Evaluation of association between center colorectal neuroendocrine neoplasm volume and survival among patients with colorectal neuroendocrine carcinoma

Although correlation between center volume and survival has been reported for several complex cancers, it remains unknown if this is true for colorectal neuroendocrine carcinomas (CRNECs). We hypothesized that higher center annual volume of colorectal neuroendocrine neoplasm resections would be associated with overall survival (OS) for patients with CRNECs.


| INTRODUCTION
Colorectal neuroendocrine neoplasms consist of neuroendocrine tumors (NETs), neuroendocrine carcinomas (NECs), and mixed neuroendocrine-non-neuroendocrine neoplasms (MiNENs).[7] Factors such as surgical resection and use of adjuvant chemotherapy have been associated with improved survival in CRNEC; however, there remains an urgent need to identify additional avenues to improve patient survival given their poor prognosis. 3,8,9Center resection volume has gained intense interest as a proxy for center experience and likelihood of providing guideline-concordant care for a given disease. 10,113][14][15][16][17][18][19][20][21] The impact of center resection volume on survival for CRNEC patients has yet to be explored.
In this study, we investigated the association between center colorectal neuroendocrine neoplasm resection volume and overall survival (OS) among patients with CRNEC.We hypothesized that the higher center mean annual volume of colorectal neuroendocrine neoplasms resection would be associated with better OS for patients with CRNECs.Patients with multiple primary tumors, unknown receipt of adjuvant chemotherapy, or who received radiation or hormonal therapy were excluded (Supporting Information S1: Figure 1).
Colorectal neuroendocrine neoplasm resection volume was calculated by summing all resected cases for each center between 2006 and 2018 that corresponded with neuroendocrine neoplasm histologic codes based on previously published literature, that is, 8240, 8241, 8242, 8243, 8244, 8245, 8249, as well as the codes for NEC. 3,23Average number of cases per year was calculated by dividing the total by the interval number of years.

| Statistical analyses
The relationship between the mean annual colorectal neuroendocrine neoplasm resection volume for each center and hazard ratio (HR) was determined using restricted cubic splines in a random effects univariable Cox model. 24Restricted cubic splines is a modeling approach that generates a smooth curve to explain the complex relationship between a continuous variable and an outcome using a piecewise cubic polynomial function over several intervals within the data set. 25alysis variables were selected a priori based on previous literature, clinical knowledge, and availability.Comparisons between variables were performed using the χ 2 test (categorical variables), Student's t test (continuous variables), or Kruskal-Wallis test (continuous variables).Unadjusted survival was compared by the Kaplan-Meier method and log-rank tests.Adjusted survival analysis was performed using Cox regression.Bidirectional selection considered age, sex, facility type, race, insurance status, income median, tumor site, tumor size, receipt of chemotherapy status, surgical margin status, Charlson-Deyo Index, Tumor-Nodes-Metastasis stage, hospital length of stay (LOS), distance from a hospital, number of positive regional nodes, and readmission within 30 days for inclusion in multivariable models.Age, facility type, and hospital distance were not selected in the original model but were added to robustly control for confounders.Covariate estimates are reported as HRs with 95% confidence intervals (CIs).All tests were two-sided and assessed for significance at p < 0.05 using RStudio (R 4.3.2).This study was deemed exempt from review by the Institutional Review Board.

| Overall patient cohort and center volume stratification
There were 694 patients with a diagnosis of NEC.The median age was 69 (interquartile range [IQR]: 59-79) years, with the majority having right-sided (74%) and stage III disease (73%).Only 50% received adjuvant chemotherapy.Using restricted cubic splines, we determined that the threshold at which colorectal neuroendocrine neoplasm resection volume was associated with higher mortality hazard was ≤0.37 cases/year (Figure 1).The centers below this threshold (511 centers) were classified as low-volume (LV), while those above this threshold (718 centers) were classified as highvolume (HV).Resection at an academic center was the only factor significantly predictive of receiving treatment at an HV center (5.42   [95% CI: 2.07-14.2],p = 0.001) in the multivariable model (Supporting Information S1: Table 1).

| CRNEC patients' characteristics based on treatment at HV or LV centers
A similar proportion of patients with comorbidities and stage III disease were treated at LV and HV centers (Table 1).Age distribution and distance traveled were also similar for both groups.However, compared to patients treated at HV centers, a significantly greater proportion of patients at LV centers were treated at a nonacademic center (62% LV vs. 49% HV, p < 0.001), underwent an open procedure (49% LV vs. 37% HV, p < 0.001), and had a positive margin (20% LV vs. 12% HV, p = 0.02).Short-term outcomes including median LOS, number of nodes positive, 30-day readmission, and 30-day mortality did not differ significantly between both groups.

| OS outcomes based on treatment at HV or LV centers
In the unadjusted survival analysis, patients treated at LV centers had significantly lower OS (Figure 1) than those treated at HV centers (p < 0.001).The 1-year, 3-year, and 5-year OS for patients treated at LV centers were 54%, 34%, and 25%, respectively, while for HV centers they were 70%, 50%, and 43%, respectively (Figure 2).Additionally, the median OS among patients treated at LV centers was 14 months (95% CI: 10-19 months), while for those treated at HV centers it was 33 months (95% CI: 25-49 months).In multivariable analyses (Table 2)

| DISCUSSION
Using a large national database, we found that CRNEC patients resected at an HV center had better long-term OS than those resected at an LV center, although there were no significant differences in short-term outcomes such as LOS, 30-day readmission, and 30-day mortality.Nonetheless, a significantly greater proportion of patients resected at HV centers underwent a minimally invasive procedure and had negative margins.Only resection at an academic center was associated with resection at an HV center.
Few extant studies have evaluated the impact of center volume outcomes for patients with gastrointestinal neuroendocrine neoplasms. 20,26,27Based on a thorough literature review, our study is the first to specifically evaluate the association between center colorectal neuroendocrine neoplasm resection volume and OS among patients with CRNEC.Nevertheless, our findings are generally consistent with the few previous studies.
F I G U R E 1 Cubic spline demonstrating relationship between center annual volume for colorectal neuroendocrine neoplasm and hazard for patients with colorectal neuroendocrine carcinoma.The inflection point at which hazard was significantly diminished was detected at >0.37 cases/year.CI, confidence interval.

Baeg et al. used the Surveillance, Epidemiology, and End Results
(SEER) database to examine the impact of center volume on outcomes for patients with gastroenteropancreatic neuroendocrine tumors (GEP-NETs) diagnosed between 1995 and 2010. 26This heterogeneous cohort comprised of 899 patients with stages I-IV disease of the stomach, pancreas, small bowel, and colorectum.
Volume was arbitrarily defined to evenly distribute patients between three groups, that is, low, middle, and high.They found that patients T A B L E 1 Clinicodemographic characteristics of patients with colorectal NEC treated at low-and high-volume colorectal neoplasm resection centers.treated at HV centers had significantly better disease-specific survival than those treated at LV centers after adjusting for confounders.From multiple sensitivity analyses, they found the threshold for worse outcomes to be fewer than two patients with GEP-NET treated every year.Based on these findings, they additionally argued that regionalization of GEP-NET care may lead to improved outcomes for patients.
Recently, Patel et al. used a statewide registry to specifically assess the relationship between hospital pancreatic NET(PNET) volume and survival among patients with PNET.Volume was also arbitrarily defined in their study.In their unadjusted analysis, they found that patients treated at HV centers had improved survival regardless of whether their disease was locoregional or metastatic. 27However, after controlling for confounders they found that treatment at a HV center was associated with decreased risk of mortality only among those with metastatic disease but not those with locoregional disease.They opined that their findings reflected the need for greater expertise in the management of metastatic disease, which is more complex.In our current study, we have focused specifically on CRNEC, a subgroup increasing in incidence and characterized by one of the poorest prognoses among gastrointestinal neuroendocrine neoplasms. 5,28,29 note, the National Comprehensive Cancer Network (NCCN) recommends adjuvant chemotherapy for patients with resectable extrapulmonary NEC. 30 We have also highlighted in a recent study that adjuvant chemotherapy was associated with improved survival among patients with CRNEC. 9However, in the current study, the proportion of patients receiving adjuvant chemotherapy was similarly low in HV and LV centers.Thus, this may represent an avenue for improvement across all centers.Chemotherapy regimens recommended by the NCCN include cisplatin and etoposide, carboplatin and etoposide, FOLFOX (folinic acid, fluorouracil, and oxaliplatin), FOLFIRI (folinic acid, fluorouracil, and irinotecan), or temozolomide and capecitabine. 29,31major strength of our study is that the volume-outcome relationship was evaluated using a restricted cubic spline to determine the threshold of volume-outcome relationship rather than an arbitrary definition of volume, which can bias conclusions. 24wever, our findings should be considered in the context of a few limitations.Briefly, our study is retrospective and subject to potential selection bias.Additionally, Ki-67 index and/or mitotic count were not routinely collected in the NCDB during the study period to allow ascertainment of NEC diagnosis.We did use previously validated histologic codes and tumor grades to identify the NEC cohort.Also, the relationship between center volume and patient outcomes is complicated and may be influenced by several other factors that are unavailable in the NCDB such as surgeon volume, surgeon experience, operative complexity, and so on.These limitations notwithstanding, in previous studies for several complex cancer diagnoses, center volume has been associated with improved survival outcomes, presumably by virtue of greater access to resources such as multidisciplinary teams and technical expertise. 10,32 summary, similar to those with other complex cancer diagnoses, patients with CRNECs may benefit from resection at HV centers given the association with improved survival after adjustment for confounders.These findings may have implications in referral patterns for patients with CRNECs.

2 | METHODS 2 . 1 |
Data sources and study subjects The National Cancer Database (NCDB) is a joint program of the Commission on Cancer of the American College of Surgeons and the American Cancer Society, and it collects approximately 70% of cancer diagnosis in the United States.Patients in the NCDB with surgically resected stages I-III CRNEC (appendix excluded given that neuroendocrine neoplasms of the appendix are considered to behave differently from other sites of the colorectum) 22 from 2006 through 2018 were identified based on ICD-O-3 codes for NEC including 8246, 8013, and 8041.Only those also coded to have poorly differentiated tumors were included.Patients had to have undergone surgical resection including partial colectomy/segmental resection, subtotal colectomy/hemicolectomy, total colectomy, or proctectomy.

F
I G U R E 2 Kaplan-Meier curve of 5-year overall survival (OS) of patients with CRNEC treated at LV and HV colorectal neuroendocrine neoplasm resection centers.CI, confidence interval; CRNEC, colorectal neuroendocrine carcinoma; HV, high volume; LV, low volume.T A B L E 2 Multivariable analysis of center colorectal neuroendocrine neoplasm resection volume and association with survival in patients with colorectal NEC.