Simplified 2-tier histologic grading system accurately predicts outcomes in goblet cell carcinoid of the appendix

Human Pathology (2015) 46, 1881–1889

www.elsevier.com/locate/humpath

Original contribution

Simplified 2-tier histologic grading system accurately predicts outcomes in goblet cell carcinoid of the appendix☆,☆☆ Lawrence H. Lee MD a,b , Yarrow J. McConnell MD b,c , Erica Tsang MD b,d , Siham Zerhouni MD b , Caroline Speers BA e , Hagen Kennecke MD b,d , David F. Schaeffer MD a,b,⁎ a

Division of Anatomic Pathology, Vancouver General Hospital, Vancouver, BC, Canada V5Z 1M9 The University of British Columbia, 317-2194 Health Sciences Mall, Vancouver, BC, Canada V6T 1Z3 c Surgical Oncology, Division of General Surgery, Vancouver General Hospital, Vancouver, BC, Canada V5Z 1M9 d Division of Medical Oncology, BC Cancer Agency, Vancouver, BC, Canada, V5Z 1L3 e Gastrointestinal Cancer Outcomes Unity, BC Cancer Agency, Vancouver, BC, Canada, V5Z 1L3 b

Received 19 June 2015; revised 31 July 2015; accepted 13 August 2015

Keywords: Goblet cell carcinoid; Adenocarcinoid; Adenoneuroendocrine; Crypt cell carcinoma; Histopathologic grading; Mixed adenoneuroendocrine carcinoma

Summary Goblet cell carcinoid (GCC) is a rare appendiceal malignancy with both neuroendocrine and glandular features. Clinical outcomes of patients with GCC vary widely and a histology-based 3-tiered prognostic scheme has been previously suggested; however, this scheme is subjective and challenging to apply in day-to-day practice. We sought to construct a simplified and prognostic grading system based on objective histologic features with specific criteria. A continuous population-based cohort of GCC with clinical outcome data and archival tissue available for review was extracted from regional databases. For the 78 patients with confirmed appendiceal GCC, specific histologic features, including cytologic atypia, peritumoral stromal desmoplasia, and solid growth pattern, were recorded, and a scoring system was devised, which separates patients with GCC into low-grade (n = 55; 71%) or highgrade (n = 23; 29%) histology. Correspondingly, clinical follow-up data show good prognosis in those with low-grade histology with median and 10-year overall survival of 51.0 months and 80.5%, respectively, whereas those with high-grade histology have a poor prognosis with median and 10-year overall survival of 16.5 months (P = .006) and 0% (P b .001), respectively. Multivariate Cox proportional hazard modeling demonstrates that this 2-tier histologic system remains predictive of overall survival when controlled for TNM clinicopathological stage. These data show that a simple and objective histologic scoring system separates GCC into low- and high-grade histology with divergent clinical outcomes. © 2015 Elsevier Inc. All rights reserved.

☆

Competing interests: None. Funding/Support: Funding for this work was provided in part by the BC Cancer Foundation (Vancouver, British Columbia, Canada). ⁎ Corresponding author at: Division of Anatomic Pathology, Vancouver General Hospital, 910 W. 10th Avenue, Vancouver, BC, Canada V5Z 1 M9. E-mail address: [email protected] (D. F. Schaeffer). ☆☆

http://dx.doi.org/10.1016/j.humpath.2015.08.005 0046-8177/© 2015 Elsevier Inc. All rights reserved.

1. Introduction Goblet cell carcinoids (GCCs) are rare malignancies that arise almost exclusively in the vermiform appendix and have both neuroendocrine and glandular histologic features [1-3].

1882 When initially described, they were thought to be slow growing and clinically indolent—akin to classic appendiceal carcinoid tumors [4]. However, adenocarcinomas have been reported to arise from GCC [5], and several multicenter cohorts have shown GCC to have a prognosis similar to that of appendiceal adenocarcinoma [1,6,7]. GCCs are, therefore, generally considered to behave more similarly to adenocarcinomas and are staged using the TNM system for appendiceal carcinoma [8]. Similar to adenocarcinoma, outcomes for patients with GCC cover a wide range and are closely related to TNM stage at time of diagnosis [9]. However, TNM staging does not fully predict outcome, and a histologic grading system that is reproducible and provides additional prognostic information would be very useful. Unlike typical neuroendocrine tumors, the Ki-67 proliferative index is not prognostic in GCC [10]. Moreover, the standard colonic adenocarcinoma grading system based on gland formation [11] is not readily applicable to GCCs given their inherently nonglandular architecture. To address this deficit, Tang et al [5,12] suggested classifying GCCs of the appendix into 3 histologic subtypes: typical GCC (group A); adenocarcinoma ex-GCC, signet ring cell type (group B); and adenocarcinoma ex-GCC, poorly differentiated carcinoma type (group C). However, the “Tang classification” poses several difficulties in dayto-day practice: First, in GCC, the clear delineation of signet ring cells from goblet cells can be challenging and is prone to significant interobserver variation [12]. Second, the classification relies on the holistic evaluation of several morphologic criteria. If some but not all the criteria are met, the correct classification is unclear. In addition, the classification scheme was based on a case series collected in a specialty referral center and may not represent the true spectrum of GCC. Finally, the 3 proposed histologic subgroups contained patients with markedly different TNM stage, and this factor was not accounted for in the survival analysis. In this study, we present a histopathologic analysis of a population-based cohort of patients with GCC, controlling for TNM stage, with full treatment and outcome data. We propose an alternative histologic grading system that stratifies GCC into low- and high-grade subtypes and predicts clinical outcome.

2. Materials and methods 2.1. Patient selection A continuous population-based cohort of patients with GCC of the vermiform appendix was identified from the British Columbia Cancer Agency's Gastrointestinal Cancers Outcome Unit (GICOU) database and the Vancouver Lower Mainland pathology archive. The GICOU database was searched for the years 1986 to 2012, based on the availability

L. H. Lee et al. of data. International Classification of Diseases for Oncology-3 (ICD-O-3) site identifier 18.1 with histology codes 82433, 82443, 82453, and 82463 was included. The Vancouver Lower Mainland pathology archive was searched for additional cases between 1984 and 2014 using the following search terms: “appendix,” “goblet cell carcinoid,” “crypt cell carcinoma,” “adenoneuroendocrine,” “signet ring cell,” and “neuroendocrine.” After accounting for duplicates, these searches identified 108 potential patients. The pathology reports for all cases were reviewed to exclude non-GCC appendiceal neoplasms and metastases from other sites. Clinical data, including patient demographics, treatment modalities, tumor recurrence, and overall survival, were obtained from the GICOU and through a detailed review of health records. Institutional ethics review board approval was obtained for the study.

2.2. Histologic assessment All microscopic slides were retrieved from all cases where a primary resection (either right hemicolectomy or appendectomy) was performed. The average number of slides available for review was 21 per case (range, 1-89). The cases underwent detailed histologic assessment by a pathology resident (L. H. L.) under the supervision of a gastrointestinal pathologist (D. F. S.). To establish a diagnosis of GCC, the tumor had to (1) arise within the wall of the vermiform appendix; (2) have at least 1 focus, greater than 1 mm2 in area, showing discohesive clusters of monotonous cells with a single prominent intracytoplasmic mucin vacuole that displaced the nucleus to the periphery of the cell (goblet or signet ring cell morphology); and (3) not be associated with dysplasia of the overlying appendiceal epithelium [5,13] (Fig. 1A-C). The presence or absence of extracellular mucin was not a diagnostic criterion. Each case was assigned a clinicopathological stage according to the 7th edition American Joint Committee on Cancer TNM staging system for appendiceal carcinoma [8]. Additionally, the hematoxylin and eosin (H&E) slides for each of the primary appendiceal tumor were reviewed and assessed for additional histologic features: (1) Cytologic atypia, defined as an area greater than 1 mm2 containing tumor cells with increased nuclear-to-cytoplasmic ratio, reduction in or loss of intracytoplasmic mucin and enlarged, hyperchromatic nuclei with irregular shape and nuclear contour (Fig. 1D-F). Four contiguous high-power fields (magnification ×400) with a 0.55-mm field diameter were used to assess a 1 mm2 area, and at least 1 cytologically atypical tumor cell was required to be in each high-power field. Cytologic atypia was distinguished from nuclear changes associated with acute appendicitis, by the lack of nuclear hyperchromasia and irregular nuclear contour, and the presence of vesicular nuclei and numerous infiltrating acute inflammatory cells (Fig. 2). (2) Peritumoral stromal desmoplasia, defined as dense fibrous connective tissue surrounding tumor cell clusters or individual tumor cells that

2-tier classification of goblet cell carcinoids

1883

Fig. 1 H&E photomicrographs of GCC exhibiting adverse histologic features. First row: Typical low-grade GCC with no adverse histologic features showing infiltration of appendiceal muscularis propria by tumor cells (arrow) with no disruption of appendiceal architecture (A, original magnification ×4) and uniform cells with goblet cell/signet ring cell morphology in well-formed individual clusters (B and C, ×40). Second row (D-F, ×40): GCC showing cytologic atypia with enlarged hyperchromatic nuclei with irregular nuclear shape and contour and variable loss of intracytoplasmic mucin (arrow). Third row (G, ×10; H and I, ×20): GCC showing prominent peritumoral stromal desmoplasia that replaces the normal smooth muscle of the appendiceal wall. Fourth row (J, ×10; K and L, ×20): GCC showing solid growth pattern with loss of distinct cell cluster architecture and tightly packed cells with minimal or no intervening stroma.

replaced the surrounding smooth muscle of the muscularis propria with distortion of the normal, circumferential, parallel, fascicular architecture (Fig. 1G-I). Desmoplasia of

the submucosa or subserosal fat or serosal adhesions are insufficient. (3) Solid growth pattern, defined as an area greater than 1 mm2 with closely packed tumor cells, minimal

1884

L. H. Lee et al.

Fig. 2 H&E photomicrographs of low-grade GCC involving an acutely inflamed appendix (×40). Tumor nuclei show reactive atypia with vesicular chromatin, visible nucleoli, smooth nuclear contours, reduced intracytoplasmic mucin, and lack of hyperchromasia (arrows). Note the numerous infiltrating neutrophils.

or no intervening stroma, and loss of distinct cell cluster architecture (Fig. 1J-L). Spatially separate small foci of solid growth pattern, which aggregate to a total area of 1 mm2, are insufficient. A histologic scoring system was created whereby 1 point was given for the presence of each of cytologic atypia, peritumoral stromal desmoplasia, and solid growth pattern (score ranges from 0 to 3), summarized in Table 1. The histologic scoring system was only applied to the primary tumor site (ie, appendix). Each case was also classified as either group A, B, or C GCC according to criteria published by Tang et al [5]. The histopathologic review and scoring by the 2 classification schemes was conducted in a blinded fashion. The reviewers (D. F. S. and L. H. L.) classified all cases, in random order, according to the Tang criteria without access to TNM or clinical data. The cases were then sorted

Table 1

into a new random order and reviewed after a temporal delay according to the proposed histologic grading criteria without access to TNM, clinical, or Tang classification data.

2.3. Statistical analysis Clinical and pathologic characteristics were compared across classification groups using χ2 testing for nominal variables and Wilcoxon rank sum testing for continuous variables. Overall survival (OS) was examined using Kaplan-Meier analysis and log-rank testing. Multivariate analysis was conducted using Cox proportional hazard modeling. Receiver operating characteristic curve analysis was used to compare the predictive ability of survival models. All statistics were conducted using Stata/IC 13.1 (StataCorp, College Station, TX).

Description of histologic scoring system for GCC

Feature Cytologic atypia

Description a

Stromal desmoplasia

Solid growth pattern

Total score

Scoring

At least 1 focus N1 mm in size High nuclear-to-cytoplasmic ratio with reduction in or loss of intracytoplasmic mucin Nuclei are enlarged and hyperchromatic with irregular nuclear shape and contours Dense fibrous connective tissue surrounding tumor cell clusters or individual tumor cells Replaces surrounding smooth muscle of the muscularis propria Results in distortion of the normal appendiceal architecture At least 1 focus N1 mm2 in size Loss of distinct cell cluster architecture Cells tightly packed together with no or minimal intervening stroma Sum of above points 2

0: Absent 1: Present 0: Absent 1: Present 0: Absent 1: Present /3 Low grade: 0-1/3 High grade: 2-3/3

Abbreviation: GCC, goblet cell carcinoid. a This is distinguished from tumor cells with nuclear changes associated with acute appendicitis (see “Histologic assessment” section).

2-tier classification of goblet cell carcinoids Table 2

1885

Clinical and demographic characteristics of patients with goblet cell carcinoid included in the study cohort

n Age, y (median) Sex, male (%) Appendicitis, % LVI present (%) PNI present (%) Mitoses/10 HPF (median) T stage 2 3 4 N stage 0 1 2 M stage 0 1 Histologic score ≤1 ≥2

n (%)

Tang classification

Histologic score

A

B

C

78 53 39 (50%) 47 (60%) 17 (22%) 60 (77%) 1 2 (2%) 42 (54%) 34 (44%) 58 (77%) 10 (13%) 7 (9%) 62 (79%) 16 (20%) 55 23

48 53 52% 75% 15% 67% 1 4% 63% 33% 92% 8% 0 96% 4% 98% 2%

20 52 55% 55% 20% 90% 1 0 55% 45% 50% 28% 22% 70% 30% 40% 60%

10 58 30% 0 60% 100% 7 0 10% 90% 55% 11% 33% 20% 80% 0 100%

P .7 .4 b.001 .007 .02 b.001 .02

b.001 b.001 b.001

≤1

≥2

55 53 56% 87% 11% 69% 1 4% 67% 29% 89% 9% 2% 98% 2% – –

23 56 35% 13% 48% 96% 3 0 22% 78% 45% 25% 30% 35% 65% – –

P .5 .1 b.001 b.001 .01 b.001 b.001 b.001 b.001 –

NOTE. Only 75 patients had lymph nodes sampled; Wilcoxon rank sum test for continuous variables; χ2 test for nominal variables. Abbreviations: LVI, lymphovascular invasion; PNI, perineural invasion; HPF, high-power field.

3. Results From the identified cohort of 108 potential patients, 20 were excluded on review of pathology reports (13 clearly had a non-GCC malignancy, and 1 had no surgery performed), and 1 had no pathology report available. Archival specimens were sought for the remaining 87 potential GCC patients. Archival material was either lost or destroyed in 2 cases, a non-GCC primary was found on histopathologic review in a further 6 cases, and 1 case had no clinical outcome data available. A diagnosis of primary appendiceal GCC was confirmed in the remaining 78 patients, all with matched clinical data. Median follow-up was 38.4 months, median age was 53 years, 39 cases (50%) occurred in female patients, and 47 patients (60%) presented with appendicitis. The T stage of most tumors were either T3 (54%) or T4 (44%). Of the 75 patients who had lymph nodes resected, 17 (22%) were positive for tumor deposits. Of the patients presenting with metastatic disease (n = 16), all had peritoneal deposits. All patients underwent surgery and/or chemotherapy at the discretion of the treating physicians.

3.1. Histopathologic classification With application of the Tang histologic criteria, 48 patients (62%) had typical GCC (group A); 20 patients

Fig. 3

(26%) had adenocarcinoma ex-GCC, signet ring cell type (group B); and 10 patient (13%) had adenocarcinoma ex-GCC, poorly differentiated carcinoma type (group C). Patients in group B or C were less likely to present with appendicitis (P b .001) and were more likely to have perineural invasion (P = .02) and to have positive nodes (P b .001), compared to patients in group A. Patients in group C were more likely to have lymphovascular invasion (P = .007), to have a high mitotic rate (median, 7 versus 1; P b .001), and to present with T4 primary and peritoneal metastases (P = .02 and P b .001, respectively) compared to group A and B patients. See Table 2 for detailed results. Applying the proposed GCC histologic scoring system, cytologic atypia was identified in 31 cases (40%); stromal desmoplasia, in 23 cases (29%); and solid growth pattern, in 15 cases (19%). When desmoplasia was present, it was always present in the muscularis propria with or without extension into the subserosal fat, mucosa, or serosa. In 45 cases (49%), none of the targeted histologic features were present, and a score of zero was assigned. A single feature (either cytologic atypia or stromal desmoplasia) was identified in 10 cases (13%); solid growth pattern was never the sole finding (Fig. 3). In the remaining 23 cases, 10 (13%) had 2 features, whereas 13 cases (17%) displayed all 3 features. Based on Kaplan-Meier analysis (see below), patients were divided into 2 groups (low grade with

Histogram of histologic features present in each case of GCC examined; each column represents 1 case.

1886

L. H. Lee et al.

histologic score 0 or 1; high grade with histologic score 2 or 3). Patients with low-grade histology were more likely to present with appendicitis (P b .001). Patients with high-grade histology were more likely to have lymphovascular invasion (P b .001), perineural invasion (P = .01), and higher mitotic rate (median, 3 versus 1; P b .001). High-grade histology was also positively associated with higher T stage (P b .001), N stage (P b .001), and M stage (P b .001). See Table 2 for detailed results.

3.2. Survival analysis On univariate analysis, a clinical presentation with appendicitis was associated with better median (51.0 versus 17.8 months; P = .04) and 10-year OS (23.0% versus 79.9%; P b .001). The presence of lymphovascular invasion, perineural invasion, a higher mitotic rate, higher T stage, higher N stage, and higher M stage were all associated with worse 10-year OS on univariate analysis (see Table 3). On multivariate analysis of these clinical and standard pathologic variables, only M stage remained as an independent predictor of survival (hazard ratio [HR], 9.5; P = .002; 95% confidence interval, 2.3-39.7).

Table 3

The Tang classification scheme and our proposed histologic grading system were both associated with a gradient in overall survival (Table 3 and Fig. 4). Tang group A patients had a median OS of 60.4 months, compared to 27.5 months in group B (P = .03). Group C patients had a median OS of 9.4 months, significantly lower than that of Group B (P = .002). Estimated 10-year OS followed a similar pattern: group A, 79.3%; group B, 20.2%; and group C, 0% (P b .001 for all comparisons). On multivariate analysis, Tang classification and M stage remained as independent predictors of OS (Table 4). A model constructed using these results gave an area under the receiver operating characteristic curve (AUROC) value of 0.908. Similarly for the proposed histologic grading system, each additional increase in score was associated with worse survival (Fig. 4B). The most marked difference was between those with a score less than or equal to 1 and those with score greater than or equal to 2. Patients were, therefore, classified into 2 groups for further analysis: low-grade histology (score 0 or 1) versus high-grade histology (score 2 or 3). Median OS was 51.0 versus 16.5 months in patients with low- versus high-grade histology (P = .006), whereas 10-year OS was 80.5% versus 0% (P b .001). On multivariate analysis, histologic score, M stage, and high mitotic rate in the primary tumor all remained as independent predictors of survival, but histologic score was the strongest factor (HR, 15.0;

Overall survival of patients with goblet cell carcinoid

Age, y Sex Appendicitis LVI present PNI present Mitoses/10 HPF T stage

N stage

M stage Tang classification

Histologic score

≤50 N50 Male Female Yes No Yes No Yes No ≥2 ≤1 2 3 4 0 1 2 0 1 A B C ≤1 ≥2

n (%)

Median OS (mo)

P

10-y OS (%)

27 (36%) 49 (64%) 39 (50%) 39 (50%) 47 (60%) 31 (40%) 17 (22%) 61 (78%) 60 (77%) 18 (23%) 27 (35%) 51 (65%) 2 (2%) 42 (54%) 34 (44%) 58 (77%) 10 (33%) 7 (10%) 62 (79%) 16 (21%) 48 (62%) 20 (26%) 10 (12%) 55 (71%) 23 (29%)

39.1 36.8 64.3 26.2 51.0 17.8 23.9 38.3 32.5 52.7 24.8 38.4 36.5 38.4 33.3 47.2 21.2 16.5 50.0 15.9 60.4 27.5 9.4 51.0 16.5

.47

49.3% 58.1% 69.6% 41.1% 79.9% 23.0% 25.5% 66.9% 41.2% 100% 38.5% 67.2% 100% 64.4% 45.6% 71.7% 66.7% 0% 78.5% 0% 79.3% 20.2% 0% 80.5% 0%

.01 .04 .41 .28 .23 .90

.17

.005 .03 a .002 b .006

P .97 .14 b.001 b.001 .006 .001 .02

b.001 b.001 b.001 a b.001 b b.001

NOTE. Only 75 patients had lymph nodes sampled; Wilcoxon rank sum test for medians; log-rank test for Kaplan-Meier estimated survival functions. Abbreviations: OS, overall survival; LVI, lymphovascular invasion; PNI, perineural invasion; HPF, high-power field. a Group A compared to group B. b Group B compared to group C.

2-tier classification of goblet cell carcinoids

1887 findings. Nearly all (98%) of Tang group A tumors were classified as histologically low grade. All (100%) of Tang group C tumors were classified as histologically high grade (Table 2). However, Tang group B tumors were divided between low-grade (40%) and high-grade (60%) histology. Although the median OS (32.6 versus 25.2 months; P = .3) was no different between low- and high-grade classified patients within Tang group B, the 5-year estimated OS was substantively different and approached significance in this small subset analysis (100% versus 48.5%; P = .09). Within the subgroup of patients presenting without metastatic disease (n = 62), a minority (8 patients; 13%) had high-grade histology. However, this clinically important subgroup had a worse 5-year OS (87.5% versus 98.1%) compared to those with low-grade histology. The analysis of Tang classification within nonmetastatic patients was less useful. It demonstrated no difference between group A and B patients, whereas there were only 2 group C patient who presented with nonmetastatic disease and both died within 10 months. Among patients with metastatic disease at presentation, all but 1 had high-grade histology, and all but 2 had group B or group C Tang classification. The 1 low-grade/group A patient had an OS of 111 months, whereas the high-grade/ group A patient had an OS of 45 months—but both have died as a result of GCC.

4. Discussion

Fig. 4 Kaplan-Meier overall survival curves of patients with GCC as a function of Tang classification system (A), histologic score (B), and low-grade (score ≤1) versus high-grade (score ≥2) histology (C).

P = .002). This model gave an AUROC of 0.906 for predicting OS—not significantly different from the predictive ability of the Tang-based model (P = .9).

3.3. Comparing subgroups When case allocation was compared between the 2 classification schemes, there was general alignment of the

We sought to derive a histologic grading system for goblet cell carcinoids of the appendix that not only predicts clinical outcomes independent of clinicopathological stage but also is objective and easily applicable by nonsubspecialized pathologists in day-to-day practice. In our cohort, GCCs of the vermiform appendix most commonly presented with symptoms related to acute appendicitis. In such patients, the disease was typically localized to the appendix without metastatic disease, which parallels other large epidemiologic studies of GCC [7]. This often presents a challenge for surgical pathologists where the acute inflammation obscures the low-grade, infiltrative, neoplastic process. Additionally, GCCs are often an incidental finding that initially lacks a complete staging procedure, and the role and type of additional therapy remain unclear [14]. Some patients experience extended disease-free periods, whereas others rapidly progress. This wide variation in clinical behavior has prompted the need for a histopathologic scheme to predict outcomes and guide therapy. As mentioned, a study by Tang et al [5] proposed a 3-tier histologic grading system that stratifies patients into 3 prognostic groups. Notably, the application of the Tang criteria to our cohort has successfully stratified patients with GCC into different prognostic categories, therefore validating the Tang system. However, the histologic criteria for distinguishing between typical GCC (group A) and signet

1888 Table 4

L. H. Lee et al. Multivariate Cox proportional hazard modeling for overall survival

Covariate Mitoses (/10 HPF) M stage Tang classification

AUROC

HR ≤1 ≥2 M0 M1 A B C 0.908

1 1.84 1 15.2 1 9.87 16.15

P .26 b.001

Covariate Mitoses (/10 HPF) M stage Histologic score

.002 .002 AUROC

≤1 ≥2 M0 M1 ≤1 ≥2

HR

P

1 2.27 1 9.45 1 15.02

.09 .001

.002

0.906

Abbreviations: AUROC, area under receiver operator curve; HR, hazard ratio.

ring cell adenocarcinoma ex-GCC (group B) can be subjective and challenging to apply [12]. This is especially true in cases where transmural inflammation of acute appendicitis results in distortion of the appendiceal wall and nuclear changes, which can be problematic. Proliferative index as measured by Ki-67 labeling is reported to be closely associated with Tang histologic classification [5]; however, a recent attempt to use Ki-67 labeling to predict outcomes was unsuccessful [10]. Unfortunately, Ki-67 values were only available for 19 (24.3%) patients in this study, making analysis of this factor impossible. A more recent study attempted to stratify the prognosis of patients with GCC by quantifying the proportion of coexisting adenocarcinoma, if present, and classifying the tumors into 3 groups based on the amount of adenocarcinoma present [15]. However, the authors do not directly address the inherent subjectivity of differentiating “adenocarcinoma” from typical “goblet cell carcinoid tumor.” Moreover, using multivariate analysis, there was no statistically significant difference in OS between the second and third groups, further supporting a 2-tier grading system. Given the similarities between the proposed grading system and the Tang classification system, there was, not surprisingly, substantial similarity in their allocation of patients and resulting survival estimates. The notable finding is that a clear distinction in prognosis was achieved without reliance on a distinction between goblet and signet ring cells. In addition, the use of a histologic scoring system is familiar to pathologists and reliably integrated into synoptic reporting initiatives. Although identifying cytologic atypia can be challenging, we outlined a minimum size requirement, specific nuclear features, and how to distinguish it from nuclear changes related to acute appendicitis. Although every effort was made to minimize the subjectivity of identifying cytologic atypia, a certain degree of subjectivity is unavoidable. The minimum proportion of cytologically atypical tumor cells required within the 1 mm2 area has not been quantified, although a minimum of 1 per high-power field is used as a criterion in this study for ease of applicability. Moreover, the qualitative measures of nuclear hyperchromasia or nuclear contour irregularity are inherently subject to a degree of interobserver variability. Nonetheless, the use of cytologic atypia in histologic grading schemes is an accepted practice and is

used widely in other areas of anatomical pathology including breast carcinomas [16], renal cell carcinomas [17], salivary gland neoplasms [18], and mesotheliomas [19] with clinically applicable prognostic significance. Its significance in the histologic grading of GCC was highlighted in the Tang classification and is reinforced here. To the best of our knowledge, this study reports the largest cohort of patients with GCC and detailed histopathologic review. That said, given the rarity of this disease, the numbers remain small and multicenter cohorts are needed to further expand our understanding. Additionally, our study draws attention to the importance of metastatic disease at presentation as a strong predictor of OS on univariate and multivariate analyses. Combining histologic grade with metastatic disease status allows for the identification of a subgroup of patients with excellent prognosis who may benefit from less aggressive treatment regimens. The interobserver reliability of the Tang histology classification has been questioned and hence is a possible limitation of the current study as well. All specimens were reviewed by 2 pathologists, and disagreements were settled by consensus after review of multiple slides. Every effort was made to classify specimens according to the published criteria, and pathologists did not have access to outcome data while conducting the histology review. Further formal blinded analyses of the comparative reliability of the 2 classification schemes are warranted.

5. Conclusion The clinical spectrum of goblet cell carcinoid of the vermiform appendix varies widely, and a reproducible and accurate means of predicting outcomes would be of great clinical utility. We present an objective, readily applicable histologic grading system that predicts OS independent of TNM staging. The proposed histologic grading system separates GCC into low- versus high-grade histology using 3 histologic criteria that can be assessed without use of special tests and does not rely on the distinction between goblet and signet ring cells. These characteristics make it superior to the previously proposed classification by Tang et

2-tier classification of goblet cell carcinoids al, which has not been demonstrated to predict survival independent of TNM stage and depends on the identification of signet ring cells as a cardinal distinguishing feature.

References [1] Butler JA, Houshiar A, Lin F, Wilson SE. Goblet cell carcinoid of the appendix. Am J Surg 1994;168:685-7. [2] Holt N, Gronbaek H. Goblet cell carcinoids of the appendix. TheScientificWorldJournal 2013;2013:543696. [3] Roy P, Chetty R. Goblet cell carcinoid tumors of the appendix: an overview. World J Gastrointest Oncol 2010;2:251-8. [4] Subbuswamy SG, Gibbs NM, Ross CF, Morson BC. Goblet cell carcinoid of the appendix. Cancer 1974;34:338-44. [5] Tang LH, Shia J, Soslow RA, et al. Pathologic classification and clinical behavior of the spectrum of goblet cell carcinoid tumors of the appendix. Am J Surg Pathol 2008;32:1429-43. [6] McConnell YJ, Mack LA, Gui X, et al. Cytoreductive surgery with hyperthermic intraperitoneal chemotherapy: an emerging treatment option for advanced goblet cell tumors of the appendix. Ann Surg Oncol 2014;21:1975-82. [7] McCusker ME, Cote TR, Clegg LX, Sobin LH. Primary malignant neoplasms of the appendix: a population-based study from the surveillance, epidemiology and end-results program, 1973-1998. Cancer 2002;94:3307-12. [8] Edge S, Byrd D, Compton C, Fritz A, Greene F, Trotti A. AJCC cancer staging manual. New York, NY: Springer; 2010. [9] Turaga KK, Pappas SG, Gamblin T. Importance of histologic subtype in the staging of appendiceal tumors. Ann Surg Oncol 2012;19:1379-85.

1889 [10] Liu E, Telem DA, Warner RR, Dikman A, Divino CM. The role of Ki-67 in predicting biological behavior of goblet cell carcinoid tumor in appendix. Am J Surg 2011;202:400-3. [11] Compton CC, Fielding LP, Burgart LJ, et al. Prognostic factors in colorectal cancer. College of American Pathologists Consensus Statement 1999. Arch Pathol Lab Med 2000;124:979-94. [12] Wang HL, Dhall D. Goblet or signet ring cells: that is the question. Adv Anat Pathol 2009;16:247-54. [13] Bosman F, Carneiro F, Hruban R, Thiese N. WHO classification of tumours of the digestive system. Lyon, France: IARC Press; 2010. [14] Varisco B, McAlvin B, Dias J, Franga D. Adenocarcinoid of the appendix: is right hemicolectomy necessary? A meta-analysis of retrospective chart reviews. Am Surg 2004;70:593-9. [15] Taggart MW, Abraham SC, Overman MJ, Mansfield PF, Rashid A. Goblet cell carcinoid tumor, mixed goblet cell carcinoid-adenocarcinoma, and adenocarcinoma of the appendix: comparison of clinicopathologic features and prognosis. Arch Pathol Lab Med 2015;139: 782-90. [16] Elston CW, Ellis IO. Pathological prognostic factors in breast cancer. I. The value of histological grade in breast cancer: experience from a large study with long-term follow-up. Histopathology 1991;19: 403-10. [17] Fuhrman SA, Lasky LC, Limas C. Prognostic significance of morphologic parameters in renal cell carcinoma. Am J Surg Pathol 1982;6:655-63. [18] Auclair PL, Goode RK, Ellis GL. Mucoepidermoid carcinoma of intraoral salivary glands. Evaluation and application of grading criteria in 143 cases. Cancer 1992;69:2021-30. [19] Kadota K, Suzuki K, Colovos C, et al. A nuclear grading system is a strong predictor of survival in epitheloid diffuse malignant pleural mesothelioma. Mod Pathol 2012;25:260-71.