Preoperative Serum Chromogranin-A is Predictive of Survival in Locoregional Jejuno-ileal Small Bowel Neuroendocrine Tumors

Abstract

Background:

Small bowel neuroendocrine tumors (SB-NET) frequently metastasize to regional lymphatic or distant sites. While most prognostication of SB-NET focuses on lymph node involvement, findings from studies of NETs from other primary sites have suggested that preoperative serum chromogranin-A (CgA) levels may provide a more accurate metric.

Study Design:

Using the National Cancer Database (2004–2016), we analyzed patients with locoregional SB-NET who underwent curative resection including an adequate lymphadenectomy (n = 1,274). A statistically optimized cut-point was used to dichotomize CgA cohort based on preoperative serum CgA levels.

Results:

We determined that a CgA ≥139ng/mL identified patients with significantly shorter estimated mean overall survival (6.6 years vs. 7.6 years, log-rank p = 0.00001). These patients were also older (63 vs. 57 years, p < 0.001) and more likely to have poorly-differentiated tumors (2.1% vs. 0.7%, p = 0.04) or primary tumors >1cm (88.2% vs. 79.2%, p = 0.001). Clinical features associated with shorter overall survival included preoperative CgA ≥139ng/mL (HR = 2.19, 95% CI 1.22 – 3.92; p = 0.009), age at diagnosis (HR = 1.06, 95% CI 1.03 – 1.09; p < 0.001), Charlson-Deyo score ≥2 (HR = 3.93, 95% CI 1.71 – 9.01; p = 0.001), and poorly-differentiated tumors (HR = 11.22, 95% CI 4.16 – 30.24; p < 0.001). Neither lymph node metastasis nor T-stage were independently associated with shorter overall survival in patients with locoregional SB-NET.

Conclusions:

Elevated preoperative serum CgA is an adverse prognostic marker associated with shorter overall survival in patients with locoregional SB-NET.

Two Sentence Summary:

“Elevated preoperative serum chromogranin-A is an adverse prognostic marker independently associated with shorter overall survival in patients with locoregional SB-NET. Neither lymph node metastasis nor T-stage is independently associated with shorter overall survival.”

Introduction

Small bowel neuroendocrine tumors (SB-NETs) arise from enterochromaffin cells and account for over one-third of all small bowel tumors.¹ Their incidence has risen 6-fold over the last 40 years, and the presence of associated regional metastases at presentation has more than doubled.² Characteristic patterns of SB-NET growth and spread parallel those of NETs from other visceral organs through regional lymphatic and distant hepatic metastases, but SB-NETs also seem to be biochemically unique, as they are more commonly associated with carcinoid syndrome and less often with mutational driver events.^3–6 Despite their propensity for metastasis, NETs originating from most primary sites often exhibit an indolent course that allows for survival-prolonging cytoreductive procedures even in the presence of distant disease. Furthermore, the advancement of systemic treatment options for NETs is promising and metastatic SB-NETs have the highest median overall survival of NETs arising from any visceral organ.² To provide appropriate expectations with respect to treatment and surveillance for patients with NET, it is imperative to better understand the disease’s biological behavior and response to available therapy. Thus, advancements in staging accuracy and prognostication, longstanding areas of intense clinical investigation, remain of the utmost importance.^{7, 8}

While there are limits to radiographic metrics to detect disease, the use of biomarkers has shown promise in the prognostication of select tumors, including NETs. Much like NETs of other primary sites, SB-NETs can secrete one or more active peptides with potential functional and prognostic relevance. One such peptide is Chromogranin A (CgA). CgA is a hydrophilic glycoprotein that is cleaved into bioactive peptides upon release in chromaffin granules of neuroendocrine cells.⁹ Serum CgA levels accurately reflect the clinical status of patients with advanced disease, and have demonstrated utility in both tumor surveillance and assessment of treatment response.¹⁰ An elevated serum CgA is frequently used as both a diagnostic and prognostic adjunct for NETs, given its correlation with metastatic disease burden and association with disease-specific-survival in several NET types; however, its significance in locoregional disease remains poorly defined.^11–15 Elevated preoperative serum CgA levels in localized pancreatic NETs has been used to identify patients with tumors who may benefit from resection, though a similar prognostic role for preoperative CgA levels in SB-NET has yet to be established.¹³

In recent years, most investigations aiming to improve prognostication in SB-NET have centered on assaying lymphatic metastasis through histopathologic examination. While a regional lymphadenectomy remains standard surgical practice, the prognostic significance of lymph node metastases and the staging system used to classify them remains heavily contested.^16–18 A recent evaluation of the American Joint Committee on Cancer (AJCC) 8^th edition staging classification using 1,925 patients with locoregional SB-NET found no association between lymph node stage and overall survival.¹⁹ Additionally, patients with low-volume lymph node metastases (<4 nodes involved) have demonstrated similar recurrence-free survival to those without lymphatic spread following resection with adequate lymphadenectomy (≥8 nodes resected).¹⁶ Furthermore, a novel classification system utilizing lymph-node-ratio more accurately stratified patients with locoregional SB-NETs by overall survival than the present AJCC classification.¹⁷ Given the unvalidated prognostic value of lymph node metastasis and the lack of data on the utility of preoperative CgA in patients with locoregional SB-NET, we aimed to further elucidate the prognostic capacity of clinical features in SB-NET by using a large, national cancer registry database. Specifically, we sought to evaluate the prognostic relevance of lymph node spread and the utility of elevated preoperative serum CgA in predicting outcomes for locoregional SB-NET. We hypothesized that preoperative serum chromogranin-A would be more prognostically useful than nodal metastasis in predicting overall survival in patients with locoregional SB-NET.

Materials & Methods

Data Source

The National Cancer Database participant user files (NCDB PUFs) were the source of all data in our study. The NCDB is a nationwide repository of de-identified patient data related to cancer care metrics and outcomes in the United States derived from the submissions of over 1,500 Commission on Cancer (CoC)-accredited programs.²⁰ The NCDB captures over 70% of new cancer diagnoses in the United States per year. The CoC is a multidisciplinary association maintained by the American College of Surgeons and the American Cancer Society that accredits US hospitals based on various aspects of cancer care. Due to our study’s inclusion of only de-identified data, it was exempt from institutional review board approval at the National Institutes of Health.

Patient Selection

Patient selection criteria are outlined in Figure 1. The NCDB was queried for patients from 2010 to 2016 with neuroendocrine tumors of the jejunum and ileum using the International Classification of Diseases for Oncology, 3^rd Edition (ICD-O-3) topography codes (C171, C172, C178, and C179) and histology codes (8240, 8241, 8246, 8249, and 8467). Preoperative serum chromogranin-A values are largely missing from the NCDB prior to 2010, and thus, only 5 patients were excluded by this cut-off. Of note, we excluded patients with duodenal tumors due to the high incidence of gastrinoma (and thus, proton-pump inhibitor usage) from the proximal small bowel due to potential false elevation of serum CgA levels. Additionally, we excluded all patients with evidence of distant metastatic disease to control for the effect of tumor burden on CgA levels. Patients who had a prior history of more than one prior malignancy, or those who did not undergo resection, underwent a margin-positive resection (R1 or R2), those with missing nodal (or other pertinent) information, and those who underwent an inadequate lymphadenectomy (<8 lymph nodes examined) were excluded.²¹ Staging was determined as per the American Joint Committee on Cancer (AJCC), 8^th Edition clinical staging system. Among the 4,589 patients who met our inclusion criteria, 1,269 (27.6%) had preoperative serum CgA levels recorded. CgA levels reported in the NCDB represent the highest pre-treatment level in the medical record and are recorded to the nearest nanogram per milliliter (ng/mL). We collected demographic, clinicopathologic, and outcome-related data for all patients. Survival analysis for patients without reported preoperative CgA levels was performed separately.

Figure 1.

Consort diagram of patient inclusion and exclusion criteria.

Statistical Methods

The primary endpoint of the study was overall survival. Cut-point determination for optimal preoperative serum CgA level was performed using Contal & O’Quigley methods.²² Briefly, this model calculates the largest difference in patient mortality at each of the available preoperative serum CgA cut-points using the absolute log-rank statistic and tests the statistical significance at each point using a Q (maximization) statistic. The preoperative serum CgA level with the highest absolute log-rank and Q-statistic represents the optimal cut-off point, which, in this case, was 139ng/mL. Patients were divided into two groups based on their serum CgA levels, either ≥139ng/mL (high) or <139ng/mL (low).

Descriptive statistics were calculated for clinicopathologic and treatment variables, and associations between variables were analyzed by Pearson’s chi-square test and Mann-Whitney U test for normally and non-normally distributed data, respectively. This statistical method was also used to evaluate associations between patient and tumor characteristics and either high or low preoperative serum CgA levels. Univariable and multivariable Cox proportional hazards regression models were used to test for associations between clinicopathologic characteristics and overall survival. A two-tailed P <0.05 was considered statistically significant. The statistically significant clinical features associated with overall survival with a P <0.1 on the univariable analysis were then included in the multivariable analysis. Survival was further estimated using the Kaplan-Meier method and groups were compared using the log-rank test. Statistical analyses were performed using R 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria) and SPSS version 26.0 (IBM Corp., Armonk, N.Y., USA).

Results

A total of 4,589 patients met the inclusion criteria. Sociodemographic, clinicopathologic, and treatment characteristics of these patients are presented in Table 1. Most patients were Caucasian with a well-balanced sex distribution and had a median age of 60 years. Most tumors were well-differentiated, ileal lesions of at least 1cm in maximal diameter, and had a high rate of lymph node metastasis (pN1). Patients with higher N-stages had correspondingly increasing median numbers of regional lymph nodes examined (14 nodes [pN0] vs. 16 nodes [pN1] vs. 26 nodes [pN2], p < 0.001).

Table 1.

Demographics and clinical characteristics of patients with locoregional small bowel neuroendocrine tumors.

Variable	All Patients		Patients with CgA “CgA Cohort”		Patients without CgA		p-value
	n = 4589	%	n = 1269	%	n = 3320	%
Age at Diagnosis							0.0001
Mean (Range)	60 (18 – 90)		59 (22 – 90)		60 (18 – 90)
Median (IQR)	60 (51 – 69)		59 (51 – 67)		60 (52 – 69)
Sex							0.508
Male	2242	48.9%	630	49.6%	1612	48.6%
Female	2347	51.1%	639	50.4%	1708	51.4%
Race							0.029
White	3915	85.3%	1106	87.2%	2809	84.6%
Non-white	674	14.7%	163	12.8%	511	15.4%
Charlson-Deyo Score							0.57
0	3507	76.4%	976	76.9%	2531	76.2%
1	820	17.9%	228	18.0%	592	17.8%
≥2	262	5.7%	65	5.1%	197	5.9%
Number of Prior Malignancies							0.284
0	4361	95.0%	1213	95.6%	3148	94.8%
1	228	5.0%	56	4.4%	172	5.2%
Primary Site							0.0001
Jejunum	166	3.6%	52	4.1%	114	3.4%
Ileum	3383	73.7%	985	77.6%	2398	72.2%
Overlapping Lesions of SB	39	0.8%	13	1.0%	26	0.8%
SB NOS	1001	21.8%	219	17.3%	782	23.6%
Surgical Procedure							0.04
Simple / Partial Resection	2961	64.5%	790	62.3%	2171	65.4%
Total Resection	1307	28.5%	371	29.2%	936	28.2%
Debulking Surgery	11	0.2%	5	0.4%	6	0.2%
Radical Resection	310	6.8%	103	8.1%	207	6.2%
Tumor Size							0.143
≤1cm	931	20.7%	235	18.9%	696	21.3%
>1cm	3572	79.3%	1008	81.1%	2564	78.7%
Differentiation							0.452
Well Differentiated	3300	80.7%	918	79.8%	2382	81.0%
Moderately Differentiated	744	18.2%	221	19.2%	523	17.8%
Poorly Differentiated	47	1.1%	11	1.0%	36	1.2%
Median Pre-op CgA (IQR)	44 (8 – 117)		44 (8 – 117)		n/a		n/a
Mitotic Count							0.176
<2 / 10HPFs	2528	84.1%	751	82.3%	1777	84.9%
2–20 / 10HPFs	462	15.4%	157	17.2%	305	14.6%
>20 / 10HPFS	15	0.5%	4	0.4%	11	0.5%
pT Stage							0.069
T1	581	12.8%	145	11.5%	436	13.2%
T2	1582	34.7%	472	37.4%	1110	33.7%
T3	1750	38.4%	464	36.8%	1286	39.1%
T4	641	14.1%	180	14.3%	461	14.0%
pN Stage							0.0001
N0	599	13.1%	117	9.2%	482	14.5%
N1	3809	83.0%	1090	86.0%	2719	81.9%
N2	179	3.9%	61	4.8%	118	3.6%
Regional Nodes Examined							0.0001
Mean (Range)	19	(8–90)	20	(8–90)	18	(8–90)
Median (IQR)	16	(12–22)	17	(13–24)	16	(12–22)

The clinical and demographic characteristics of patients with preoperative serum CgA levels (the “CgA cohort”) were similar to those of the patients without CgA values available, except a lower rate of non-white patients (12.8% vs. 15.4%, p = 0.029) and a higher rate of pN1–2 disease (90.8% vs. 85.5%, p = 0.0001), likely attributable to an increased mean number of regional lymph nodes examined (20 vs. 18, p = 0.0001) (Table 1). The median CgA level was 44ng/mL, and the statistically optimized cut-point was 139ng/mL (Supplementary Figure 1). Overall, 1,269 patients were included in the preoperative serum CgA analysis, and most patients fell into the CgA-Low subgroup (n = 998, 78.6%). Compared to patients in the CgA-Low subgroup, patients in the CgA-High subgroup had a statistically significantly higher median age (63 vs. 57, p < 0.001), had higher rates of tumor size >1cm (88.2% vs. 79.2%, p = 0.001). A pairwise chi-square analysis showed no significant difference in the rate of well-differentiated NET by CgA groups. However, we observed a significantly higher rate of poorly differentiated (2.1% vs. 0.7%, p = 0.04) in the CgA-High group. Inversely, the CgA-Low group had a significantly higher rate of moderately-differentiated NETs (20.6% vs. 13.9%, p = 0.02). We found no statistically significant differences in T-stage, pathologic nodal stage, number of regional nodes examined, mitotic count, surgical procedure undergone, or comorbidity score by CgA subgroup (Table 2).

Table 2.

Demographics and clinical characteristics of patients with recorded preoperative serum chromogranin-A (CgA).

Variables	CgA Cohort		CgA Level
			<139		≥139		P
	n = 1269	%	n = 998	%	n = 271	%
Age at Diagnosis							<0.001
Mean (Range)	59 (22 – 90)		58 (22 – 89)		63 (30 – 90)
Median (IQR)	59 (51 – 67)		57 (50 – 65)		63 (53 – 72)
Sex							0.534
Male	630	49.6%	500	50.1%	130	47.8%
Female	639	50.4%	498	49.9%	142	52.2%
Race							0.391
White	1106	87.2%	874	87.6%	232	85.6%
Non-white	163	12.8%	124	12.4%	39	14.4%
Charlson-Deyo Score							0.12
0	976	76.9%	779	78.1%	197	72.7%
1	228	18.0%	173	17.3%	55	20.3%
≥2	65	5.1%	46	4.6%	19	7.0%
Number of Prior Malignancies							0.728
0	1213	95.6%	955	95.7%	258	95.2%
1	56	4.4%	43	4.3%	13	4.8%
Primary Site							0.001
Jejunum	52	4.1%	42	4.2%	10	3.7%
Ileum	985	77.6%	792	79.4%	193	71.2%
Overlapping Lesions of SB	13	1.0%	6	0.6%	7	2.6%
SB NOS	219	17.3%	158	15.8%	61	22.5%
Surgical Procedure							0.37
Simple / Partial Resection	790	62.3%	618	61.9%	172	63.5%
Total Resection	371	29.2%	288	28.9%	83	30.6%
Debulking Surgery	5	0.4%	4	0.4%	1	0.4%
Radical Resection	103	8.1%	88	8.8%	15	5.5%
Tumor Size							0.001
≤1cm	235	18.9%	204	20.8%	31	11.8%
>1cm	1008	81.1%	777	79.2%	231	88.2%
Adequate Lymphadenectomy (Y/N)
No	0	0.0	0	-	0	0.0%
Yes	1269	100.0	1002	100.0%	272	100.0%
Differentiation							0.01
Well Differentiated	918	79.8%	719	78.8%	199	84.0%
Moderately Differentiated	221	19.2%	188	20.6%	33	13.9%
Poorly Differentiated	11	1.0%	6	0.7%	5	2.1%
Median Pre-op Serum CgA (IQR)	44 (8 – 117)		27 (5 – 62)		314 (198 – 602)		<0.001
Preoperative Serum CgA
<139	998	78.6%	998	100.0%	0	-
≥139	271	21.4%	0	-	271	100%
Mitotic Count							0.14
<2 / 10HPFs	751	82.3%	608	83.3%	143	78.6%
2–20 / 10HPFs	157	17.2%	120	16.4%	37	20.3%
>20 / 10HPFS	4	0.4%	2	0.3%	2	1.1%
pT Stage							0.055
T1	145	11.5%	126	12.7%	19	7.1%
T2	472	37.4%	371	37.4%	101	37.5%
T3	464	36.8%	360	36.3%	104	38.7%
T4	180	14.3%	135	13.6%	45	16.7%
pN Stage							0.493
N0	117	9.2%	97	9.7%	20	7.4%
N1	1090	86.0%	854	85.6%	236	87.4%
N2	61	4.8%	47	4.7%	14	5.2%
Regional Nodes Examined							0.58
Mean (Range)	20	(8–90)	20	(8–90)	20	(8–65)
Median (IQR)	17	(13–24)	17	(13–24)	16	(13–24)

Of the patients in the CgA cohort (n = 1,269) 1,010 were included in the survival analysis with others excluded due to lack of follow-up information. With a median follow-up for living patients of 44 months (interquartile range: 24–67 months), the actual median overall survival was not reached in either subgroup (Figure 2). However, patients in the CgA-High group had a statistically significantly shorter estimated mean overall survival compared to those in the CgA-Low group (6.6 years [95% CI 6.27 – 6.91] vs. 7.6 years [95% CI 7.35 – 7.61], log-rank p = 0.00001).

Figure 2.

Kaplan-Meier analysis of patients in CgA-High and CgA-Low subgroups (n = 1010).

The clinical variables significantly associated with overall survival in univariate analysis included age at diagnosis, Charlson-Deyo score ≥2, prior malignancy, primary tumor in ileum or overlapping lesions in small bowel, tumor size >1cm, preoperative serum CgA ≥139ng/mL, pT3 and pT4 tumors (Table 3). Multivariate Cox proportional hazards analysis for the CgA cohort is summarized in Table 3. Age at diagnosis (HR = 1.06, 95% CI 1.03 – 1.09; p < 0.001), Charlson-Deyo score ≥2 (HR = 3.93, 95% CI 1.71 – 9.01; p = 0.001), poorly differentiated tumors (HR = 11.22, 95% CI 4.16 – 30.24; p < 0.001) and preoperative serum CgA ≥139ng/mL (HR = 2.19, 95% CI 1.22 – 3.92; p = 0.009) were independently associated with shorter overall survival on multivariable Cox analysis. Of note, pathologic T-stage and N-stage were not independently associated with overall survival (p > 0.05). When a survival analysis was performed on all patients, without the preoperative CgA variable (Supplementary Table 1), pN1 was independently associated with an improved overall survival compared to pN0 (HR = 0.47, 95% CI 0.33 – 0.65, p < 0.001). Kaplan Meier analysis of all patients (n = 3,793) for pathologic N-stage demonstrated longer mean overall survival in patients with pN1 compared to pN0 (7.07 years vs. 6.86 years, p = 0.003) and pT1 compared to pT3 (7.14 years vs. 6.93 years, p = 0.01), but not pT2 (7.13 years, p = 0.23) or pT4 (7.03 years, p = 0.053) (Supplementary Figure 2).

Table 3.

Cox proportional hazards survival analysis of patients in the CgA cohort.

	HR	Univariate 95.0% CI		P	HR	Multivariable 95.0% CI		P
		Lower	Upper			Lower	Upper
Age at Diagnosis	1.09	1.06	1.11	<0.001	1.06	1.03	1.09	<0.001
Sex
Male	Ref
Female	0.83	0.50	1.37	0.457
Race
White	Ref
Non-white	1.70	0.88	3.27	0.112
Charlson-Deyo Score
0	Ref				Ref
1	1.47	0.79	2.76	0.227	1.45	0.68	3.07	0.335
≥2	4.35	2.11	8.99	<0.001	3.93	1.71	9.01	0.001
Number of Prior Malignancies
0	Ref				Ref
1	2.51	1.14	5.52	0.022	1.11	0.41	3.00	0.844
Primary Site
Jejunum	Ref				Ref
Ileum	0.44	0.17	1.13	0.089	0.55	0.15	1.98	0.362
Overlapping Lesions of SB	4.62	1.10	19.48	0.037	3.38	0.61	18.75	0.163
SB NOS	0.79	0.29	2.15	0.638	0.55	0.14	2.06	0.373
Surgical Procedure
Simple / Partial Resection	Ref
Total Resection	0.93	0.52	1.67	0.803
Debulking Surgery	3.33	0.46	24.31	0.236
Radical Resection	1.27	0.54	3.00	0.590
Tumor Size
≤1cm	Ref				Ref
>1cm	3.53	1.28	9.74	0.015	1.41	0.39	5.08	0.596
Differentiation
Well Differentiated	Ref				Ref
Moderately Differentiated	1.18	0.59	2.39	0.640	1.69	0.82	3.48	0.157
Poorly Differentiated	17.68	7.81	40.02	<0.001	11.22	4.16	30.24	<0.001
Preoperative Serum CgA
<139ng/mL	Ref				Ref
≥139ng/mL	3.09	1.87	5.12	<0.001	2.19	1.22	3.92	0.009
Mitotic Count
<2 / 10HPFs	Ref
2–20 / 10HPFs	1.42	0.62	3.25	0.409
>20 / 10HPFS	0.00	0.00	n/a	0.980
pT Stage
T1	Ref				Ref
T2	3.54	0.45	27.68	0.228	1.81	0.18	18.60	0.617
T3	7.60	1.04	55.47	0.046	2.91	0.27	30.96	0.376
T4	8.64	1.14	65.45	0.037	3.09	0.28	34.15	0.358
pN Stage
N0	Ref
N1	1.41	0.51	3.89	0.509
N2	0.82	0.15	4.45	0.814

Discussion

Our study demonstrates the prognostic utility of preoperative serum CgA in patients with locoregional SB-NETs who are surgical candidates. We highlighted that, unlike preoperative serum CgA, common pathological features such as T-stage and lymph node metastasis poorly predicted overall survival in patients with locoregional SB-NETs. Neither Cox proportional hazards nor Kaplan-Meier analyses demonstrated shorter overall survival in patients with lymph node metastasis (pN1–2) in patients with locoregional SB-NET. These findings call into question the prognostic relevance of lymph node metastases as defined by the current staging system and suggest that preoperative serum CgA may provide both a more accurate and accessible prognosticator for SB-NET.

Although the utility of CgA is well-established in the diagnosis and surveillance of patients with neuroendocrine tumors, its prognostic significance in the non-metastatic setting remains ill-defined.¹¹ The significant elevations in CgA levels have been associated with shorter overall survival in small pancreatic NETs while, conversely, it has been proposed that patients whose values are below this threshold may safely defer resection of small, well-differentiated tumors.^{13, 23} Our study demonstrates the independent association between the elevated preoperative CgA levels and shorter overall survival in patients with locoregional SB-NET which can be useful in preoperative counseling. Our findings may help identify a subset of patients with more aggressive tumor biology despite a relatively benign clinical presentation, such as those with possible subclinical or occult metastatic disease. Furthermore, using an accepted statistical method of dichotomization, we determined 139ng/mL to be the optimal cut-off above which patients in this cohort experienced the largest disparities in overall survival. Even when compared to potentially more accurate, ratio-based nodal staging systems, the advantages of preoperative serum CgA are 1) it informs a preoperative discussion of prognosis and 2) unlike histopathology, it is not subject to interpretation or adequacy of the surgical specimen. Additionally, our CgA cut-off exposed a difference in overall survival within 5 years of follow-up, delineating a subgroup of patients with unfavorable biology despite reported 5-year survival rates surpassing 94%.¹⁷ Thus, patients with locoregional SB-NET with an elevated preoperative CgA may benefit from a sensitive metastatic workup such as 68-Ga DOTATATE PET/CT (FDA-approved in 2016) or 18F-FDOPA PET/CT (currently investigational only), which were not used clinically in this study. Additional detection of metastatic sites can alter the extent of the operation and treatment strategy. The clinical utility of preoperative serum CgA levels, beyond preoperative discussion of prognosis, warrants further investigation. To date, ours is the largest study evaluating the prognostic implications of serum CgA in locoregional SB-NET and the only one to focus on its use in the preoperative setting. Others have examined the prognostic utility of CgA in SB-NET as part of a nomogram, thereby failing to examine the variable’s independent association with survival.^7,24 Furthermore, these studies utilized arbitrary cut-offs for CgA lacking clinical significance or statistical derivation. While our cut-off of 139ng/mL was identified from a proper statistical analysis, the CgA levels in this cohort may be from various assays. Thus, the cut-off of 139ng/mL may not be clinically applicable to other cohorts and requires prospective validation using a common assay. Our approach to dichotomization of the variable emphasizes the prognostic value of elevated preoperative CgA in patients with locoregional SB-NET.

While a regional lymphadenectomy that includes at least 8 lymph nodes represents the standard of care advocated by the National Comprehensive Cancer Network guidelines, the implications of lymph node metastases themselves among patients with SB-NET remain incompletely understood.^{21, 25} The first study to establish eight harvested nodes as the minimal number required for an adequate lymphadenectomy also attributed its associated survival benefit to the minimization of the lymph node ratio, and not the resection of nodal tissue itself.²⁵ This finding may suggest that the resection of nodal metastases likely does not improve long-term outcomes in these patients, but rather that these patients were spared from the inevitable understaging that would ensue if few or no lymph nodes were harvested during resection of the primary tumor. Similarly, a 2018 Surveillance, Epidemiology, and End Results (SEER) database analysis of patients with non-metastatic SB-NETs demonstrated the superior prognostic value of positive-lymph-node ratio to absolute nodal involvement and validated a novel lymph node staging system.¹⁷ Finally, in a 2019 study, patients with SB-NETs who received an adequate lymphadenectomy only exhibited shortened recurrence-free survival if ≥4 nodes were positive. Of note, recurrence-free survival was not associated with nodal involvement in patients who received sub-optimal lymphadenectomies. Adequate lymphadenectomy allows the more accurate prognostication as the extent of the tumor progression can be more precisely assessed.¹⁶ These data argue for further discrimination amongst, and a more objective assessment of, the overwhelming majority of SB-NETs staged as pN1. Our study, the largest appraisal of the current pathologic nodal staging system to date, utilizes a nationwide database to evaluate the importance of nodal involvement in SB-NETs. While our analysis found no evidence of association between pathologic node stage and overall survival, it is possible that our study was underpowered to assess this variable, given that only 13.1% of all patients had no lymph node metastases. Given the size of our study, however, it is more likely that this finding indicates a lack of granularity in the current staging system that may hinder its prognostic ability in patients with locoregional disease. This is supported by the fact that 83% of patients in our study had pN1 disease. Other well-known negative prognostic indicators, such as advanced age, Charlson-Deyo comorbidity score, and poorly-differentiated tumors, retained their established independence with overall survival, lending confidence to our model as an assessment tool. Furthermore, when chromogranin-A level was removed from the multivariable analysis to facilitate the assessment of pN in all patients, pN1 was paradoxically associated with improved overall survival; though this may be explained by more complete surgery and an upstaging due to a more extensive lymphadenectomy, it might also highlight a lack of power due to the small number of pN0 patients. This supplementary analysis also failed to demonstrate an independent association between pathologic T-stage and overall survival, despite finding such association with tumor size >1cm. While this study was not designed specifically to assess the accuracy to the existing T- and N-staging systems, these data further add to the uncertainty surrounding the current AJCC classifications and highlight the need for adjuvant metrics to guide discussions of prognosis with patients who present with locoregional SB-NETs, a role that we believe serum CgA levels can help to fulfill.

In addition to demonstrating the value of serum CgA levels, our study has a number of pertinent ancillary findings. Unsurprisingly, patients with poorly-differentiated tumors had a significantly shorter overall survival than their better-differentiated counterparts. Mitotic count, a variable used in contemporary staging systems of other neuroendocrine tumors, did not demonstrate an association with overall survival, which may be due to type II statistical error. These data, coupled with the unconventional relationships between T-, N-stage, and overall survival, reflect the unique biology underlying SB-NETs and indicate the need for a tailored staging system. Alternative staging systems, such as the 2010 WHO classification of pancreatic neuroendocrine tumors, prioritize measures of biologic aggressiveness over conventional appraisals, such as the ENETS and AJCC ‘TNM-based’ systems. Similarly, our findings demonstrate a stronger association of overall survival with histopathologic and biochemical markers of aggression, like poor differentiation and elevated CgA, thereby advocating for the prospective reassessment of prognostic variables in SB-NETs.

Despite our use of a large national database, our study is limited by its relatively small yield of patients with node-negative (pN0) or advanced nodal disease (pN2), and thus, may be underpowered to detect an overall survival difference. The retrospective nature of our analysis can be particularly problematic due to missing data, selection bias, non-standardized follow-up, absence of disease-specific survival, and the variability in laboratory assessment of serum CgA levels, which may contribute to differences in measured values. The NCDB also obscures information about non-neuroendocrine etiologies of CgA elevation, such as false elevations secondary to gastric acid suppression drugs, chronic atrophic gastritis, chronic kidney disease, systemic inflammatory diseases (i.e. rheumatoid arthritis, systemic lupus erythematosus), and cardiovascular diseases.²⁶ The lack of postoperative CgA in the NCDB limited our ability to assess the tumor-derived CgA and the effect of cytoreduction on CgA levels. Furthermore, because the current T-staging system is more discriminatory on the depth of invasion, our attempts to control for tumor burden by eliminating patients with metastatic disease leaves room for undetected variability in primary tumor size. Based on our findings, we support the call for a reassessment of the current pathologic nodal staging system in SB-NETs and suggest prospective evaluation of preoperative serum chromogranin-A, a standardized, non-invasive, and objective biochemical measure, as a superior prognosticator.

Conclusion

Elevated preoperative serum CgA is an adverse prognostic marker associated with shorter overall survival in patients with locoregional SB-NET. Thus, it can be useful in preoperative counseling. Neither lymph node metastasis nor T-stage was independently associated with shorter overall survival. We believe that our findings are cause for consideration of serum CgA as an established prognosticator of SB-NET, and indicate the need for dedicated prospective reassessment of SB-NET staging in a comprehensive, multicenter study to establish clinicopathologic variables with the strongest association with long-term survival.

Supplementary Material

1

Supplementary Figure 1. Graphical representation of Contal O’Quigley method of dichotomization.

2

Supplementary Figure 2. (A) Kaplan Meier analysis of all patients (n = 3793) based on pathologic nodal staging (AJCC 8^th Edition). P-values reflect log-rank comparison to pN0. (B) Kaplan Meier analysis of all patients (n = 3778) based on pathologic T-staging (AJCC 8^th Edition). P-values reflect log-rank comparison to pT1.

5

Supplementary Table 1. Cox proportional hazards survival analysis (excluding the CgA variable) of all patients (n = 3778).