Literature survey on epidemiology and pathology of gangliocytic paraganglioma

Gangliocytic paraganglioma (GP) was first reported as ganglioneuroma by Dahl et al. in 1957 [1] and Kepes et al. named this disease entity as GP in 1971 [2]. Although GP has generally been regarded as a neuroendocrine tumor, some authors have reported that GP is a hamartoma developing in misplaced embryonic pancreatic tissue [3‐5]. Despite these investigations, the origin of GP remains unclear. However, a few cases of GP showing lymph node metastasis are well known [Table 1], and we previously reported such a case [6]. That case prompted us to survey publications of GP to investigate mechanisms for GP. In the present study, we searched literature databases and selected publications of GP in English and Japanese from 1957 to the early trimester of 2010 in order to investigate details of the disease.

Few studies reported the epidemiology, pathology, and clinical characteristics of the disease. We then evaluated the details of the disease by carefully analyzing and extracting common features of the disease as presented in these selected publications.

We searched English and Japanese cases of GP published from 1957 to the early trimester of 2010 using the PubMed http://​www.​ncbi.​nlm.​nih.​gov/​pubmed/​ and IgakuChuoZasshi http://​www.​jamas.​or.​jp/​ databases on August 2010 by conducting a search of "gangliocytic paraganglioma" with case report options as an additional tool. As a result, 4574 English and 27 Japanese publications were retrieved.

We checked the abstracts and 73 English and 24 Japanese (total 97) publications were regarded as cases of GP and 4504 publications were excluded.

Since this disease entity has been reported under other names, prior to Kepes et al. naming it as a GP [2], such as "ganglioneuroma", "non-chromaffin paraganglioma", and "paraganglioneuroma", we checked the references of 97 selected publications.

Finally, we added 8 publications as GP reports because they met the following criteria:

These 105 publications contained reports on 192 patients with GP. In the present study, we conducted a non-systematic literature review using 173 patients with duodenal GP.

We extracted and sampled raw data from the selected publications, such as clinical findings (age, sex, clinical symptoms, operation method, and outcome), histopathological findings (site, maximum diameter of the tumor, diagnostic rate from biopsy before operation, the depth of the tumor invasion, and with or without lymph node metastasis), and immunohistochemical findings.

In addition, we performed appropriate statistical analyses using the extracted data. Statistical analyses were performed using the non-parametric Mann-Whitney U test or Chi-square test.

Differences were considered significant at P < 0.05.

Additionally, we evaluated gender differences based on the results of our statistical analyses. Moreover, we evaluated the expression of hormone receptors in relation to estrogen and progesterone based on our previously reported case [6].

Finally, we declare that our study may be affected by publication bias because it comprises a cumulating case series.

We collected as many publications written in English or Japanese involving GP and examined them to extract characteristic features of the tumor as presented in clinical and histopathological findings.

We focused on determining the significant factors associated with the progression of the tumor. Here we discuss the clinical, histopathological, and immunohistochemical findings in reference to matters that emerged from our research.

In the present study, we found gastrointestinal bleeding as the most common symptom, followed by abdominal pain. This fact has been previously accepted and our data confirm it. All patients without autopsies and having clinically incidental focus had surgical interventions, including 15 patients with endoscopic intervention. Indeed, one patient required additional surgical intervention due to a residue of the tumor at her first endoscopic procedure [36], while other patients showed a good outcome without recurrence or metastasis. Furthermore, there was no record of a patient dying from GP, and patients with this tumor, therefore, have an extremely good prognosis. However, one patient has been reported as showing a recurrence due to a residue of a previous tumor at his initial surgical intervention [24]. Accordingly, we emphasize the importance of both a histopathological assessment of extensive tumor components at the surgical margin and imaging examinations to monitor for recurrence or metastasis after the operation.

Although one patient received irradiation after surgical intervention [25], we maintain that patients without residual tumor require no adjuvant therapy because no recurrence or metastasis has been reported in such patients. By contrast, it is still unclear whether a residual tumor can be controlled by irradiation or chemotherapy alone without surgical intervention.

The immunohistochemical findings on the tumors need consideration since the identification of three cellular components are essential for diagnosis. Epithelioid and ganglion-like cells showed a high positive rate for several kinds of immunohistochemical neuroendocrine markers, such as synaptophysin, chromogranin A, and NSE. In addition, epithelioid cells showed a high positive rate for PP. In contrast, spindle-shaped cells had the highest positive rate for S-100. These results are consistent with other previous publications. Furthermore, positive rates for each hormone, such as somatostatin, serotonin, gastrin, glucagon, and insulin in epithelioid cells, were significantly different and the meaning of this finding is worth investigating. However, these extracted immunohistochemical findings should be regarded as hints or suggestions because it is thought that results involving negative data have not been described in previous publications. In fact, our previous case report did not describe the immunohistochemical evaluation of each hormone according to negative reactivity.

Incidentally, we previously tried to establish the immunohistochemical prognostic indicators of GP using bcl-2, p53, and Ki-67, which are acceptable prognostic indicators in several kinds of neuroendocrine tumors [41‐44]. However, all of these indicators showed a negative reactivity.

Unfortunately, there were no other cases of GP that included an immunohistochemical evaluation using bcl-2 and p53, and the value of these factors as prognostic indicators of GP remains unclear.

In contrast, two cases without lymph node metastasis were described with the Ki-67 labeling index in GP [45, 46]. However, both showed extremely low Ki-67 labeling index values. Therefore, we suggest that immunohistochemical evaluation using Ki-67 may have a limited prognostic value in GP.

Finally, we gained insight into the progression of GP tumors and related factors. It has been accepted that GP usually arises from the submucosal or muscular layer, which may make the diagnosis difficult using a forceps biopsy prior to surgical intervention. In fact, we revealed that the diagnostic rate by biopsy before surgical intervention was only 11.4% (4/35). In addition, we showed that many more cases of GP exceeding the submucosal layer were reported (61.1%, 66/108) than expected, and GP exceeding the submucosal layer is a risk factor for lymph node metastasis. These facts emphasize the importance of imaging examinations prior to surgical interventions.

It is interesting to note that significant differences were found for gender between GP within the submucosal layer and exceeding the submucosal layer. On the basis of our investigation, a hypothesis emerged that asserted no significant relationship for gender and that female gender induces vertical growth of the tumor. To confirm part of our hypothesis, we focused on female-specific factors and initially evaluated tumor cells immunohistochemically using anti-estrogen and progesterone receptor antibodies. As a result, our immunohistochemical evaluation revealed that epithelioid cells showed positive reactivity for the progesterone receptor. Furthermore, some investigators reported that progesterone regulates neural differentiation [47, 48]. These facts suggest that the vertical growth of GP might be affected by progesterone exposure.

Additionally, it is interesting to note that normal pancreatic islet cells also showed positive reactivity for the progesterone receptor. It has been reported that normal pancreatic islet cells and pancreatic neuroendocrine tumors showed positive reactivity for the progesterone receptor [49], and our literature survey demonstrated that epithelioid cells showed a high positive rate for PP (89.7%, 70/78) immunohistochemically. These facts indicate the relationship between GP and pancreatic islet cells. However, immunohistochemical evaluation for estrogen receptors differs between epithelioid cells and normal pancreatic islet cells, and our immunohistochemical evaluation was based on only one patient. Furthermore, it is strongly suspected that our data have been affected by publication bias.

To confirm these hypotheses, further investigation is required (e.g. compare the positivity between the metastatic cases [6, 24, 25, 36, 46, 50‐56] and non-metastatic cases for the immunohistochemical antigen expression).

Our literature survey revealed that there were many more cases of GP exceeding the submucosal layer than were expected, and significant differences were found in gender between GP within the submucosal layer and exceeding the submucosal layer. Furthermore, our immunohistochemical evaluation showed that both epithelioid cells and normal pancreatic islet cells showed strongly positive reactivity for progesterone receptors. These immunohistochemical results support the following two hypotheses. First, vertical growth of GP might be affected by progesterone exposure. Second, the origin of GP might be the pancreatic islet cells. However, further investigation is required to confirm these hypotheses.

Since the data of this study were extracted and sampled from previous publications, written informed consent for publication from patients does not exist with the exception of our previous publication. As for our previously reported cases, written informed consent was obtained from the patients for publication of this study as well as any accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal. Furthermore, the anonymity of all patients was strictly protected.