nodes     Yes  36 (27.9)   No  89 (69.0)   Unknown  4 (3.1)  Involvement of radial margins     Yes  15 (11.6)   No  106 (82.2)   Unknown  8 (6.2)  Tumor stage     I  6 (4.7)   IIA  12 (9.3)   IIB  2 (1.5)   IIIA  5 (3.9)   IIIB  99 (76.7)   Unknown  5 (3.9)  Tumor grade     G1  107 (82.9)   G2  20 (15.5)   Unknown  2 (1.6)  Lymphovascular invasion     Yes  70 (54.3)   No  13 (10.1)   Unknown  46 (35.6)  Perineural invasion     Yes  53 (41.1)   No  20 (15.5)   Unknown  56 (43.4)  Follow-up, mo     Median  81   Range  1–295  Recurrent disease     Yes  40 (31.0)   No  89 (69.0)  Sites of recurrence     Liver  21 (52.5)   Mesentery  11 (27.5)   Peritoneum  3 (7.5)   Pelvic lymph nodes  6 (15.0)   Others  9 (22.5)  Recurrence Rates and Characteristics After a median postoperative follow-up of 81 months (range = 1–295 months), recurrence was diagnosed in 40 of 129 patients (31.0%, 95% confidence interval [CI] = 23.0% to 39.0%) (Table 1). Among patients who relapsed, the cumulative rates of recurrence at one, three, five, and 10 years were 15.0% (95% CI =  3.9% to 26.1%), 37.5% (95% CI =  22.5% to 52.5%), 50.0% (95% CI =  34.5% to 65.5%), and 85.0% (95% CI =  73.9% to 96.0%), respectively. As shown in Figure 1, the risk of recurrence substantially declined eight years after surgery among relapsing patients. Nearly all recurrences were detected within 12 years of surgery, with the exception of one patient, whose primary tumor had the largest diameter recorded in this cohort (7 cm). No relapse was observed among patients (n = 6) with stage I tumors. Among patients with stage II–IIIA tumors (n = 19), one-, three-, five-, and 10-year cumulative relapse rates were 10.5% (95% CI =  0.0% to 24.3%), 21.1% (95% CI =  2.8% to 39.4%), 21.1% (95% CI =  2.8% to 39.4%), and 31.6% (95% CI =  10.7% to 52.5%), respectively, whereas recurrence rates of 4.0% (95% CI =  0.2% to 7.8%), 10.1% (95% CI =  4.2% to 16.0%), 15.1% (95% CI =  8.0% to 22.1%), and 26.3% (95% CI =  17.6% to 35.0%) were estimated for patients with stage IIIB (lymph node–positive) midgut NETs (n = 99). The majority of recurrences were metastatic, with liver (52.5%) and pelvic lymph nodes (15.0%) being the main sites of distant relapse (Table 1). Thirteen patients (32.5%) presented with locoregional recurrences (11 in the mesentery, two in the small intestine). All patients recurring in the mesentery had surgical margins free of disease. At the time of relapse, 19 of 40 patients (47.5%) were asymptomatic, whereas abdominal pain, new-onset/worsened diarrhea or flushing, and/or diaphoresis prompted further work-up in 32.5%, 22.5%, 12.5%, and 5.0% of patients, respectively. Disease recurrence was mainly diagnosed by computed tomography (80.0%) or magnetic resonance imaging (20.0%). These investigations were usually limited to the abdomen and pelvis (85.0%). As depicted in Figure 2, 5-HIAA levels at relapse were statistically significantly different when compared with their first value after curative surgery (P < .001). There was no difference between CgA levels after surgery and at relapse (P = .08). Figure 1. View largeDownload slide Frequency distribution of postoperative recurrence in relapsing stage I–III midgut NETs. Recurrences (white circles) are represented as a function of time. Figure 1. View largeDownload slide Frequency distribution of postoperative recurrence in relapsing stage I–III midgut NETs. Recurrences (white circles) are represented as a function of time. Figure 2. View largeDownload slide Tumor markers and recurrence detection. Differences in (A) median chromogranin A serum concentrations and (B) median urinary 5-HIAA after surgery and at relapse were calculated by Wilcoxon matched pairs signed-rank test. Paired row values, median change, and interquartile range are represented. All statistical tests were two-sided. CgA = chromogranin A; 5-HIAA = 5-hydroxyindoleacetic acid. Figure 2. View largeDownload slide Tumor markers and recurrence detection. Differences in (A) median chromogranin A serum concentrations and (B) median urinary 5-HIAA after surgery and at relapse were calculated by Wilcoxon matched pairs signed-rank test. Paired row values, median change, and interquartile range are represented. All statistical tests were two-sided. CgA = chromogranin A; 5-HIAA = 5-hydroxyindoleacetic acid. Predictors of Recurrence As detailed in Table 2, the number of lymph nodes excised at the time of surgical resection was the only clinic-pathologic feature predictive of relapse in univariate analysis. However, the relative homogeneity of the study cohort and the consequent presence of subgroups with small numbers may have limited our ability to detect statistically significant differences, as suggested by wide confidence intervals around recurrence rates. To identify the optimal cutoff number of harvested lymph nodes capable of predicting relapse, ROC curve analysis was performed (Figure 3). In our cohort, a lymph node threshold of 17 or fewer distinguished between relapsing and nonrelapsing patients with a sensitivity of 88.2% (95% CI =  72.5% to 96.7%) and a specificity of 35.4% (95% CI =  25.1% to 46.7%), with an AUC of 0.63 (95% CI =  0.54 to 0.72, P = .01). By univariate analysis, a number of harvested lymph nodes of 17 or fewer statistically significantly predicted recurrence (P = .01). After adjusting for variables that showed a P value of less than .20 in univariate analysis (incidental diagnosis, N stage, tumor size), dissection of more than 17 lymph nodes remained associated with reduced risk of tumor recurrence in multivariable analysis (odds ratio = 0.21, 95% CI =  0.06 to 0.80, P = .02). Table 2. Univariate analysis of potential predictors of recurrence Characteristics  Recurrence rate (95% CI), %  P*  Median DFS (95% CI), mo  P†  Age at diagnosis, y    .98    .72   20–40  25.0 (7.1 to 59.1)    NR (11 to NR)     40–60  31.8 (21.8 to 43.8)    138 (124 to 139)     >60  30.9 (20.3 to 44)    117 (86 to 223)    Sex    .71    .83   Male  32.8 (22.3 to 45.3)    134 (87 to 223)     Female  29.4 (19.9 to 41.1)    NR (96 to NR)    Race    .22    .06   White  31.1 (23.1 to 40.5)    144 (124 to 223)     Black  46.1 (23.2 to 70.8)    87 (36 to 117)     Hispanic, Asian, unknown  11.1 (2.0 to 43.5)    NR (139 to NR)    Incidental diagnosis    .17    .33   Yes  37.5 (26.7 to 49.7)    NR (87 to NR)     No  21.6 (12.5 to 34.6)    124 (87 to 144)     Unknown  35.7 (16.3 to 61.2)    NR (96 to NR)    Carcinoid syndrome    .78    .79   Yes  35.3 (17.3 to 58.7)    134 (86 to 134)     No  30.4 (22.6 to 39.4)    139 (117 to 223)    Tumor location    .58    .33   Ileum  32.4 (24.4 to 41.6)    138 (96 to 223)     Others  22.2 (9.0 to 45.2)    NR (87 to NR)    Tumor multifocality    .30    .09   Yes  30.8 (18.6 to 46.4)    139 (68 to 223)     No  32.9 (23.9 to 43.5)    134 (96 to 144)    Type of surgery    .70    .33   Right hemicolectomy  29.0 (19.6 to 40.6)    138 (96 to 144)     Partial small bowel resection  32.8 (22.1 to 45.6)    139 (87 to 223)    Tumor size, cm    .07    .08   <1  19.2 (8.5 to 37.9)    124 (124 to 125)     1–2  34.8 (22.7 to 49.2)    87 (68 to 138)     2–4  37.2 (24.3 to 52.1)    134 (87 to 144)     >4  60.0 (23.1 to 88.2)    223 (3 to NR)     Unknown  0 (0 to 30)    NR (NR to NR)    Degree of parietal invasion    .49    .26   Submucosa  16.7 (3.0 to 56.3)    NR (134 to NR)     Muscularis propria  31.3 (18.0 to 48.6)    96 (87 to 139)     Subserosa  47.4 (27.3 to 68.3)    144 (36 to NR)     Serosa/mesenteric fat  29.2 (18.2 to 43.2)    124 (87 to 223)    Lymph node metastases    .20    .65   N0  15.8 (5.5 to 37.6)    NR (NR to NR)     N1  32.3 (23.9 to 42.0)    138 (96 to 223)     Nx  45.4 (21.3 to 72.0)    NR (24 to NR)    Lymph nodes harvested    .01    .03   ≤17  36.1 (26.6 to 46.9)    134 (87 to 144)     >17  12.1 (4.8 to 27.3)    NR (87 to NR)    Macroscopic mesenteric lymph nodes    .39    .64   Yes  33.3 (20.2 to 49.7)    139 (87 to 223)     No  31.5 (22.7 to 41.7)    134 (96 to 144)    Tumor stage    .58    .40   I  0 (0 to 39)    NR (NR to NR)     II  35.7 (16.3 to 61.2)    NR (25 to NR)     III  31.7 (23.6 to 41.2)    138 (96 to 223)    Tumor grade    .88    .25   G1  31.8 (23.7 to 41.1)    138 (117 to 223)     G2  30.0 (14.5 to 51.9)    81 (66 to 96)    Involvement of radial margins    .50    .33   Yes  33.3 (15.2 to 58.3)    139 (41 to NR)     No  32.1 (23.9 to 41.5)    134 (96 to 144)    Lymphovascular invasion    .78    .55   Yes  32.9 (23.0 to 44.5)    139 (87 to 144)     No  23.1 (8.2 to 50.3)    87 (87 to 88)     Unknown  30.4 (19.1 to 44.8)    138 (117 to 223)    Perineural invasion    .24    .31   Yes  34.0 (22.7 to 47.4)    144 (78 to 144)     No  15.0 (5.2 to 36.0)    124 (96 to NR)     Unknown  33.9 (22.9 to 47.0)    139 (117 to 223)    Characteristics  Recurrence rate (95% CI), %  P*  Median DFS (95% CI), mo  P†  Age at diagnosis, y    .98    .72   20–40  25.0 (7.1 to 59.1)    NR (11 to NR)     40–60  31.8 (21.8 to 43.8)    138 (124 to 139)     >60  30.9 (20.3 to 44)    117 (86 to 223)    Sex    .71    .83   Male  32.8 (22.3 to 45.3)    134 (87 to 223)     Female  29.4 (19.9 to 41.1)    NR (96 to NR)    Race    .22    .06   White  31.1 (23.1 to 40.5)    144 (124 to 223)     Black  46.1 (23.2 to 70.8)    87 (36 to 117)     Hispanic, Asian, unknown  11.1 (2.0 to 43.5)    NR (139 to NR)    Incidental diagnosis    .17    .33   Yes  37.5 (26.7 to 49.7)    NR (87 to NR)     No  21.6 (12.5 to 34.6)    124 (87 to 144)     Unknown  35.7 (16.3 to 61.2)    NR (96 to NR)    Carcinoid syndrome    .78    .79   Yes  35.3 (17.3 to 58.7)    134 (86 to 134)     No  30.4 (22.6 to 39.4)    139 (117 to 223)    Tumor location    .58    .33   Ileum  32.4 (24.4 to 41.6)    138 (96 to 223)     Others  22.2 (9.0 to 45.2)    NR (87 to NR)    Tumor multifocality    .30    .09   Yes  30.8 (18.6 to 46.4)    139 (68 to 223)     No  32.9 (23.9 to 43.5)    134 (96 to 144)    Type of surgery    .70    .33   Right hemicolectomy  29.0 (19.6 to 40.6)    138 (96 to 144)     Partial small bowel resection  32.8 (22.1 to 45.6)    139 (87 to 223)    Tumor size, cm    .07    .08   <1  19.2 (8.5 to 37.9)    124 (124 to 125)     1–2  34.8 (22.7 to 49.2)    87 (68 to 138)     2–4  37.2 (24.3 to 52.1)    134 (87 to 144)     >4  60.0 (23.1 to 88.2)    223 (3 to NR)     Unknown  0 (0 to 30)    NR (NR to NR)    Degree of parietal invasion    .49    .26   Submucosa  16.7 (3.0 to 56.3)    NR (134 to NR)     Muscularis propria  31.3 (18.0 to 48.6)    96 (87 to 139)     Subserosa  47.4 (27.3 to 68.3)    144 (36 to NR)     Serosa/mesenteric fat  29.2 (18.2 to 43.2)    124 (87 to 223)    Lymph node metastases    .20    .65   N0  15.8 (5.5 to 37.6)    NR (NR to NR)     N1  32.3 (23.9 to 42.0)    138 (96 to 223)     Nx  45.4 (21.3 to 72.0)    NR (24 to NR)    Lymph nodes harvested    .01    .03   ≤17  36.1 (26.6 to 46.9)    134 (87 to 144)     >17  12.1 (4.8 to 27.3)    NR (87 to NR)    Macroscopic mesenteric lymph nodes    .39    .64   Yes  33.3 (20.2 to 49.7)    139 (87 to 223)     No  31.5 (22.7 to 41.7)    134 (96 to 144)    Tumor stage    .58    .40   I  0 (0 to 39)    NR (NR to NR)     II  35.7 (16.3 to 61.2)    NR (25 to NR)     III  31.7 (23.6 to 41.2)    138 (96 to 223)    Tumor grade    .88    .25   G1  31.8 (23.7 to 41.1)    138 (117 to 223)     G2  30.0 (14.5 to 51.9)    81 (66 to 96)    Involvement of radial margins    .50    .33   Yes  33.3 (15.2 to 58.3)    139 (41 to NR)     No  32.1 (23.9 to 41.5)    134 (96 to 144)    Lymphovascular invasion    .78    .55   Yes  32.9 (23.0 to 44.5)    139 (87 to 144)     No  23.1 (8.2 to 50.3)    87 (87 to 88)     Unknown  30.4 (19.1 to 44.8)    138 (117 to 223)    Perineural invasion    .24    .31   Yes  34.0 (22.7 to 47.4)    144 (78 to 144)     No  15.0 (5.2 to 36.0)    124 (96 to NR)     Unknown  33.9 (22.9 to 47.0)    139 (117 to 223)    * Calculated by χ2 or Fisher’s exact test, as appropriate. Both tests were two-sided. CI = confidence interval; DFS = disease free survival; NR = not reached. † Calculated by two-sided log-rank test. Table 2. Univariate analysis of potential predictors of recurrence Characteristics  Recurrence rate (95% CI), %  P*  Median DFS (95% CI), mo  P†  Age at diagnosis, y    .98    .72   20–40  25.0 (7.1 to 59.1)    NR (11 to NR)     40–60  31.8 (21.8 to 43.8)    138 (124 to 139)     >60  30.9 (20.3 to 44)    117 (86 to 223)    Sex    .71    .83   Male  32.8 (22.3 to 45.3)    134 (87 to 223)     Female  29.4 (19.9 to 41.1)    NR (96 to NR)    Race    .22    .06   White  31.1 (23.1 to 40.5)    144 (124 to 223)     Black  46.1 (23.2 to 70.8)    87 (36 to 117)     Hispanic, Asian, unknown  11.1 (2.0 to 43.5)    NR (139 to NR)    Incidental diagnosis    .17    .33   Yes  37.5 (26.7 to 49.7)    NR (87 to NR)     No  21.6 (12.5 to 34.6)    124 (87 to 144)     Unknown  35.7 (16.3 to 61.2)    NR (96 to NR)    Carcinoid syndrome    .78    .79   Yes  35.3 (17.3 to 58.7)    134 (86 to 134)     No  30.4 (22.6 to 39.4)    139 (117 to 223)    Tumor location    .58    .33   Ileum  32.4 (24.4 to 41.6)    138 (96 to 223)     Others  22.2 (9.0 to 45.2)    NR (87 to NR)    Tumor multifocality    .30    .09   Yes  30.8 (18.6 to 46.4)    139 (68 to 223)     No  32.9 (23.9 to 43.5)    134 (96 to 144)    Type of surgery    .70    .33   Right hemicolectomy  29.0 (19.6 to 40.6)    138 (96 to 144)     Partial small bowel resection  32.8 (22.1 to 45.6)    139 (87 to 223)    Tumor size, cm    .07    .08   <1  19.2 (8.5 to 37.9)    124 (124 to 125)     1–2  34.8 (22.7 to 49.2)    87 (68 to 138)     2–4  37.2 (24.3 to 52.1)    134 (87 to 144)     >4  60.0 (23.1 to 88.2)    223 (3 to NR)     Unknown  0 (0 to 30)    NR (NR to NR)    Degree of parietal invasion    .49    .26   Submucosa  16.7 (3.0 to 56.3)    NR (134 to NR)     Muscularis propria  31.3 (18.0 to 48.6)    96 (87 to 139)     Subserosa  47.4 (27.3 to 68.3)    144 (36 to NR)     Serosa/mesenteric fat  29.2 (18.2 to 43.2)    124 (87 to 223)    Lymph node metastases    .20    .65   N0  15.8 (5.5 to 37.6)    NR (NR to NR)     N1  32.3 (23.9 to 42.0)    138 (96 to 223)     Nx  45.4 (21.3 to 72.0)    NR (24 to NR)    Lymph nodes harvested    .01    .03   ≤17  36.1 (26.6 to 46.9)    134 (87 to 144)     >17  12.1 (4.8 to 27.3)    NR (87 to NR)    Macroscopic mesenteric lymph nodes    .39    .64   Yes  33.3 (20.2 to 49.7)    139 (87 to 223)     No  31.5 (22.7 to 41.7)    134 (96 to 144)    Tumor stage    .58    .40   I  0 (0 to 39)    NR (NR to NR)     II  35.7 (16.3 to 61.2)    NR (25 to NR)     III  31.7 (23.6 to 41.2)    138 (96 to 223)    Tumor grade    .88    .25   G1  31.8 (23.7 to 41.1)    138 (117 to 223)     G2  30.0 (14.5 to 51.9)    81 (66 to 96)    Involvement of radial margins    .50    .33   Yes  33.3 (15.2 to 58.3)    139 (41 to NR)     No  32.1 (23.9 to 41.5)    134 (96 to 144)    Lymphovascular invasion    .78    .55   Yes  32.9 (23.0 to 44.5)    139 (87 to 144)     No  23.1 (8.2 to 50.3)    87 (87 to 88)     Unknown  30.4 (19.1 to 44.8)    138 (117 to 223)    Perineural invasion    .24    .31   Yes  34.0 (22.7 to 47.4)    144 (78 to 144)     No  15.0 (5.2 to 36.0)    124 (96 to NR)     Unknown  33.9 (22.9 to 47.0)    139 (117 to 223)    Characteristics  Recurrence rate (95% CI), %  P*  Median DFS (95% CI), mo  P†  Age at diagnosis, y    .98    .72   20–40  25.0 (7.1 to 59.1)    NR (11 to NR)     40–60  31.8 (21.8 to 43.8)    138 (124 to 139)     >60  30.9 (20.3 to 44)    117 (86 to 223)    Sex    .71    .83   Male  32.8 (22.3 to 45.3)    134 (87 to 223)     Female  29.4 (19.9 to 41.1)    NR (96 to NR)    Race    .22    .06   White  31.1 (23.1 to 40.5)    144 (124 to 223)     Black  46.1 (23.2 to 70.8)    87 (36 to 117)     Hispanic, Asian, unknown  11.1 (2.0 to 43.5)    NR (139 to NR)    Incidental diagnosis    .17    .33   Yes  37.5 (26.7 to 49.7)    NR (87 to NR)     No  21.6 (12.5 to 34.6)    124 (87 to 144)     Unknown  35.7 (16.3 to 61.2)    NR (96 to NR)    Carcinoid syndrome    .78    .79   Yes  35.3 (17.3 to 58.7)    134 (86 to 134)     No  30.4 (22.6 to 39.4)    139 (117 to 223)    Tumor location    .58    .33   Ileum  32.4 (24.4 to 41.6)    138 (96 to 223)     Others  22.2 (9.0 to 45.2)    NR (87 to NR)    Tumor multifocality    .30    .09   Yes  30.8 (18.6 to 46.4)    139 (68 to 223)     No  32.9 (23.9 to 43.5)    134 (96 to 144)    Type of surgery    .70    .33   Right hemicolectomy  29.0 (19.6 to 40.6)    138 (96 to 144)     Partial small bowel resection  32.8 (22.1 to 45.6)    139 (87 to 223)    Tumor size, cm    .07    .08   <1  19.2 (8.5 to 37.9)    124 (124 to 125)     1–2  34.8 (22.7 to 49.2)    87 (68 to 138)     2–4  37.2 (24.3 to 52.1)    134 (87 to 144)     >4  60.0 (23.1 to 88.2)    223 (3 to NR)     Unknown  0 (0 to 30)    NR (NR to NR)    Degree of parietal invasion    .49    .26   Submucosa  16.7 (3.0 to 56.3)    NR (134 to NR)     Muscularis propria  31.3 (18.0 to 48.6)    96 (87 to 139)     Subserosa  47.4 (27.3 to 68.3)    144 (36 to NR)     Serosa/mesenteric fat  29.2 (18.2 to 43.2)    124 (87 to 223)    Lymph node metastases    .20    .65   N0  15.8 (5.5 to 37.6)    NR (NR to NR)     N1  32.3 (23.9 to 42.0)    138 (96 to 223)     Nx  45.4 (21.3 to 72.0)    NR (24 to NR)    Lymph nodes harvested    .01    .03   ≤17  36.1 (26.6 to 46.9)    134 (87 to 144)     >17  12.1 (4.8 to 27.3)    NR (87 to NR)    Macroscopic mesenteric lymph nodes    .39    .64   Yes  33.3 (20.2 to 49.7)    139 (87 to 223)     No  31.5 (22.7 to 41.7)    134 (96 to 144)    Tumor stage    .58    .40   I  0 (0 to 39)    NR (NR to NR)     II  35.7 (16.3 to 61.2)    NR (25 to NR)     III  31.7 (23.6 to 41.2)    138 (96 to 223)    Tumor grade    .88    .25   G1  31.8 (23.7 to 41.1)    138 (117 to 223)     G2  30.0 (14.5 to 51.9)    81 (66 to 96)    Involvement of radial margins    .50    .33   Yes  33.3 (15.2 to 58.3)    139 (41 to NR)     No  32.1 (23.9 to 41.5)    134 (96 to 144)    Lymphovascular invasion    .78    .55   Yes  32.9 (23.0 to 44.5)    139 (87 to 144)     No  23.1 (8.2 to 50.3)    87 (87 to 88)     Unknown  30.4 (19.1 to 44.8)    138 (117 to 223)    Perineural invasion    .24    .31   Yes  34.0 (22.7 to 47.4)    144 (78 to 144)     No  15.0 (5.2 to 36.0)    124 (96 to NR)     Unknown  33.9 (22.9 to 47.0)    139 (117 to 223)    * Calculated by χ2 or Fisher’s exact test, as appropriate. Both tests were two-sided. CI = confidence interval; DFS = disease free survival; NR = not reached. † Calculated by two-sided log-rank test. Figure 3. View largeDownload slide Lymph node count and tumor recurrence. Receiver operating characteristic curve showing the relationship between sensitivity and false-positive rate. Figure 3. View largeDownload slide Lymph node count and tumor recurrence. Receiver operating characteristic curve showing the relationship between sensitivity and false-positive rate. We also evaluated a lower lymph node threshold of 12 based on data on minimal adequate evaluation in colorectal cancer (14). On univariate analysis, a number of harvested lymph nodes of less than 12 was associated with increased tumor recurrence in a statistically significant fashion (P = .04). Disease-Free Survival DFS was measured from time of resection until the earliest evidence of relapse, death from any cause, or last clinical contact. The median DFS was 138 months (95% CI =  117 to 223 months) (Figure 4A). The five- and 10-year DFS rates were 81.6% (±3.8%) and 59.5% (±5.9%). There was no difference between TTP and DFS because no death was recorded before tumor recurrence. As illustrated in Figure 4B, patients who underwent resection of more than 17 lymph nodes had a superior DFS as compared with those who were subjected to dissection of 17 or fewer lymph nodes (median DFS = 134 months vs not reached, respectively, P = .04). In multivariable analysis, statistical significance was retained after controlling for race, tumor multifocality, and tumor size (P = .04). All remaining clinical-pathologic factors were not statistically significantly associated with changes in DFS (Table 2). Figure 4. View largeDownload slide Kaplan-Meier estimates of disease-free survival (DFS). A) DFS in the overall population. B) DFS by number of harvested lymph nodes. Figure 4. View largeDownload slide Kaplan-Meier estimates of disease-free survival (DFS). A) DFS in the overall population. B) DFS by number of harvested lymph nodes. Immortal Time Bias To mitigate the immortal time bias, patients who were seen at our institution within 12 months of their diagnosis (n = 109, 84.5%) were analyzed separately (data not shown). The median DFS of patients referred within 12 months of diagnosis was 134 months (95% CI =  87 to 144 months), whereas the median DFS was not reached (95% CI =  223 months to not reached) in patients referred after 12 months (P = .01). Overall Survival Five deaths occurred among the 129 patients in this cohort. Of the four deaths recorded among relapsing patients, two deaths were known to be directly related to progressive metastatic disease, one was possibly related to disease progression, and one was not tumor related (data not shown). Of the two deaths clearly attributable to progressive metastatic disease, one occurred nine years after initial surgery and seven years after metastatic relapse, while the other was recorded 12 years after initial surgery and seven years after relapse. All the remaining 36 patients with relapsed disease were alive at the time of data cutoff. Among relapsing patients, the 10-year and 15-year OS rates were 92.9% (±5.0%) and 74.0% (±13.3%). Discussion Data on risk of recurrence after resection of stage I–III midgut NETs are scarce. Retrospective series (5,6) have been limited by small size, relatively short median follow-up, and inclusion of patients who recurred prior to referral to the institution conducting the study. These factors can lead to either overestimation or underestimation of recurrence risk. Yet accurate information on probability of recurrence is critically important for several reasons: postoperative surveillance guidelines need to draw upon these data to develop recommendations on timing and duration of follow-up evaluations, adjuvant studies require estimates of baseline risk of recurrence, and patients seek accurate information on their personal risk of recurrence after surgery. Past multi-institutional series have demonstrated that the large majority of patients presenting with midgut NETs have evidence of lymph node involvement on surgical resection. Metastatic recurrence rates in these patients have been reported to be roughly 50%, with an increase in risk associated with more distant mesenteric lymph node metastases and multifocal primary tumors (which are often associated with gross lymphadenopathy) (5,6,13). Our database of 129 patients who underwent resection of stage I–III midgut NETs represents, to our knowledge, the largest single or multi-institutional series to examine this specific population. Results of our analysis show that 76.8% of patients presented with lymph node–positive disease. This is similar to a rate of 79% observed in a multi-institutional French study of 100 patients (5). The median follow-up of 81 months allows us to evaluate long-term risk of recurrence with reasonable accuracy, although estimates of cumulative recurrence risk after five years are limited by large confidence intervals due to small numbers. Among all patients, the recurrence rate in our series was 31.0% (95% CI =  23.0% to 39.0%). Among lymph node–positive patients (n = 99), the long-term recurrence rate was 32.3%, with a 95% confidence interval of 23.1% to 41.5%. This is also in accordance with the prior 100-patient series (5) which revealed a 42% (95% CI =  32% to 52%) cumulative rate of recurrence (median follow-up = 57.5 months). An interesting finding in our study is that patients with stage I tumors (n = 6) had no evidence of recurrence, suggesting that such patients may be able to forgo surveillance or undergo a relatively low-intensity postoperative surveillance schedule. However, larger series of stage I patients are needed to draw firmer conclusions regarding recurrence risk. The rate of recurrence in our database was relatively steady over an eight-year period, with a subsequent dropoff. It is uncertain whether this finding reflects a true decline in recurrence after eight years or a statistical artifact (only 25% of patients had surveillance beyond eight years). We conclude from these data that surveillance for recurrence does not need to be particularly frequent during the initial years after surgery. Indeed, an annual surveillance interval will likely allow early detection of most patients with metastatic disease. It is hard to recommend an outer limit for duration of surveillance given the fact that recurrences can occur after 10 years. However, given the apparent decline in recurrence after eight years, a decade-long duration of active surveillance will likely enable detection of most recurrences. In our series, the small number of patients with baseline elevated 5-HIAA or CgA hinders reliable conclusions regarding the role of tumor markers in the postoperative surveillance of patients with midgut carcinoid. While CgA did not appear to be a robust tumor marker for detection of metastatic recurrence, urine 5-HIAA levels did increase at the time of diagnosis of recurrence. However, it is difficult to ascertain whether measurement of urine 5-HIAA adds diagnostic value beyond cross-sectional imaging. Among the clinical-pathologic features evaluated in this study, only the number of harvested lymph nodes statistically significantly predicted recurrence in both univariate (P = .01) and multivariable analysis (P = .02). While ROC analysis identified 17 lymph nodes as most predictive of recurrence, even removal of 12 lymph nodes was associated with a statistically significant reduction in recurrence risk. Consistent with the evidence that removal of mesenteric lymph node metastases positively impacts survival in patients with midgut NETs (15), our data suggest that an appropriate lymphadenectomy, rather than simple segmental bowel resection, is preferable in midgut NETs. With respect to other pathological prognostic factors, the statistical power of our study is limited by relatively small numbers of patients with grade 2 and lymph node–negative tumors. In our analysis, we were only able to confirm two deaths that were clearly attributable to disease progression. Given the median age of diagnosis (57 years) and very long survival of patients with early detection of metastatic disease on surveillance, it is unclear to what extent a diagnosis of early-stage midgut NET impacts overall life expectancy. It is likely that competing causes of death exceed NET-related deaths in this population. This has important implications regarding adjuvant therapy trials as it would be difficult to demonstrate that delay in recurrence impacts survival. The main limitation of our study is incomplete follow-up, either due to loss of patients from the practice (patients undergoing surveillance at other institutions or choosing to forgo surveillance), or due to the fact that some patients were diagnosed within the past decade and therefore have not completed a 10-year course of surveillance. Therefore, it is possible that our recurrence rate of 31.0% represents an underestimation of the true long-term recurrence rate. An ideal estimation of recurrence risk would require a prospective database with a fixed surveillance schedule and a very long duration of follow-up. Note The authors declare no conflicts of interest. References 1 Frilling A, Akerström G, Falconi Met al.  , Neuroendocrine tumor disease: An evolving landscape. Endocr Relat Cancer . 2012; 19: R163– R185. Google Scholar CrossRef Search ADS PubMed  2 Yao JC, Hassan M, Phan Aet al.  , One hundred years after “carcinoid”: Epidemiology of and prognostic factors for neuroendocrine tumors in 35,825 cases in the United States. J Clin Oncol . 2008; 26: 3063– 3072. Google Scholar CrossRef Search ADS PubMed  3 Cives M, Strosberg J. An update on gastroenteropancreatic neuroendocrine tumors. Oncology (Williston Park) . 2014; 28: 749– 756. Google Scholar PubMed  4 Mocellin S, Nitti D. Gastrointestinal carcinoid: Epidemiological and survival evidence from a large population-based study (n = 25 531). Ann Oncol.  2013; 24: 3040– 3044. Google Scholar CrossRef Search ADS PubMed  5 Le Roux C, Lombard-Bohas C, Delmas Cet al.  , Relapse factors for ileal neuroendocrine tumours after curative surgery: A retrospective French multicentre study. Dig Liver Dis . 2011; 43: 828– 833. Google Scholar CrossRef Search ADS PubMed  6 Dieckhoff P, Runkel H, Daniel Het al.  , Well-differentiated neuroendocrine neoplasia: relapse-free survival and predictors of recurrence after curative intended resections. Digestion . 2014; 90: 89– 97. Google Scholar CrossRef Search ADS PubMed  7 Rindi G, Klöppel G, Couvelard Aet al.  , TNM staging of midgut and hindgut (neuro) endocrine tumors: A consensus proposal including a grading system. Virchows Arch.  2007; 451: 757– 762. Google Scholar CrossRef Search ADS PubMed  8 Edge S, Byrd DR, Compton CCet al.  , , eds. AJCC Cancer Staging Manual . 7th ed. Chicago: Springer; 2010. 9 Rindi G, Arnold R. Nomenclature and classification of neuroendocrine neoplasms of the digestive system. In: Bosman F, Carneiro F, Hruban R, Theise N, eds. World Health Organization Classification of Tumours of the digestive system . Lyon: IARC Press; 2010: 13– 14. 10 Strosberg JR, Weber JM, Feldman Met al.  , Prognostic validity of the American Joint Committee on Cancer staging classification for midgut neuroendocrine tumors. J Clin Oncol . 2013; 31: 420– 425. Google Scholar CrossRef Search ADS PubMed  11 DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: A nonparametric approach. Biometrics . 1988; 44 3: 837– 845. Google Scholar CrossRef Search ADS PubMed  12 Shariff SZ, Cuerden MS, Jain AKet al.  , The secret of immortal time bias in epidemiologic studies. J Am Soc Nephrol.  2008; 19: 841– 843. Google Scholar CrossRef Search ADS PubMed  13 Welin S, Stridsberg M, Cunningham Jet al.  , Elevated plasma chromogranin A is the first indication of recurrence in radically operated midgut carcinoid tumors. Neuroendocrinology . 2009; 89: 302– 307. Google Scholar CrossRef Search ADS PubMed  14 Nelson H, Petrelli N, Carlin Aet al.  , Guidelines 2000 for colon and rectal cancer surgery. J Natl Cancer Inst.  2001; 93: 583– 596. Google Scholar CrossRef Search ADS PubMed  15 Hellman P, Lundström T, Ohrvall Uet al.  , Effect of surgery on the outcome of midgut carcinoid disease with lymph node and liver metastases. World J Surg . 2002; 26: 991– 997. Google Scholar CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png JNCI: Journal of the National Cancer Institute Oxford University Press

Analysis of Postoperative Recurrence in Stage I–III Midgut Neuroendocrine Tumors

Loading next page...
 
/lp/ou_press/analysis-of-postoperative-recurrence-in-stage-i-iii-midgut-4I0L6pavZP
Publisher
Oxford University Press
Copyright
© The Author 2017. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0027-8874
eISSN
1460-2105
D.O.I.
10.1093/jnci/djx174
Publisher site
See Article on Publisher Site

Abstract

Abstract Background Surgery represents the only curative treatment for stage I–III midgut neuroendocrine tumors (NETs). At present, there are very limited data on the risk of postoperative recurrence. The optimal modality, duration and frequency of surveillance have not been well established. In this work, we investigated the long-term risk of recurrence, peak timing of recurrence, and potential predictors of relapse in patients with stage I–III midgut NETs. Methods We retrospectively evaluated 129 patients with stage I–III midgut NETs who were seen at the Moffitt Cancer Center between 2000 and 2010 following an R0/R1 resection. Disease-free survival (DFS) was estimated using the Kaplan-Meier method. Demographic, clinical, and pathological features were assessed as potential predictors of recurrence. All statistical tests were two-sided. Results After a median postoperative follow-up of 81 months (range = 1–295 months), recurrence was diagnosed in 40 out of 129 patients (31.0%, 95% confidence interval [CI] = 23.0% to 39.0%). Liver, mesentery, and pelvic lymph nodes were the main sites of relapse. The median DFS was 138 months (95% CI = 117 to 223 months). Resection of 17 or fewer lymph nodes predicted relapse (P = .01) and shorter DFS (P = .04). Among patients who relapsed, the cumulative risks of recurrence at one, five, and 10 years were 15.0% (95% CI = 3.9% to 26.1%), 50.0% (95% CI = 34.5% to 65.5%), and 85.0% (95% CI = 73.9% to 96.0%). No recurrence was observed among patients (n = 6) with stage I tumors, whereas similar rates of relapse were noted in patients with stage II or III NETs (n = 118). Conclusions An annual surveillance interval may allow early detection of recurrence. Given the apparent decline in recurrence after eight years from surgery, a decade-long duration of active surveillance may be proposed. Neuroendocrine tumors (NETs) of the small intestine and proximal colon, also known as midgut carcinoid tumors, represent the most common type of gastrointestinal NET (1). Their incidence has increased substantially in recent years (2). Midgut NETs tend to progress indolently and produce serotonin among other vasoactive substances. In patients with distant metastases, this hormonal output often results in the malignant carcinoid syndrome (3). Analyses of the Surveillance, Epidemiology, and End Results (SEER) database suggest that the majority of patients present initially with tumors that are either local or metastatic to locoregional lymph nodes (4). Symptoms related to intraluminal and mesenteric tumors include abdominal pain, bleeding, or nausea/vomiting. Mesenteric lymph node metastases often produce a dense fibrotic reaction that can tether surrounding bowel and mesenteric vessels, resulting in bowel obstruction or ischemia. Occasionally, tumors of the ileocecum are discovered incidentally on colonoscopy, while mesenteric masses may also be detected incidentally on imaging studies. Surgical resection, typically via right hemicolectomy or partial small bowel resection with lymphadenectomy, represents the standard of care for most patients with early-stage disease (3). While surgery is potentially curative, there are limited data on the risk of postoperative recurrence, and therefore very little evidence upon which to base guidelines for long-term surveillance. One reason for the dearth in literature on recurrence risk is the need for long-term follow-up: tumors can metastasize more than 10 years after primary tumor resection (5,6). In the past, the ability to predict recurrence was also limited by the absence of a commonly accepted staging classification. However, in 2007, the European Neuroendocrine Tumor Society (ENETS) developed a tumor/node/metastasis (TNM)-based staging classification for small bowel NETs of the jejunum and ileum (7). This classification distinguished between tumors confined to the submucosa and smaller than 1 cm (stage I), tumors invading the muscularis propria or subserosa or larger than 1 cm (stage II), tumors invading the visceral peritoneum (serosa) or other organs (stage IIIA), and tumors metastatic to locoregional lymph nodes (stage IIIB). This staging system was subsequently adopted by the American Joint Committee on Cancer (AJCC) for the 7th edition of its staging manual (8). Both ENETS and the AJCC have also adopted a grading classification proposed in 2010 by the World Health Organization (WHO) that distinguishes between low-grade tumors (0–1 mitoses per 10 high-powered fields, or Ki-67 index ≤ 2%), intermediate-grade tumors (mitotic rate 2–20 per 10 high-powered fields, or Ki-67 index 3%–20%), and high-grade tumors (mitotic rate >20 per 10 high-powered fields, or Ki-67 index > 20%) (9). Most small bowel NETs are low grade (10). In order to help counsel patients on risk of postoperative recurrence and to aid in development of surveillance guidelines, we searched a retrospective database for all patients with pathologically confirmed jejunal, ileal, or ileocecal NETs, identifying patients who presented without evidence of distant metastases (stage I–III). The primary goal of our study was to determine stage-based recurrence risk and to ascertain the peak timing of recurrence. In doing so, we hoped to answer the following questions: which patients benefit from postoperative surveillance?; what tests should be performed?; what is the appropriate duration of surveillance?; and what is the appropriate frequency? Methods Patients We searched a retrospective database of patients with midgut NETs seen at our institution (H. Lee Moffitt Cancer Center, Tampa, FL) between January 2000 and December 2010. Within this group, we identified patients who underwent R0/R1 surgical resection for tumors lacking both radiographic and intraoperative evidence of synchronous distant metastases (stage I–III). Patients who were lost to follow-up immediately after the surgery, patients with unresectable mesenteric lymph nodes, and patients with mixed adenoneuroendocrine carcinomas (including goblet cell carcinoids) were excluded. In order to limit referral bias, we also excluded patients who had undergone surgery and subsequently developed metastatic disease prior to referral to our institution. The following information was collected by review of patient medical records: demographics, date of initial diagnosis, date of surgery, presenting symptoms, surgery characteristics, initial tumor stage according to the ENETS/AJCC classification (7), location of primary tumor, presence of a functional hormonal syndrome at diagnosis, elevation of tumor markers such as chromogranin A (CgA) and 5-hydroxindoloacetic acid (5-HIAA), and follow-up data, including date and site of recurrence or date of last contact without recurrence. Pathological information including tumor grade by WHO 2010 criteria (9), tumor size and uni- or multifocality, extent of parietal invasion, clearance of margins and presence of perineural or lymphovascular invasion were obtained by review of surgical pathology reports. The NET diagnosis was considered incidental when patients were being worked up for signs and symptoms unrelated to their midgut NET. Institutional approval was obtained for this study. Statistical Analysis Descriptive statistics were used for patient demographics. The association between recurrence risk and patients’ clinicopathological features was evaluated by χ2 or Fisher’s test, as appropriate. Factors showing a P value of .2 or less at univariate analysis were introduced in a multivariable logistic regression model, in which variables were selected using backward stepwise elimination with statistical significance at a P value of .05 or less. Receiver operating characteristics (ROC) curve analysis was used to set the optimal cutoff point for possible predictors of relapse. DeLong’s test was used to compare the area under the ROC curves (AUC) (11). Disease-free survival (DFS) was calculated from date of surgery (R0/R1) until evidence of macroscopic recurrence by imaging or surgical exploration, or death. Time to progression (TTP) was calculated from date of surgery until evidence of recurrence. Overall survival (OS) was measured from date of initial diagnosis until death from any cause or last known follow-up. All time-to-event functions were estimated by the Kaplan-Meier method and compared by the log-rank test. Multivariable analysis was performed using Cox proportional hazards regression. The assumption of proportionality was verified by log-log plot. Only variables with a P value of .2 or less at univariate analysis were included in the Cox model. Exact 95% confidence intervals were calculated for each proportion of interest. All tests were two-sided, and statistical significance was declared at a P value of .05 or less. Statistical analysis was conducted using MedCal statistical software 12.7 (MedCalc Software bvba, Ostend, Belgium). A potential confounding factor in an institutional survival analysis is the immortal time bias, also known as left truncation bias. For the purpose of this study, immortal time bias refers to the span of time in the observation period of a cohort during which relapse could not have occurred (12). Therefore, here we defined immortal time bias as the interval of time between diagnosis and the original presentation to our institution. To mitigate this bias, we carried out separate analyses for patients who were seen at the Moffitt Cancer Center within 12 months of their diagnosis. Results Demographics and Tumor Characteristics Demographic variables and clinicopathological characteristics of 129 patients included in the study are listed in Table 1. The majority of patients (110/129, 85.3%) were referred to our institution within one year of their diagnosis. The median age at diagnosis was 57 years (range = 33–85 years). The diagnosis was incidental in 39.5% of patients. Presence of the carcinoid syndrome was documented in 17 patients (13.2%) at baseline. The majority of patients (86.0%) had ileal primaries, and multifocal tumors were detected in 30.2% of cases. Right hemicolectomy and partial small bowel resection were performed in 53.5% and 45.0% of the patient population, respectively. The median tumor size was 1.8 cm (range = 0.3–7 cm), and more than three-quarters of patients (76.8%) had lymph node metastases and consequently stage IIIB tumors according to ENETS/AJCC criteria. The median number of lymph nodes harvested was 12 (range = 0–52). Grossly involved mesenteric lymph nodes were present in 27.9% of patients. Proximal and distal bowel margins were always free of disease, whereas radial margins were microscopically negative in 82.2% of the cohort. The majority of tumors (82.9%) were G1, and lymphovascular and perineural invasion were described in 54.3% and 41.1% of cases, respectively. Before surgery, CgA and 5-HIAA were available only in 19 and 26 cases, respectively, and were elevated in 26.3% and 30.8% of patients. Table 1. Patient demographics and tumor characteristics Characteristics  No. of patients (%) (n = 129)  Age at diagnosis, y     Median  57   Range  33–85  Sex     Male  61 (47.3)   Female  68 (52.7)  Race     White  106 (82.2)   Black  13 (10.1)   Hispanic  3 (2.3)   Asian  1 (0.8)   Unknown  6 (4.6)  Incidental diagnosis     Yes  51 (39.5)   No  64 (49.6)   Unknown  14 (10.9)  Carcinoid syndrome     Yes  17 (13.2)   No  112 (86.8)  Tumor location     Jejunum  6 (4.7)   Ileum  111 (86.0)   Right colon and appendix  4 (3.1)   Unknown  8 (6.2)  Tumor multifocality     Yes  39 (30.2)   No  85 (65.9)   Unknown  5 (3.9)  Type of surgery     Right hemicolectomy  69 (53.5)   Partial small bowel resection  58 (45.0)   Unknown  2 (1.5)  Tumor size, cm     Median  1.8   Range  0.3–7  Degree of parietal invasion     Submucosa  6 (4.7)   Muscularis propria  32 (24.8)   Subserosa  19 (14.7)   Serosa/mesenteric fat  48 (37.2)   Unknown  24 (18.6)  Lymph node metastases     N0  19 (14.7)   N1  99 (76.8)   Nx  11 (8.5)  Lymph nodes harvested     ≤17  83 (64.3)   >17  33 (25.6)   Unknown  13 (10.1)  Macroscopic mesenteric lymph nodes     Yes  36 (27.9)   No  89 (69.0)   Unknown  4 (3.1)  Involvement of radial margins     Yes  15 (11.6)   No  106 (82.2)   Unknown  8 (6.2)  Tumor stage     I  6 (4.7)   IIA  12 (9.3)   IIB  2 (1.5)   IIIA  5 (3.9)   IIIB  99 (76.7)   Unknown  5 (3.9)  Tumor grade     G1  107 (82.9)   G2  20 (15.5)   Unknown  2 (1.6)  Lymphovascular invasion     Yes  70 (54.3)   No  13 (10.1)   Unknown  46 (35.6)  Perineural invasion     Yes  53 (41.1)   No  20 (15.5)   Unknown  56 (43.4)  Follow-up, mo     Median  81   Range  1–295  Recurrent disease     Yes  40 (31.0)   No  89 (69.0)  Sites of recurrence     Liver  21 (52.5)   Mesentery  11 (27.5)   Peritoneum  3 (7.5)   Pelvic lymph nodes  6 (15.0)   Others  9 (22.5)  Characteristics  No. of patients (%) (n = 129)  Age at diagnosis, y     Median  57   Range  33–85  Sex     Male  61 (47.3)   Female  68 (52.7)  Race     White  106 (82.2)   Black  13 (10.1)   Hispanic  3 (2.3)   Asian  1 (0.8)   Unknown  6 (4.6)  Incidental diagnosis     Yes  51 (39.5)   No  64 (49.6)   Unknown  14 (10.9)  Carcinoid syndrome     Yes  17 (13.2)   No  112 (86.8)  Tumor location     Jejunum  6 (4.7)   Ileum  111 (86.0)   Right colon and appendix  4 (3.1)   Unknown  8 (6.2)  Tumor multifocality     Yes  39 (30.2)   No  85 (65.9)   Unknown  5 (3.9)  Type of surgery     Right hemicolectomy  69 (53.5)   Partial small bowel resection  58 (45.0)   Unknown  2 (1.5)  Tumor size, cm     Median  1.8   Range  0.3–7  Degree of parietal invasion     Submucosa  6 (4.7)   Muscularis propria  32 (24.8)   Subserosa  19 (14.7)   Serosa/mesenteric fat  48 (37.2)   Unknown  24 (18.6)  Lymph node metastases     N0  19 (14.7)   N1  99 (76.8)   Nx  11 (8.5)  Lymph nodes harvested     ≤17  83 (64.3)   >17  33 (25.6)   Unknown  13 (10.1)  Macroscopic mesenteric lymph nodes     Yes  36 (27.9)   No  89 (69.0)   Unknown  4 (3.1)  Involvement of radial margins     Yes  15 (11.6)   No  106 (82.2)   Unknown  8 (6.2)  Tumor stage     I  6 (4.7)   IIA  12 (9.3)   IIB  2 (1.5)   IIIA  5 (3.9)   IIIB  99 (76.7)   Unknown  5 (3.9)  Tumor grade     G1  107 (82.9)   G2  20 (15.5)   Unknown  2 (1.6)  Lymphovascular invasion     Yes  70 (54.3)   No  13 (10.1)   Unknown  46 (35.6)  Perineural invasion     Yes  53 (41.1)   No  20 (15.5)   Unknown  56 (43.4)  Follow-up, mo     Median  81   Range  1–295  Recurrent disease     Yes  40 (31.0)   No  89 (69.0)  Sites of recurrence     Liver  21 (52.5)   Mesentery  11 (27.5)   Peritoneum  3 (7.5)   Pelvic lymph nodes  6 (15.0)   Others  9 (22.5)  Table 1. Patient demographics and tumor characteristics Characteristics  No. of patients (%) (n = 129)  Age at diagnosis, y     Median  57   Range  33–85  Sex     Male  61 (47.3)   Female  68 (52.7)  Race     White  106 (82.2)   Black  13 (10.1)   Hispanic  3 (2.3)   Asian  1 (0.8)   Unknown  6 (4.6)  Incidental diagnosis     Yes  51 (39.5)   No  64 (49.6)   Unknown  14 (10.9)  Carcinoid syndrome     Yes  17 (13.2)   No  112 (86.8)  Tumor location     Jejunum  6 (4.7)   Ileum  111 (86.0)   Right colon and appendix  4 (3.1)   Unknown  8 (6.2)  Tumor multifocality     Yes  39 (30.2)   No  85 (65.9)   Unknown  5 (3.9)  Type of surgery     Right hemicolectomy  69 (53.5)   Partial small bowel resection  58 (45.0)   Unknown  2 (1.5)  Tumor size, cm     Median  1.8   Range  0.3–7  Degree of parietal invasion     Submucosa  6 (4.7)   Muscularis propria  32 (24.8)   Subserosa  19 (14.7)   Serosa/mesenteric fat  48 (37.2)   Unknown  24 (18.6)  Lymph node metastases     N0  19 (14.7)   N1  99 (76.8)   Nx  11 (8.5)  Lymph nodes harvested     ≤17  83 (64.3)   >17  33 (25.6)   Unknown  13 (10.1)  Macroscopic mesenteric lymph nodes     Yes  36 (27.9)   No  89 (69.0)   Unknown  4 (3.1)  Involvement of radial margins     Yes  15 (11.6)   No  106 (82.2)   Unknown  8 (6.2)  Tumor stage     I  6 (4.7)   IIA  12 (9.3)   IIB  2 (1.5)   IIIA  5 (3.9)   IIIB  99 (76.7)   Unknown  5 (3.9)  Tumor grade     G1  107 (82.9)   G2  20 (15.5)   Unknown  2 (1.6)  Lymphovascular invasion     Yes  70 (54.3)   No  13 (10.1)   Unknown  46 (35.6)  Perineural invasion     Yes  53 (41.1)   No  20 (15.5)   Unknown  56 (43.4)  Follow-up, mo     Median  81   Range  1–295  Recurrent disease     Yes  40 (31.0)   No  89 (69.0)  Sites of recurrence     Liver  21 (52.5)   Mesentery  11 (27.5)   Peritoneum  3 (7.5)   Pelvic lymph nodes  6 (15.0)   Others  9 (22.5)  Characteristics  No. of patients (%) (n = 129)  Age at diagnosis, y     Median  57   Range  33–85  Sex     Male  61 (47.3)   Female  68 (52.7)  Race     White  106 (82.2)   Black  13 (10.1)   Hispanic  3 (2.3)   Asian  1 (0.8)   Unknown  6 (4.6)  Incidental diagnosis     Yes  51 (39.5)   No  64 (49.6)   Unknown  14 (10.9)  Carcinoid syndrome     Yes  17 (13.2)   No  112 (86.8)  Tumor location     Jejunum  6 (4.7)   Ileum  111 (86.0)   Right colon and appendix  4 (3.1)   Unknown  8 (6.2)  Tumor multifocality     Yes  39 (30.2)   No  85 (65.9)   Unknown  5 (3.9)  Type of surgery     Right hemicolectomy  69 (53.5)   Partial small bowel resection  58 (45.0)   Unknown  2 (1.5)  Tumor size, cm     Median  1.8   Range  0.3–7  Degree of parietal invasion     Submucosa  6 (4.7)   Muscularis propria  32 (24.8)   Subserosa  19 (14.7)   Serosa/mesenteric fat  48 (37.2)   Unknown  24 (18.6)  Lymph node metastases     N0  19 (14.7)   N1  99 (76.8)   Nx  11 (8.5)  Lymph nodes harvested     ≤17  83 (64.3)   >17  33 (25.6)   Unknown  13 (10.1)  Macroscopic mesenteric lymph nodes     Yes  36 (27.9)   No  89 (69.0)   Unknown  4 (3.1)  Involvement of radial margins     Yes  15 (11.6)   No  106 (82.2)   Unknown  8 (6.2)  Tumor stage     I  6 (4.7)   IIA  12 (9.3)   IIB  2 (1.5)   IIIA  5 (3.9)   IIIB  99 (76.7)   Unknown  5 (3.9)  Tumor grade     G1  107 (82.9)   G2  20 (15.5)   Unknown  2 (1.6)  Lymphovascular invasion     Yes  70 (54.3)   No  13 (10.1)   Unknown  46 (35.6)  Perineural invasion     Yes  53 (41.1)   No  20 (15.5)   Unknown  56 (43.4)  Follow-up, mo     Median  81   Range  1–295  Recurrent disease     Yes  40 (31.0)   No  89 (69.0)  Sites of recurrence     Liver  21 (52.5)   Mesentery  11 (27.5)   Peritoneum  3 (7.5)   Pelvic lymph nodes  6 (15.0)   Others  9 (22.5)  Recurrence Rates and Characteristics After a median postoperative follow-up of 81 months (range = 1–295 months), recurrence was diagnosed in 40 of 129 patients (31.0%, 95% confidence interval [CI] = 23.0% to 39.0%) (Table 1). Among patients who relapsed, the cumulative rates of recurrence at one, three, five, and 10 years were 15.0% (95% CI =  3.9% to 26.1%), 37.5% (95% CI =  22.5% to 52.5%), 50.0% (95% CI =  34.5% to 65.5%), and 85.0% (95% CI =  73.9% to 96.0%), respectively. As shown in Figure 1, the risk of recurrence substantially declined eight years after surgery among relapsing patients. Nearly all recurrences were detected within 12 years of surgery, with the exception of one patient, whose primary tumor had the largest diameter recorded in this cohort (7 cm). No relapse was observed among patients (n = 6) with stage I tumors. Among patients with stage II–IIIA tumors (n = 19), one-, three-, five-, and 10-year cumulative relapse rates were 10.5% (95% CI =  0.0% to 24.3%), 21.1% (95% CI =  2.8% to 39.4%), 21.1% (95% CI =  2.8% to 39.4%), and 31.6% (95% CI =  10.7% to 52.5%), respectively, whereas recurrence rates of 4.0% (95% CI =  0.2% to 7.8%), 10.1% (95% CI =  4.2% to 16.0%), 15.1% (95% CI =  8.0% to 22.1%), and 26.3% (95% CI =  17.6% to 35.0%) were estimated for patients with stage IIIB (lymph node–positive) midgut NETs (n = 99). The majority of recurrences were metastatic, with liver (52.5%) and pelvic lymph nodes (15.0%) being the main sites of distant relapse (Table 1). Thirteen patients (32.5%) presented with locoregional recurrences (11 in the mesentery, two in the small intestine). All patients recurring in the mesentery had surgical margins free of disease. At the time of relapse, 19 of 40 patients (47.5%) were asymptomatic, whereas abdominal pain, new-onset/worsened diarrhea or flushing, and/or diaphoresis prompted further work-up in 32.5%, 22.5%, 12.5%, and 5.0% of patients, respectively. Disease recurrence was mainly diagnosed by computed tomography (80.0%) or magnetic resonance imaging (20.0%). These investigations were usually limited to the abdomen and pelvis (85.0%). As depicted in Figure 2, 5-HIAA levels at relapse were statistically significantly different when compared with their first value after curative surgery (P < .001). There was no difference between CgA levels after surgery and at relapse (P = .08). Figure 1. View largeDownload slide Frequency distribution of postoperative recurrence in relapsing stage I–III midgut NETs. Recurrences (white circles) are represented as a function of time. Figure 1. View largeDownload slide Frequency distribution of postoperative recurrence in relapsing stage I–III midgut NETs. Recurrences (white circles) are represented as a function of time. Figure 2. View largeDownload slide Tumor markers and recurrence detection. Differences in (A) median chromogranin A serum concentrations and (B) median urinary 5-HIAA after surgery and at relapse were calculated by Wilcoxon matched pairs signed-rank test. Paired row values, median change, and interquartile range are represented. All statistical tests were two-sided. CgA = chromogranin A; 5-HIAA = 5-hydroxyindoleacetic acid. Figure 2. View largeDownload slide Tumor markers and recurrence detection. Differences in (A) median chromogranin A serum concentrations and (B) median urinary 5-HIAA after surgery and at relapse were calculated by Wilcoxon matched pairs signed-rank test. Paired row values, median change, and interquartile range are represented. All statistical tests were two-sided. CgA = chromogranin A; 5-HIAA = 5-hydroxyindoleacetic acid. Predictors of Recurrence As detailed in Table 2, the number of lymph nodes excised at the time of surgical resection was the only clinic-pathologic feature predictive of relapse in univariate analysis. However, the relative homogeneity of the study cohort and the consequent presence of subgroups with small numbers may have limited our ability to detect statistically significant differences, as suggested by wide confidence intervals around recurrence rates. To identify the optimal cutoff number of harvested lymph nodes capable of predicting relapse, ROC curve analysis was performed (Figure 3). In our cohort, a lymph node threshold of 17 or fewer distinguished between relapsing and nonrelapsing patients with a sensitivity of 88.2% (95% CI =  72.5% to 96.7%) and a specificity of 35.4% (95% CI =  25.1% to 46.7%), with an AUC of 0.63 (95% CI =  0.54 to 0.72, P = .01). By univariate analysis, a number of harvested lymph nodes of 17 or fewer statistically significantly predicted recurrence (P = .01). After adjusting for variables that showed a P value of less than .20 in univariate analysis (incidental diagnosis, N stage, tumor size), dissection of more than 17 lymph nodes remained associated with reduced risk of tumor recurrence in multivariable analysis (odds ratio = 0.21, 95% CI =  0.06 to 0.80, P = .02). Table 2. Univariate analysis of potential predictors of recurrence Characteristics  Recurrence rate (95% CI), %  P*  Median DFS (95% CI), mo  P†  Age at diagnosis, y    .98    .72   20–40  25.0 (7.1 to 59.1)    NR (11 to NR)     40–60  31.8 (21.8 to 43.8)    138 (124 to 139)     >60  30.9 (20.3 to 44)    117 (86 to 223)    Sex    .71    .83   Male  32.8 (22.3 to 45.3)    134 (87 to 223)     Female  29.4 (19.9 to 41.1)    NR (96 to NR)    Race    .22    .06   White  31.1 (23.1 to 40.5)    144 (124 to 223)     Black  46.1 (23.2 to 70.8)    87 (36 to 117)     Hispanic, Asian, unknown  11.1 (2.0 to 43.5)    NR (139 to NR)    Incidental diagnosis    .17    .33   Yes  37.5 (26.7 to 49.7)    NR (87 to NR)     No  21.6 (12.5 to 34.6)    124 (87 to 144)     Unknown  35.7 (16.3 to 61.2)    NR (96 to NR)    Carcinoid syndrome    .78    .79   Yes  35.3 (17.3 to 58.7)    134 (86 to 134)     No  30.4 (22.6 to 39.4)    139 (117 to 223)    Tumor location    .58    .33   Ileum  32.4 (24.4 to 41.6)    138 (96 to 223)     Others  22.2 (9.0 to 45.2)    NR (87 to NR)    Tumor multifocality    .30    .09   Yes  30.8 (18.6 to 46.4)    139 (68 to 223)     No  32.9 (23.9 to 43.5)    134 (96 to 144)    Type of surgery    .70    .33   Right hemicolectomy  29.0 (19.6 to 40.6)    138 (96 to 144)     Partial small bowel resection  32.8 (22.1 to 45.6)    139 (87 to 223)    Tumor size, cm    .07    .08   <1  19.2 (8.5 to 37.9)    124 (124 to 125)     1–2  34.8 (22.7 to 49.2)    87 (68 to 138)     2–4  37.2 (24.3 to 52.1)    134 (87 to 144)     >4  60.0 (23.1 to 88.2)    223 (3 to NR)     Unknown  0 (0 to 30)    NR (NR to NR)    Degree of parietal invasion    .49    .26   Submucosa  16.7 (3.0 to 56.3)    NR (134 to NR)     Muscularis propria  31.3 (18.0 to 48.6)    96 (87 to 139)     Subserosa  47.4 (27.3 to 68.3)    144 (36 to NR)     Serosa/mesenteric fat  29.2 (18.2 to 43.2)    124 (87 to 223)    Lymph node metastases    .20    .65   N0  15.8 (5.5 to 37.6)    NR (NR to NR)     N1  32.3 (23.9 to 42.0)    138 (96 to 223)     Nx  45.4 (21.3 to 72.0)    NR (24 to NR)    Lymph nodes harvested    .01    .03   ≤17  36.1 (26.6 to 46.9)    134 (87 to 144)     >17  12.1 (4.8 to 27.3)    NR (87 to NR)    Macroscopic mesenteric lymph nodes    .39    .64   Yes  33.3 (20.2 to 49.7)    139 (87 to 223)     No  31.5 (22.7 to 41.7)    134 (96 to 144)    Tumor stage    .58    .40   I  0 (0 to 39)    NR (NR to NR)     II  35.7 (16.3 to 61.2)    NR (25 to NR)     III  31.7 (23.6 to 41.2)    138 (96 to 223)    Tumor grade    .88    .25   G1  31.8 (23.7 to 41.1)    138 (117 to 223)     G2  30.0 (14.5 to 51.9)    81 (66 to 96)    Involvement of radial margins    .50    .33   Yes  33.3 (15.2 to 58.3)    139 (41 to NR)     No  32.1 (23.9 to 41.5)    134 (96 to 144)    Lymphovascular invasion    .78    .55   Yes  32.9 (23.0 to 44.5)    139 (87 to 144)     No  23.1 (8.2 to 50.3)    87 (87 to 88)     Unknown  30.4 (19.1 to 44.8)    138 (117 to 223)    Perineural invasion    .24    .31   Yes  34.0 (22.7 to 47.4)    144 (78 to 144)     No  15.0 (5.2 to 36.0)    124 (96 to NR)     Unknown  33.9 (22.9 to 47.0)    139 (117 to 223)    Characteristics  Recurrence rate (95% CI), %  P*  Median DFS (95% CI), mo  P†  Age at diagnosis, y    .98    .72   20–40  25.0 (7.1 to 59.1)    NR (11 to NR)     40–60  31.8 (21.8 to 43.8)    138 (124 to 139)     >60  30.9 (20.3 to 44)    117 (86 to 223)    Sex    .71    .83   Male  32.8 (22.3 to 45.3)    134 (87 to 223)     Female  29.4 (19.9 to 41.1)    NR (96 to NR)    Race    .22    .06   White  31.1 (23.1 to 40.5)    144 (124 to 223)     Black  46.1 (23.2 to 70.8)    87 (36 to 117)     Hispanic, Asian, unknown  11.1 (2.0 to 43.5)    NR (139 to NR)    Incidental diagnosis    .17    .33   Yes  37.5 (26.7 to 49.7)    NR (87 to NR)     No  21.6 (12.5 to 34.6)    124 (87 to 144)     Unknown  35.7 (16.3 to 61.2)    NR (96 to NR)    Carcinoid syndrome    .78    .79   Yes  35.3 (17.3 to 58.7)    134 (86 to 134)     No  30.4 (22.6 to 39.4)    139 (117 to 223)    Tumor location    .58    .33   Ileum  32.4 (24.4 to 41.6)    138 (96 to 223)     Others  22.2 (9.0 to 45.2)    NR (87 to NR)    Tumor multifocality    .30    .09   Yes  30.8 (18.6 to 46.4)    139 (68 to 223)     No  32.9 (23.9 to 43.5)    134 (96 to 144)    Type of surgery    .70    .33   Right hemicolectomy  29.0 (19.6 to 40.6)    138 (96 to 144)     Partial small bowel resection  32.8 (22.1 to 45.6)    139 (87 to 223)    Tumor size, cm    .07    .08   <1  19.2 (8.5 to 37.9)    124 (124 to 125)     1–2  34.8 (22.7 to 49.2)    87 (68 to 138)     2–4  37.2 (24.3 to 52.1)    134 (87 to 144)     >4  60.0 (23.1 to 88.2)    223 (3 to NR)     Unknown  0 (0 to 30)    NR (NR to NR)    Degree of parietal invasion    .49    .26   Submucosa  16.7 (3.0 to 56.3)    NR (134 to NR)     Muscularis propria  31.3 (18.0 to 48.6)    96 (87 to 139)     Subserosa  47.4 (27.3 to 68.3)    144 (36 to NR)     Serosa/mesenteric fat  29.2 (18.2 to 43.2)    124 (87 to 223)    Lymph node metastases    .20    .65   N0  15.8 (5.5 to 37.6)    NR (NR to NR)     N1  32.3 (23.9 to 42.0)    138 (96 to 223)     Nx  45.4 (21.3 to 72.0)    NR (24 to NR)    Lymph nodes harvested    .01    .03   ≤17  36.1 (26.6 to 46.9)    134 (87 to 144)     >17  12.1 (4.8 to 27.3)    NR (87 to NR)    Macroscopic mesenteric lymph nodes    .39    .64   Yes  33.3 (20.2 to 49.7)    139 (87 to 223)     No  31.5 (22.7 to 41.7)    134 (96 to 144)    Tumor stage    .58    .40   I  0 (0 to 39)    NR (NR to NR)     II  35.7 (16.3 to 61.2)    NR (25 to NR)     III  31.7 (23.6 to 41.2)    138 (96 to 223)    Tumor grade    .88    .25   G1  31.8 (23.7 to 41.1)    138 (117 to 223)     G2  30.0 (14.5 to 51.9)    81 (66 to 96)    Involvement of radial margins    .50    .33   Yes  33.3 (15.2 to 58.3)    139 (41 to NR)     No  32.1 (23.9 to 41.5)    134 (96 to 144)    Lymphovascular invasion    .78    .55   Yes  32.9 (23.0 to 44.5)    139 (87 to 144)     No  23.1 (8.2 to 50.3)    87 (87 to 88)     Unknown  30.4 (19.1 to 44.8)    138 (117 to 223)    Perineural invasion    .24    .31   Yes  34.0 (22.7 to 47.4)    144 (78 to 144)     No  15.0 (5.2 to 36.0)    124 (96 to NR)     Unknown  33.9 (22.9 to 47.0)    139 (117 to 223)    * Calculated by χ2 or Fisher’s exact test, as appropriate. Both tests were two-sided. CI = confidence interval; DFS = disease free survival; NR = not reached. † Calculated by two-sided log-rank test. Table 2. Univariate analysis of potential predictors of recurrence Characteristics  Recurrence rate (95% CI), %  P*  Median DFS (95% CI), mo  P†  Age at diagnosis, y    .98    .72   20–40  25.0 (7.1 to 59.1)    NR (11 to NR)     40–60  31.8 (21.8 to 43.8)    138 (124 to 139)     >60  30.9 (20.3 to 44)    117 (86 to 223)    Sex    .71    .83   Male  32.8 (22.3 to 45.3)    134 (87 to 223)     Female  29.4 (19.9 to 41.1)    NR (96 to NR)    Race    .22    .06   White  31.1 (23.1 to 40.5)    144 (124 to 223)     Black  46.1 (23.2 to 70.8)    87 (36 to 117)     Hispanic, Asian, unknown  11.1 (2.0 to 43.5)    NR (139 to NR)    Incidental diagnosis    .17    .33   Yes  37.5 (26.7 to 49.7)    NR (87 to NR)     No  21.6 (12.5 to 34.6)    124 (87 to 144)     Unknown  35.7 (16.3 to 61.2)    NR (96 to NR)    Carcinoid syndrome    .78    .79   Yes  35.3 (17.3 to 58.7)    134 (86 to 134)     No  30.4 (22.6 to 39.4)    139 (117 to 223)    Tumor location    .58    .33   Ileum  32.4 (24.4 to 41.6)    138 (96 to 223)     Others  22.2 (9.0 to 45.2)    NR (87 to NR)    Tumor multifocality    .30    .09   Yes  30.8 (18.6 to 46.4)    139 (68 to 223)     No  32.9 (23.9 to 43.5)    134 (96 to 144)    Type of surgery    .70    .33   Right hemicolectomy  29.0 (19.6 to 40.6)    138 (96 to 144)     Partial small bowel resection  32.8 (22.1 to 45.6)    139 (87 to 223)    Tumor size, cm    .07    .08   <1  19.2 (8.5 to 37.9)    124 (124 to 125)     1–2  34.8 (22.7 to 49.2)    87 (68 to 138)     2–4  37.2 (24.3 to 52.1)    134 (87 to 144)     >4  60.0 (23.1 to 88.2)    223 (3 to NR)     Unknown  0 (0 to 30)    NR (NR to NR)    Degree of parietal invasion    .49    .26   Submucosa  16.7 (3.0 to 56.3)    NR (134 to NR)     Muscularis propria  31.3 (18.0 to 48.6)    96 (87 to 139)     Subserosa  47.4 (27.3 to 68.3)    144 (36 to NR)     Serosa/mesenteric fat  29.2 (18.2 to 43.2)    124 (87 to 223)    Lymph node metastases    .20    .65   N0  15.8 (5.5 to 37.6)    NR (NR to NR)     N1  32.3 (23.9 to 42.0)    138 (96 to 223)     Nx  45.4 (21.3 to 72.0)    NR (24 to NR)    Lymph nodes harvested    .01    .03   ≤17  36.1 (26.6 to 46.9)    134 (87 to 144)     >17  12.1 (4.8 to 27.3)    NR (87 to NR)    Macroscopic mesenteric lymph nodes    .39    .64   Yes  33.3 (20.2 to 49.7)    139 (87 to 223)     No  31.5 (22.7 to 41.7)    134 (96 to 144)    Tumor stage    .58    .40   I  0 (0 to 39)    NR (NR to NR)     II  35.7 (16.3 to 61.2)    NR (25 to NR)     III  31.7 (23.6 to 41.2)    138 (96 to 223)    Tumor grade    .88    .25   G1  31.8 (23.7 to 41.1)    138 (117 to 223)     G2  30.0 (14.5 to 51.9)    81 (66 to 96)    Involvement of radial margins    .50    .33   Yes  33.3 (15.2 to 58.3)    139 (41 to NR)     No  32.1 (23.9 to 41.5)    134 (96 to 144)    Lymphovascular invasion    .78    .55   Yes  32.9 (23.0 to 44.5)    139 (87 to 144)     No  23.1 (8.2 to 50.3)    87 (87 to 88)     Unknown  30.4 (19.1 to 44.8)    138 (117 to 223)    Perineural invasion    .24    .31   Yes  34.0 (22.7 to 47.4)    144 (78 to 144)     No  15.0 (5.2 to 36.0)    124 (96 to NR)     Unknown  33.9 (22.9 to 47.0)    139 (117 to 223)    Characteristics  Recurrence rate (95% CI), %  P*  Median DFS (95% CI), mo  P†  Age at diagnosis, y    .98    .72   20–40  25.0 (7.1 to 59.1)    NR (11 to NR)     40–60  31.8 (21.8 to 43.8)    138 (124 to 139)     >60  30.9 (20.3 to 44)    117 (86 to 223)    Sex    .71    .83   Male  32.8 (22.3 to 45.3)    134 (87 to 223)     Female  29.4 (19.9 to 41.1)    NR (96 to NR)    Race    .22    .06   White  31.1 (23.1 to 40.5)    144 (124 to 223)     Black  46.1 (23.2 to 70.8)    87 (36 to 117)     Hispanic, Asian, unknown  11.1 (2.0 to 43.5)    NR (139 to NR)    Incidental diagnosis    .17    .33   Yes  37.5 (26.7 to 49.7)    NR (87 to NR)     No  21.6 (12.5 to 34.6)    124 (87 to 144)     Unknown  35.7 (16.3 to 61.2)    NR (96 to NR)    Carcinoid syndrome    .78    .79   Yes  35.3 (17.3 to 58.7)    134 (86 to 134)     No  30.4 (22.6 to 39.4)    139 (117 to 223)    Tumor location    .58    .33   Ileum  32.4 (24.4 to 41.6)    138 (96 to 223)     Others  22.2 (9.0 to 45.2)    NR (87 to NR)    Tumor multifocality    .30    .09   Yes  30.8 (18.6 to 46.4)    139 (68 to 223)     No  32.9 (23.9 to 43.5)    134 (96 to 144)    Type of surgery    .70    .33   Right hemicolectomy  29.0 (19.6 to 40.6)    138 (96 to 144)     Partial small bowel resection  32.8 (22.1 to 45.6)    139 (87 to 223)    Tumor size, cm    .07    .08   <1  19.2 (8.5 to 37.9)    124 (124 to 125)     1–2  34.8 (22.7 to 49.2)    87 (68 to 138)     2–4  37.2 (24.3 to 52.1)    134 (87 to 144)     >4  60.0 (23.1 to 88.2)    223 (3 to NR)     Unknown  0 (0 to 30)    NR (NR to NR)    Degree of parietal invasion    .49    .26   Submucosa  16.7 (3.0 to 56.3)    NR (134 to NR)     Muscularis propria  31.3 (18.0 to 48.6)    96 (87 to 139)     Subserosa  47.4 (27.3 to 68.3)    144 (36 to NR)     Serosa/mesenteric fat  29.2 (18.2 to 43.2)    124 (87 to 223)    Lymph node metastases    .20    .65   N0  15.8 (5.5 to 37.6)    NR (NR to NR)     N1  32.3 (23.9 to 42.0)    138 (96 to 223)     Nx  45.4 (21.3 to 72.0)    NR (24 to NR)    Lymph nodes harvested    .01    .03   ≤17  36.1 (26.6 to 46.9)    134 (87 to 144)     >17  12.1 (4.8 to 27.3)    NR (87 to NR)    Macroscopic mesenteric lymph nodes    .39    .64   Yes  33.3 (20.2 to 49.7)    139 (87 to 223)     No  31.5 (22.7 to 41.7)    134 (96 to 144)    Tumor stage    .58    .40   I  0 (0 to 39)    NR (NR to NR)     II  35.7 (16.3 to 61.2)    NR (25 to NR)     III  31.7 (23.6 to 41.2)    138 (96 to 223)    Tumor grade    .88    .25   G1  31.8 (23.7 to 41.1)    138 (117 to 223)     G2  30.0 (14.5 to 51.9)    81 (66 to 96)    Involvement of radial margins    .50    .33   Yes  33.3 (15.2 to 58.3)    139 (41 to NR)     No  32.1 (23.9 to 41.5)    134 (96 to 144)    Lymphovascular invasion    .78    .55   Yes  32.9 (23.0 to 44.5)    139 (87 to 144)     No  23.1 (8.2 to 50.3)    87 (87 to 88)     Unknown  30.4 (19.1 to 44.8)    138 (117 to 223)    Perineural invasion    .24    .31   Yes  34.0 (22.7 to 47.4)    144 (78 to 144)     No  15.0 (5.2 to 36.0)    124 (96 to NR)     Unknown  33.9 (22.9 to 47.0)    139 (117 to 223)    * Calculated by χ2 or Fisher’s exact test, as appropriate. Both tests were two-sided. CI = confidence interval; DFS = disease free survival; NR = not reached. † Calculated by two-sided log-rank test. Figure 3. View largeDownload slide Lymph node count and tumor recurrence. Receiver operating characteristic curve showing the relationship between sensitivity and false-positive rate. Figure 3. View largeDownload slide Lymph node count and tumor recurrence. Receiver operating characteristic curve showing the relationship between sensitivity and false-positive rate. We also evaluated a lower lymph node threshold of 12 based on data on minimal adequate evaluation in colorectal cancer (14). On univariate analysis, a number of harvested lymph nodes of less than 12 was associated with increased tumor recurrence in a statistically significant fashion (P = .04). Disease-Free Survival DFS was measured from time of resection until the earliest evidence of relapse, death from any cause, or last clinical contact. The median DFS was 138 months (95% CI =  117 to 223 months) (Figure 4A). The five- and 10-year DFS rates were 81.6% (±3.8%) and 59.5% (±5.9%). There was no difference between TTP and DFS because no death was recorded before tumor recurrence. As illustrated in Figure 4B, patients who underwent resection of more than 17 lymph nodes had a superior DFS as compared with those who were subjected to dissection of 17 or fewer lymph nodes (median DFS = 134 months vs not reached, respectively, P = .04). In multivariable analysis, statistical significance was retained after controlling for race, tumor multifocality, and tumor size (P = .04). All remaining clinical-pathologic factors were not statistically significantly associated with changes in DFS (Table 2). Figure 4. View largeDownload slide Kaplan-Meier estimates of disease-free survival (DFS). A) DFS in the overall population. B) DFS by number of harvested lymph nodes. Figure 4. View largeDownload slide Kaplan-Meier estimates of disease-free survival (DFS). A) DFS in the overall population. B) DFS by number of harvested lymph nodes. Immortal Time Bias To mitigate the immortal time bias, patients who were seen at our institution within 12 months of their diagnosis (n = 109, 84.5%) were analyzed separately (data not shown). The median DFS of patients referred within 12 months of diagnosis was 134 months (95% CI =  87 to 144 months), whereas the median DFS was not reached (95% CI =  223 months to not reached) in patients referred after 12 months (P = .01). Overall Survival Five deaths occurred among the 129 patients in this cohort. Of the four deaths recorded among relapsing patients, two deaths were known to be directly related to progressive metastatic disease, one was possibly related to disease progression, and one was not tumor related (data not shown). Of the two deaths clearly attributable to progressive metastatic disease, one occurred nine years after initial surgery and seven years after metastatic relapse, while the other was recorded 12 years after initial surgery and seven years after relapse. All the remaining 36 patients with relapsed disease were alive at the time of data cutoff. Among relapsing patients, the 10-year and 15-year OS rates were 92.9% (±5.0%) and 74.0% (±13.3%). Discussion Data on risk of recurrence after resection of stage I–III midgut NETs are scarce. Retrospective series (5,6) have been limited by small size, relatively short median follow-up, and inclusion of patients who recurred prior to referral to the institution conducting the study. These factors can lead to either overestimation or underestimation of recurrence risk. Yet accurate information on probability of recurrence is critically important for several reasons: postoperative surveillance guidelines need to draw upon these data to develop recommendations on timing and duration of follow-up evaluations, adjuvant studies require estimates of baseline risk of recurrence, and patients seek accurate information on their personal risk of recurrence after surgery. Past multi-institutional series have demonstrated that the large majority of patients presenting with midgut NETs have evidence of lymph node involvement on surgical resection. Metastatic recurrence rates in these patients have been reported to be roughly 50%, with an increase in risk associated with more distant mesenteric lymph node metastases and multifocal primary tumors (which are often associated with gross lymphadenopathy) (5,6,13). Our database of 129 patients who underwent resection of stage I–III midgut NETs represents, to our knowledge, the largest single or multi-institutional series to examine this specific population. Results of our analysis show that 76.8% of patients presented with lymph node–positive disease. This is similar to a rate of 79% observed in a multi-institutional French study of 100 patients (5). The median follow-up of 81 months allows us to evaluate long-term risk of recurrence with reasonable accuracy, although estimates of cumulative recurrence risk after five years are limited by large confidence intervals due to small numbers. Among all patients, the recurrence rate in our series was 31.0% (95% CI =  23.0% to 39.0%). Among lymph node–positive patients (n = 99), the long-term recurrence rate was 32.3%, with a 95% confidence interval of 23.1% to 41.5%. This is also in accordance with the prior 100-patient series (5) which revealed a 42% (95% CI =  32% to 52%) cumulative rate of recurrence (median follow-up = 57.5 months). An interesting finding in our study is that patients with stage I tumors (n = 6) had no evidence of recurrence, suggesting that such patients may be able to forgo surveillance or undergo a relatively low-intensity postoperative surveillance schedule. However, larger series of stage I patients are needed to draw firmer conclusions regarding recurrence risk. The rate of recurrence in our database was relatively steady over an eight-year period, with a subsequent dropoff. It is uncertain whether this finding reflects a true decline in recurrence after eight years or a statistical artifact (only 25% of patients had surveillance beyond eight years). We conclude from these data that surveillance for recurrence does not need to be particularly frequent during the initial years after surgery. Indeed, an annual surveillance interval will likely allow early detection of most patients with metastatic disease. It is hard to recommend an outer limit for duration of surveillance given the fact that recurrences can occur after 10 years. However, given the apparent decline in recurrence after eight years, a decade-long duration of active surveillance will likely enable detection of most recurrences. In our series, the small number of patients with baseline elevated 5-HIAA or CgA hinders reliable conclusions regarding the role of tumor markers in the postoperative surveillance of patients with midgut carcinoid. While CgA did not appear to be a robust tumor marker for detection of metastatic recurrence, urine 5-HIAA levels did increase at the time of diagnosis of recurrence. However, it is difficult to ascertain whether measurement of urine 5-HIAA adds diagnostic value beyond cross-sectional imaging. Among the clinical-pathologic features evaluated in this study, only the number of harvested lymph nodes statistically significantly predicted recurrence in both univariate (P = .01) and multivariable analysis (P = .02). While ROC analysis identified 17 lymph nodes as most predictive of recurrence, even removal of 12 lymph nodes was associated with a statistically significant reduction in recurrence risk. Consistent with the evidence that removal of mesenteric lymph node metastases positively impacts survival in patients with midgut NETs (15), our data suggest that an appropriate lymphadenectomy, rather than simple segmental bowel resection, is preferable in midgut NETs. With respect to other pathological prognostic factors, the statistical power of our study is limited by relatively small numbers of patients with grade 2 and lymph node–negative tumors. In our analysis, we were only able to confirm two deaths that were clearly attributable to disease progression. Given the median age of diagnosis (57 years) and very long survival of patients with early detection of metastatic disease on surveillance, it is unclear to what extent a diagnosis of early-stage midgut NET impacts overall life expectancy. It is likely that competing causes of death exceed NET-related deaths in this population. This has important implications regarding adjuvant therapy trials as it would be difficult to demonstrate that delay in recurrence impacts survival. The main limitation of our study is incomplete follow-up, either due to loss of patients from the practice (patients undergoing surveillance at other institutions or choosing to forgo surveillance), or due to the fact that some patients were diagnosed within the past decade and therefore have not completed a 10-year course of surveillance. Therefore, it is possible that our recurrence rate of 31.0% represents an underestimation of the true long-term recurrence rate. An ideal estimation of recurrence risk would require a prospective database with a fixed surveillance schedule and a very long duration of follow-up. Note The authors declare no conflicts of interest. References 1 Frilling A, Akerström G, Falconi Met al.  , Neuroendocrine tumor disease: An evolving landscape. Endocr Relat Cancer . 2012; 19: R163– R185. Google Scholar CrossRef Search ADS PubMed  2 Yao JC, Hassan M, Phan Aet al.  , One hundred years after “carcinoid”: Epidemiology of and prognostic factors for neuroendocrine tumors in 35,825 cases in the United States. J Clin Oncol . 2008; 26: 3063– 3072. Google Scholar CrossRef Search ADS PubMed  3 Cives M, Strosberg J. An update on gastroenteropancreatic neuroendocrine tumors. Oncology (Williston Park) . 2014; 28: 749– 756. Google Scholar PubMed  4 Mocellin S, Nitti D. Gastrointestinal carcinoid: Epidemiological and survival evidence from a large population-based study (n = 25 531). Ann Oncol.  2013; 24: 3040– 3044. Google Scholar CrossRef Search ADS PubMed  5 Le Roux C, Lombard-Bohas C, Delmas Cet al.  , Relapse factors for ileal neuroendocrine tumours after curative surgery: A retrospective French multicentre study. Dig Liver Dis . 2011; 43: 828– 833. Google Scholar CrossRef Search ADS PubMed  6 Dieckhoff P, Runkel H, Daniel Het al.  , Well-differentiated neuroendocrine neoplasia: relapse-free survival and predictors of recurrence after curative intended resections. Digestion . 2014; 90: 89– 97. Google Scholar CrossRef Search ADS PubMed  7 Rindi G, Klöppel G, Couvelard Aet al.  , TNM staging of midgut and hindgut (neuro) endocrine tumors: A consensus proposal including a grading system. Virchows Arch.  2007; 451: 757– 762. Google Scholar CrossRef Search ADS PubMed  8 Edge S, Byrd DR, Compton CCet al.  , , eds. AJCC Cancer Staging Manual . 7th ed. Chicago: Springer; 2010. 9 Rindi G, Arnold R. Nomenclature and classification of neuroendocrine neoplasms of the digestive system. In: Bosman F, Carneiro F, Hruban R, Theise N, eds. World Health Organization Classification of Tumours of the digestive system . Lyon: IARC Press; 2010: 13– 14. 10 Strosberg JR, Weber JM, Feldman Met al.  , Prognostic validity of the American Joint Committee on Cancer staging classification for midgut neuroendocrine tumors. J Clin Oncol . 2013; 31: 420– 425. Google Scholar CrossRef Search ADS PubMed  11 DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: A nonparametric approach. Biometrics . 1988; 44 3: 837– 845. Google Scholar CrossRef Search ADS PubMed  12 Shariff SZ, Cuerden MS, Jain AKet al.  , The secret of immortal time bias in epidemiologic studies. J Am Soc Nephrol.  2008; 19: 841– 843. Google Scholar CrossRef Search ADS PubMed  13 Welin S, Stridsberg M, Cunningham Jet al.  , Elevated plasma chromogranin A is the first indication of recurrence in radically operated midgut carcinoid tumors. Neuroendocrinology . 2009; 89: 302– 307. Google Scholar CrossRef Search ADS PubMed  14 Nelson H, Petrelli N, Carlin Aet al.  , Guidelines 2000 for colon and rectal cancer surgery. J Natl Cancer Inst.  2001; 93: 583– 596. Google Scholar CrossRef Search ADS PubMed  15 Hellman P, Lundström T, Ohrvall Uet al.  , Effect of surgery on the outcome of midgut carcinoid disease with lymph node and liver metastases. World J Surg . 2002; 26: 991– 997. Google Scholar CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.

Journal

JNCI: Journal of the National Cancer InstituteOxford University Press

Published: Sep 26, 2017

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off