Overinterpretation is common in pathological diagnosis of appendix cancer during patient referral for oncologic care

Mucinous appendiceal neoplasms that may give rise to pseudomyxoma peritonei (PMP) remain a challenging tumor type for pathological classification despite their description dating back to the 19 th century when the term PMP was first used by Werth.[ 1 ] Mucinous appendiceal neoplasms comprise a spectrum from low-grade appendiceal neoplasm (LAMN) to various grades of invasive adenocarcinoma.[ 2 ] PMP is clinically defined as mucinous or gelatinous ascites and peritoneal implants associated with the pathological finding of neoplastic mucin-producing cells in the peritoneal cavity.[ 2 , 3 ] It is now understood that the vast majority of PMP arise from perforated mucinous neoplasms of the appendix, and represent their peritoneal extension, with a minority of cases being due to mucinous adenocarcinomas arising from other sites.[ 4 ] Ovarian origin of PMP is rare and generally associated with mucinous adenocarcinoma arising from mature cystic teratoma.[ 5 , 6 ] Most cases of PMP have low-grade histologic features (sometimes termed “low-grade mucinous carcinoma peritonei” according to the World Health Organization (WHO)), while close to one quarter of cases have high-grade histologic features (sometimes termed “high-grade mucinous carcinoma peritonei” according to the WHO) with a corresponding worse prognosis.[ 4 ] In addition, the presence of signet ring cells has emerged as an independent negative prognostic factor, especially if tissue invasion is seen.[ 7 , 8 ] Although appendiceal neoplasms are relatively uncommon and encountered in less than 1–2% of incidental appendectomies, mucinous neoplasms and well-differentiated neuroendocrine tumors, also known as “carcinoids”, comprise the majority. [ 9 , 10 ] In contrast to neoplasia, by far the most common appendiceal diagnosis in surgical pathology is acute appendicitis.[ 11 , 12 ] Several classification schemes have been published to help determine risk for recurrent disease in LAMN.[ 3 , 13 – 17 ] Studies have demonstrated that patients with LAMNs that are confined to the appendix remain disease free upon follow up.[ 14 , 15 , 18 , 19 ] Additionally, the presence of acellular mucin or neoplastic epithelium at the appendectomy margin of LAMNs confined to the appendix is not a predictor of peritoneal recurrence.[ 15 , 19 ] Moreover, a clear link has been established between the presence of extra-appendiceal neoplastic epithelium and risk for recurrent disease, wherein LAMNs lacking extra-appendiceal neoplastic epithelium are considered to be at low risk of peritoneal recurrence and LAMNs with extra-appendiceal neoplastic epithelium are at high risk of recurrence.[ 14 , 15 , 18 , 20 ] Clinical management of LAMN with and without extra-appendiceal neoplastic epithelium is controversial.[ 17 , 21 ] A diagnosis of LAMN with low risk of recurrence can thus be used to justify conservative clinical follow-up, while at some centers LAMN with high risk of recurrence has been used to justify the pursuit of more aggressive therapies such as right hemicolectomy, cytoreductive surgery (CRS), with or without hyperthermic intraperitoneal chemotherapy (HIPEC) upon identification of dissemination to the peritoneal surfaces.[ 18 ] Other institutions have reported successful long-term outcomes with close observation of select patients with LAMN with high risk of recurrence, including some patients with limited peritoneal dissemination.[ 22 ] Our institution has adopted an individualized approach that, in general, includes close observation of patients with LAMN with low or high risk of recurrence, in the absence of peritoneal dissemination. Because proper histologic classification can strongly influence both prognosis and treatment,[ 23 ] we performed a retrospective pathological chart review to measure the actual concordance rate between community pathologists and a single large academic medical center during patient referral.

161 cases were identified that included appendix specimens over the period of study. Forty-six of 161 cases (28.6%) contained appendiceal primary neoplasia or lesions. Of the 46 initial diagnostic case surgical specimens, 35 (76.1%) were appendectomies, 6 (13.0%) were right hemicolectomies, 2 (4.3%) were appendectomies with partial cecectomy, 1 (2.2%) was an ileocecectomy, 1 (2.2%) was an appendectomy with cecectomy and pelvis mass excision, and 1 (2.2%) was an appendectomy with hysterectomy, bilateral salpingo-oopherectomy, splenectomy, and omentectomy. The 115 excluded cases had a wide variety of diagnoses including incidental appendectomies, secondary involvement by other neoplasms including mesothelioma, leiomyosarcoma, serous ovarian carcinoma, and other diagnoses unrelated to primary appendiceal disease. No cases were excluded due to lack of availability of pathology records. Patients referred for primary appendiceal neoplasia ranged in age from 29 to 81 years old with an average of 55 years old (median 59 years old) and patients were more likely to be female (M:F ratio 1:1.56). The in-house slide review was generally requested by an attending surgical oncologist during patient referral for oncologic care. The originating pathologist practices were usually mid-sized (number of pathologists, average 10, median 7) and non-academic (91.3%) with moderate surgical case volumes (average 22,000 accessions per annum, median 18,000 accessions per annum). The primary signing pathologist was usually not subspecialized in gastrointestinal pathology (87%) and consulted another pathologist(s) in a large fraction of the cases (37%). The distribution of appendiceal diagnoses from the originating pathologist was 50% adenocarcinoma (23 of 46), 50% LAMN (23 of 46), and 0% mucocele (0 of 46, Fig 1 ). In contrast, the distribution of appendiceal diagnoses from the reference academic pathologist was 28.3% adenocarcinoma (13 0f 46), 63% LAMN (29 of 46), and 8.7% mucocele (4 of 46). Subclassification demonstrated that the adenocarcinomas were 30.8% (4 of 13) adenocarcinoma ex goblet cell carcinoid, 23.1% (3 of 13) adenocarcinomas arising from low-grade appendiceal mucinous neoplasm (LAMN), and 46.2% (6 of 13) mucinous adenocarcinomas not otherwise specified. LAMNs were predominantly high-risk with extra-appendical neoplastic epithelium (62.1%, 18 of 29), low-risk without extra-appendiceal neoplastic epithelium (24.1%, 7 of 29), or not specified or ambiguous (13.8%, 4 of 29). Greyed boxes indicate concordant diagnoses between the originating pathologist and academic pathologist. AdCA indicates adenocarcinoma; LAMN, low-grade appendiceal mucinous neoplasm. For cases in which the originating pathologist rendered a diagnosis of adenocarcinoma, the reference pathologist rendered a diagnosis of adenocarcinoma (56.5%, 13 of 23), LAMN (39.1%, 9 of 23), or simple mucocele (4.3%, 1 of 23). For cases in which the originating pathologist rendered a diagnosis of LAMN, the reference pathologist rendered a diagnosis of LAMN (87.0%, 20 of 23) or simple mucocele (13.0%, 3 of 23). Conversely, when the reference pathologist rendered a diagnosis of adenocarcinoma, the originating pathologist was universally in agreement (100%, 13 of 13). When the reference pathologist rendered a diagnosis of LAMN, the originating pathologist rendered a diagnosis of adenocarcinoma (31.0%, 9 of 29) or LAMN (69.0%, 20 of 29). When the reference pathologist rendered a diagnosis of simple mucocele (non-neoplastic), the originating pathologist rendered a diagnosis of adenocarcinoma (25%, 1 of 4) or LAMN (75%, 3 of 4, Figs 2 and 3 ). A and B, Appendiceal cross sections with concentric mucosal atrophy and architectural changes of the crypts. C, While decreased, the crypts are focally present within the lamina propria and no cytologic atypia is seen. D, Appendiceal cross section with crypt architectural changes indicating prior mucosal damage. E and F, Appendiceal tip with architectural changes and evidence of prior rupture, including transmural fibrosis and extra-appendiceal mucin. (hematoxylin& eosin, original magnifications x20 [A], x40x [B], 200x [C], 40x [D], 20x [E], and 40x [F]). A, Appendiceal base with LAMN and acellular mucin. B, Cytologic atypia seen in LAMN with subjacent abnormal fibrotic stroma without intact lamina propria. C, Florid case of LAMN. D, Cytologic atypia seen in LAMN with undulating and flat profiles and abnormal underlying stroma. E, Invasive adenocarcinoma arising from LAMN and penetrating smooth muscle of the muscularis propria. F, Invasive mucinous adenocarcinoma arising from LAMN with invasion of the smooth muscle layers seen in the bottom left portion of the micrograph. G, LAMN. H, LAMN with abnormal stroma lacking an intact lamina propria. I, Peritoneal extension of LAMN which resembles the luminal neoplasm and includes a hyalinized stroma. (hematoxylin& eosin, original magnifications x40 [A], x200 [B], x40 [C], x200 [D], x400 [E], x400 [F], x200 [G], x200 [H], and x200 [I]). The differences in distributions of major diagnostic categories resulted in an overall concordance rate of 71.7% (33 of 46, Fig 1 ). When utilizing the secondary academic pathologist evaluation as the reference evaluation, the originating pathologists diagnosed invasive adenocarcinoma with a concordance rate of 100% (13 of 13), whereas LAMN was diagnosed with a concordance rate of 69% (20 of 29), and mucocele was diagnosed with a concordance rate of 0% (0 of 4). The overall unweighted kappa value was 0.48 (95% CI, 0.27–0.69) consistent with moderate interobserver agreement. It should be noted that in no case were there disagreements in interpretation due to under-interpretation by the originating pathologist. Therefore, all of the discordance (28.3%, 13 of 46 cases) can be attributed to over-interpretation by the originating pathologist when compared to the secondary reference pathologist. Of the 29 LAMN cases as diagnosed by the reference pathology group, nine were originally diagnosed as adenocarcinoma (31.0%, 9 of 29). Additional potential diagnostic deficiencies were identified including discordant documentation of features needed for risk stratification (6.9%, 2 of 29) in which both such cases were documented as confined to the appendix at the originating pathology practice; whereas, the reference pathologist believed there to be extra-appendiceal neoplastic epithelium (i.e. high risk of recurrence). Several cases (31.0%, 9 of 29) lacked documentation of extra-appendiceal neoplastic epithelium by histological description, risk stratification (i.e. “high risk of recurrence” or “low risk of recurrence”), or TNM staging in the original pathology report; however, one of these cases (1 of 9) could not be categorized by the reference pathologist. Two LAMN cases (6.9%, 2 of 29) used non-standard terminology, namely “mucinous cystic neoplasm of uncertain malignant potential” and “appendiceal mucinous neoplasm with focal high-grade dysplasia” according to consensus guidelines which would advocate the use of LAMN and high-grade appendiceal mucinous neoplasm (HAMN), respectively.[ 3 ] It should be noted that advocacy for the use of HAMN for primary intraluminal appendiceal mucinous neoplasm that mimics LAMN but has high-grade cytologic features was strong, but not universal in the consensus statement (68%, 30 of 44).[ 3 ] Importantly, neither of these two cases had documentation of the presence or absence of extra-appendiceal neoplastic epithelium. Therefore, a total of 19 of 29 LAMN cases (65.5%) had potential diagnostic deficiencies when compared to the reference pathology evaluation.

LAMN and their peritoneal dissemination remain a difficult area for appropriate pathological classification. This study demonstrates a moderate diagnostic agreement for the diagnosis of primary appendiceal mucinous neoplasia in normative pathology and patient referral practice. Published consensus guidelines may facilitate clarification of classification and improvement of diagnostic concordance; however, clarified algorithmic diagnostic approaches based closely on the anatomy and biological potential of these neoplasms are needed. Because of the inherent difficulties of these cases, many with borderline malignant potential and disrupted anatomy, we recommend consideration of expert secondary pathological review at large academic referral centers that frequently encounter these lesions. Further studies are needed to help clarify clinical outcomes and improve current diagnostic evaluation in part by incorporating emerging biomarkers will the goal of providing improved prognostic and therapeutic information for patient care regarding this disease.

This study was approved by the Institutional Review Board Human Research Protections Program at the University of California San Diego (UCSD). Informed consent was not required for this study. Retrospective chart review was performed for a 2-year period (2014–2015) and included pathology records obtained from the anatomic pathology laboratory information system (AP LIS) and the general electronic medical record. The AP LIS was searched to identify pathology cases containing appendix resection specimens (i.e. appendectomies, right hemicolectomies, etc.) with mucinous neoplasia evaluated both by an outside pathology group and by in-house slide review at UCSD during patient referral. Cases were included if there was a primary appendiceal neoplasm or lesion, and excluded if incidental or contained secondary neoplasia or other unrelated pathology. In addition, pathology records had to be readily available at the time of chart review. A retrospective chart review of original pathology reports was performed for all cases. Prospective pathological evaluation at the reference institution, a single large academic medical center (UCSD), was performed solely by a subspecialized GI pathology clinical service via protocol in-house slide review during patient referral for oncologic care. None of the cases were secondary consultations specifically requested by the original evaluating pathologist. Reference pathologists were not provided with prior study specific training sets. The histopathological slides were reviewed prospectively in 91% of the overall cases by at least one fellowship-trained GI pathologist and 74% by a GI pathologist with special interest in appendiceal neoplasia (MV). All cases having disagreement with the originating pathologist were evaluated prospectively by the lead study pathologist (MV) and shown to more than one reference pathologist. In general, such cases are returned to the originating institution after a short grace period; therefore, a comprehensive retrospective histopathological review was not performed. Example microscopic images from selected available cases are provided for illustration. Concordance was measured as the total number of cases placed in the same diagnostic category (LAMN, adenocarcinoma, mucocele) by both the initial pathologist report and the second evaluation report. Cases were placed in the adenocarcinoma category if the term “adenocarcinoma” was included in the final pathological diagnosis. For cases without adenocarcinoma in the final pathological diagnosis, utilization of the term LAMN or other borderline malignant terminology such as “uncertain malignant potential” was placed the case in the LAMN category. “High-risk” LAMN refers to the presence of extra-appendiceal neoplastic epithelium which confers a high-risk of disease recurrence in the peritoneum, whereas “low-risk” LAMN here refers to the lack of extra-appendiceal neoplastic epithelium and includes cases with only extra-appendiceal acellular mucin and also cases thought to be entirely confined to the appendix [ 15 ]. Cases in which the location of the neoplastic epithelium could not be definitively determined were categorized as ambiguous, and cases lacking documentation of the presence or absence of extra-appendiceal neoplastic epithelium on the original pathology report were noted. In addition, mucocele as used herein refers to a non-neoplastic simple mucocele (sometimes also called a “mucous retention cyst”). Definitions of over-interpretation and under-interpretation are utilized as described previously [ 24 ]. Briefly, over-interpretation is defined as reporting a diagnosis of a higher category of malignancy than the reference; whereas, under-interpretation is defined as reporting a diagnosis of a lower category of malignancy than the reference. Unweighted Kappa values to determine interobserver agreement were calculated according to Fleiss and Cohen.[ 25 ]

(PDF) Click here for additional data file.