|Home | About | Journals | Submit | Contact Us | Français|
Distinguishing between solid-pseudopapillary neoplasms (SPNs) and pancreatic neuroendocrine tumors (PanNETs) may pose a diagnostic dilemma. Both can demonstrate solid growth patterns, and both can be immunoreactive with neuroendocrine markers such as synaptophysin and CD56. One well-established feature of SPNs is the presence of hyaline globules, which in contrast has only rarely been reported in PanNETs. Clinicopathologic features of 361 cases originally classified as PanNETs were examined. Of these, 24 tumors (6.6%) had hyaline globules, raising the possibility of SPN. Immunohistochemistry for β-catenin was performed on these 24 neoplasms, and showed nuclear labeling in 6 cases. These 6 cases, which also demonstrated cytoplasmic CD10 staining, were reclassified as SPNs. The remaining 18 cases maintained their original diagnosis as PanNETs, and the hyaline globules in these cases were periodic acid-Schiff (PAS)-positive, diastase resistant and immunoreactive with alpha-1-antitrypsin. All 24 cases were histologically re-evaluated, and the pattern of invasion, presence of clear cells, and nuclear grooves were found to be helpful in distinguishing SPNs from PanNETs. We conclude that the presence of hyaline globules should raise SPNs in the differential diagnosis of a solid cellular neoplasm of the pancreas; however, these should not be used as the sole criterion in the diagnosis of SPNs, as hyaline globules may also been seen in 5% of PanNETs. Immunohistochemical and histologic features supporting the diagnosis of SPNs over PanNETs include CD10 and nuclear β-catenin labeling, an insidious pattern of invasion, clear cells, and nuclear grooves.
Differentiating between pancreatic neuroendocrine tumors (PanNETs) and solid-pseudopapillary tumors (SPNs) may be diagnostically challenging. Although SPN is classically described as a solid and cystic neoplasm consisting of poorly cohesive cells forming degenerative papillae, foamy histiocytes, and longitudinal nuclear grooves, SPNs may also exhibit a solid pattern of growth with uniform round-to-oval cells, mimicking PanNETs17,18,27,34. Conversely, pseudopapillary areas may also be seen in PanNETs34, and PanNETs may occasionally be cystic14. Additionally, neuroendocrine markers such as synaptophysin may be expressed focally in SPNs31,35,46. Immunohistochemistry for β-catenin has proven useful, as there is nuclear labeling in SPNs and membranous labeling in PanNETs1,33,49. Among the additional immunohistochemical markers which can assist in establishing the correct diagnosis are CD10, estrogen and progesterone receptors, and alpha-1-antitrypsin21,35,40. Because the biological behavior, genetics, and therapeutic implications of PanNETs and SPNs can differ, it is imperative to make the correct diagnosis3,4,7-9,11-13,15,16,18-20,22,23,26,29,30,32,37-39,41,43,44,47,50-59.
Hyaline globules are a feature which has classically been associated with SPNs2,25,28. These globules are typically PAS-positive intracytoplasmic inclusions which are immunoreactive for alpha-1-antitrypsin. Eosinophilic globules are not specific to SPNs, but have also been reported in other pancreatic lesions, including intraductal oncocytic papillary neoplasms48, hepatoid carcinoma36, and clear cell ductal adenocarcinoma42. In contrast, eosinophilic globules have only rarely been described in PanNETs10.
We reviewed a large series of neoplasms originally diagnosed as PanNETs for hyaline globules. We show that eosinophilic globules may be seen not only in SPNs but also in PanNETs, and describe morphologic features which favor SPN in cases where the distinction between SPNs and PanNETs may be difficult.
The surgical pathology and consultation files of the Johns Hopkins Hospital were searched for all cases from May 1984 to January 2010 originally diagnosed as pancreatic neuroendocrine tumors (PanNETs). A total of 361 cases were identified. All available pathology reports and slides were reviewed. This review yielded 24 neoplasms originally diagnosed as PanNETs which contained hyaline globules. Single immunolabeling was performed using the antibodies listed in Table 1. Immunohistochemical studies for synaptophysin and β-catenin were performed on all cases with hyaline globules, and immunolabeling for CD10 was performed as a confirmatory test in cases with nuclear β-catenin labeling. PAS, PAS with diastase, and immunohistochemistry with alpha-1-antitrypsin and trypsin were performed on all cases with hyaline globules to further characterize these globules. Because these globules were focal, they were not present on levels immunolabeled with alpha-1-antitrypsin in three cases, those immunolabeled with trypsin in two cases, and those stained with PAS and PAS with diastase in two cases. Immunohistochemistry for Ki-67 and CK19 was also performed.
The Ki-67 labeling index was measured manually using custom software written in ImageJ (Wayne Rasband, NIH, Bethesda, MD) to assist in performing the nuclear counts. Five images were acquired per case at a total magnification of 400x using a Q-Color3 digital camera (Olympus, Center Valley, PA) on an Olympus B-50 microscope (Center Valley, PA). Fields were selected that represented the highest density of Ki-67 positive cells.
Using an ImageJ macro, each image was presented to the user, who then labeled each Ki-67 positive nucleus with a color marker by clicking on it with the mouse. Each Ki-67 negative nucleus was labeled using a marker of a different color. Non-neoplastic nuclei in the image were not labeled. The positive and negative markers in each image were automatically counted and a labeling index was calculated: % Ki-67 = (positive nuclei / (positive nuclei + negative nuclei)). On this basis, tumors were then classified as grade 1 (Ki-67 index: ≤ 2%), grade 2 (Ki-67 index: 3-20%), and grade 3 (Ki-67 index: > 20%)3,43.
Categorical data were compared using Chi square (or Fisher’s exact test). Continuous variables, reported as means with standard deviations (SD), were compared using a student t-test. Statistical analyses were performed using the software programs PRISM (GraphPad, San Diego) and Microsoft Excel (Redmond, WA). A P value of < 0.05 was considered statistically significant.
Three hundred sixty-one pancreatic neoplasms originally classified as PanNETs were reviewed. Of these, 24 were noted to have hyaline globules. Due to the known association between hyaline globules and SPNs, immunolabeling with synaptophysin and β-catenin was performed to confirm whether all 24 cases were truly PanNETs or in fact solid SPNs misclassified as PanNETs (Figure 1, Table 2). In 18 of the 24 cases, nuclear labeling for β-catenin was absent (not shown), and synaptophysin was positive (diffusely in 17 cases, focally in 1 case), confirming these neoplasms as true PanNETs (Figure 1A and B). In 6 of the 24 cases, there was nuclear labeling for β-catenin and cytoplasmic CD10 positivity (not shown), and these neoplasms were re-classified as SPNs (Figure 1C and D). Synaptophysin immunoreactivity was present in 5 of 6 SPNs and was typically focal (not shown). Therefore, a total of 18 (5%) of 355 histologically confirmed PanNETs in this series contained hyaline globules.
The mean age in years (±SD) for the total number of cases with hyaline globules (n=24), as well as the confirmed PanNETs (n=18) and reclassified SPNs (n=6) was 54.1 (±13.4), 56 (±12.4), and 48.3 (±14.7), respectively (PanNETs vs. SPNs, P = 0.2312). There was no difference in patient gender (PanNETs: 55.6% male vs. SPNs: 50% male, P = 1.000). In 5 (27.8%) of the 18 confirmed PanNETs, the neoplasm was clinically functioning. Three of these were insulin secreting, while two were glucagon secreting. Two (11.1%) of the 18 confirmed PanNETs arose in patients with multiple endocrine neoplasia 1 (MEN 1).
The mean tumor size in cm (±SD) for confirmed PanNETs was 4.7 (±2.9). For the reclassified SPNs, it was 6.3 (±6.0) (PanNETs vs. SPNs, P = 0.3720). The body and tail were most frequently involved in confirmed PanNETs (72.1%), while the body and head were the most common sites of the reclassified SPNs (33.3% each). Tumors were predominantly unifocal (88.9% PanNETs vs. 100% SPNs, P =1.000). Both the confirmed PanNETs and reclassified SPNs were most frequently solid (72.2% and 50%, respectively), though they occasionally showed pure cystic or a combination of solid and cystic growth. A statistical comparison of the location distribution, focality, and type of growth in PanNETs vs. SPNs was precluded due to the small sample size.
The hyaline globules in the 24 cases ranged dramatically in size, and were both intracytoplasmic and extracellular (Figure 1). They did not show any unique distribution within the tumor. In both PanNETs and SPNs, they were frequently focal, and therefore no significant differences in their prevalence in PanNETs vs. SPNs were readily recognized. In all tested cases, they were PAS-positive (Figure 2A, Table 2), diastase resistant. Hyaline globules in 12 (80%) of the 15 tested confirmed PanNETs, as well as 6 (100%) of the 6 tested reclassified SPNs had globules immunoreactive with alpha-1-antitrypsin (Figure 2B, Table 2). In one PanNET, the globules were positive for trypsin (not shown).
The 24 cases with hyaline globules were histologically re-evaluated to assess whether any features aside from CD10 and nuclear β-catenin labeling could help distinguish PanNETs with hyaline globules from SPNs. All 24 cases were histologically examined for the pattern of invasion, and presence or absence of the following features: microcystic change, hemorrhage, cholesterol clefts, clear cells, foam cells, nuclear grooves, and foci of discohesion within the tumor (Figure 3, Table 2). Clear cells, characterized by clear cytoplasm and an eccentric nucleus, were one feature more commonly seen in SPNs (Figure 3A, 100% SPNs vs. 27.8% PanNETs, P = 0.0034). An additional feature frequently observed in SPNs was an insidious pattern of invasion (Figure 3B). In this pattern, the neoplastic cells subtly penetrate around and entrap normal pancreatic elements. Insidious invasion was noted in 83.3% of the SPNs and in only 11.1% of the PanNETs (P = 0.0027). Although rare and focal, longitudinal nuclear grooves were more often present in SPNs than PanNETs, while the nuclei of PanNETs more commonly were round with speckled chromatin and lacked grooves (Figure 3C and D, 50% SPNs vs. 0% PanNETs, P = 0.0099). Foci of discohesive, single cells were only rarely present in either neoplasm (Figure 3E), but surprisingly were not statistically more frequent in SPNs (33.3% SPNs vs. 5.6% PanNETs, P = 0.1433). Other features such as foam cells (Figure 3F), hyalinization (Figure 3G), cholesterol clefting (Figure 3H), microcystic change (not shown), and hemorrhage (not shown) were compared, but showed no statistically significant differences between SPNs and PanNETs (Table 2). Of the 6 SPNs, 4 were purely solid, with no pseudopapillary or pseudoglandular features, while 2 were predominantly solid with only a small focus of discohesion imparting a pseudopapillary appearance.
The PanNETs were assigned a WHO grade of 1, 2, or 3 on the basis of the Ki-67 immunolabeling index3,43. The Ki-67 labeling indices of the SPNs were similarly grouped for the purposes of comparison. Fifty percent of confirmed PanNETs vs. 66.7% of reclassified SPNs were grade 1, 50% of PanNETs vs. 33.3 % of SPNs were grade 2, and 0% of either PanNETs or SPNs were grade 3. Using American Joint Committee on Cancer (AJCC) staging criteria6, the primary tumor in cases of confirmed PanNETs was T1 in 22.2%, T2 in 33.3%, and T3 in 44.5% of cases. In cases of reclassified SPNs, the primary tumor was T2 in 100% of cases. Five (27.8%) of 18 PanNETs showed nodal metastases at the time of diagnosis, while 3 (16.7%) of 18 PanNETs had distant metastases. In contrast, none of the SPNs demonstrated nodal or distant metastases at the time of initial diagnosis. A statistical comparison of these parameters in PanNETs vs. SPNs could not be performed due to the small sample size.
The mean follow-up time in years (±SD) for the confirmed PanNETs was 5.7 (±5). For the re-classified SPNs, it was 10.2 (±3.3) (SPNs vs. PanNETs, P = 0.0532). Fifteen (83.3%) of 18 confirmed PanNETs and 5 (83.3%) of 6 reclassified SPNs were alive with disease over the duration of the follow-up period (SPNs vs. PanNETs, P = 1.000). Eight (44.4%) of the 18 PanNETs, and 0 (0%) of the 6 SPNs immunolabeled with CK19 (PanNETs vs. SPNs, P = 0.0664). Only patients in the PanNET group developed metastases subsequent to the time of initial diagnosis. One (5.6%) of the 18 PanNETs metastasized to lymph nodes, and 3 (16.7%) had a metastasis to the liver. The mean time (± SD) in years from the initial diagnosis to the metastasis was 3.2 (±2.4).
In summary, we identified hyaline globules in 6.6% of 361 neoplasms originally diagnosed as pancreatic neuroendocrine tumors (PanNETs). Although 6 of these cases with hyaline globules were shown to be misdiagnosed solid-pseudopapillary neoplasms (SPNs), the majority were confirmed to be PanNETs with hyaline globules. Thus, although hyaline globules should raise the possibility of SPNs, hyaline globules may also be found in 5% of PanNETs.
There can be considerable morphologic overlap between SPNs and PanNETs17,18,27,34,34. Importantly, we found that several histologic features, in addition to the hyaline globules, can facilitate the distinction between PanNETs and SPNs. An insidious pattern of invasion, clear cells, and nuclear grooves are features which, in addition to immunolabeling with β-catenin, favor SPN. Features such as foam cells, hyalinization, cholesterol clefting, microcystic change, hemorrhage, and foci of discohesion were no different in PanNETs vs. SPNs. Relying on pseudopapillary or pseudoglandular features to make the distinction between PanNETs and SPNs was of limited utility: 4 of 6 SPNs were solid with no such features, and 2 of 6 SPNs were predominantly solid with only a small discohesive focus conferring a pseudopapillary appearance.
When confronted with the diagnostic dilemma of PanNET vs. SPN, a number of immunomarkers can help establish the correct diagnosis3. Notohara et. al35 have reported that CD10, CD56, vimentin, synaptophysin, chromogranin, and pancytokeratin are useful in distinguishing between SPNs and PanNETs. In the current study, we advocate a panel of immunomarkers to include synaptophysin, β-catenin, and (if the β-catenin is nuclear) CD10 to make the distinction between PanNETs and SPNs, whether hyaline globules are present or not. Trypsin and chymotrypsin can be included in the panel since acinar cell carcinomas are also solid cellular neoplasms and should be in the differential diagnosis. All 18 PanNETs strongly expressed synaptophysin (diffusely in 17 tumors and focally in 1 tumor). Although 5 of the 6 SPNs were immunoreactive for synaptophysin, such labeling was typically only focal. In contrast, all 18 PanNETs had a membranous pattern of labeling for β-catenin, while all 6 SPNs had abnormal nuclear β-catenin immunolabeling and cytoplasmic CD10 staining.
Among 361 tumors originally classified as PanNETs, 18 were true PanNETs with hyaline globules. This corresponds to an incidence of hyaline globules in PanNETs of nearly 5%. The hyaline globules in PanNETs are PAS-positive/diastase resistant, and are most frequently immunoreactive with alpha-1-antitrypsin. Garg et al. previously reported a case of a PanNET with PAS-positive hyaline globules10. This case was of ectopic ACTH syndrome in a 12-year-old boy with a pancreatic neuroendocrine tumor consisting of monomorphic cells and delicate vessels. In the current study, only 5 (27.8%) of the PanNETs were clinically functional, three of which were insulinomas, two of which were glucagonomas, and none of which were ACTH-secreting. The PanNETs with hyaline globules in our series were often large, a number of which had nodal and distant metastases at the time of initial diagnosis, as well as subsequent metastases.
Six of the 24 cases originally diagnosed as PanNETs were in fact SPNs with hyaline globules. Such globules have long been associated with SPNs, and their presence in a solid cellular neoplasm of the pancreas should raise the possibility of an SPNs in the differential diagnosis2,25,28. As with those in PanNETs, hyaline globules in SPNs were PAS-positive/diastase resistant, and were also immunoreactive with alpha-1-antitrypsin. Unlike PanNETs, however, SPNs typically showed more focal synaptophysin positivity as has been previously documented31,35,46, as well as immunoreactivity with β-catenin (nuclear) and CD10 1,33,35,49.
The clinical importance of correctly distinguishing between SPNs and PanNETs is growing, from the standpoints of therapeutics, genetics, and outcome. SPNs are treated surgically; historically, chemoradiation has not played a beneficial role24,29,58. PanNETs, too, may be surgically resected, and a number of studies have shown that they are also amenable to both chemotherapy and radiation4,26,30,41,47,51,53,54,57. Among the non-surgical approaches to the treatment of PanNETs are peptide receptor radionuclide therapy (PRRT), chemotherapy, somatostatin analogs, and interferon8. A recent study showed that PanNETs harbor frequent alterations in several genes, including those involved in the mammalian target of rapamycin (mTOR) pathway15. Specific targeted therapies, such as Everolimus, have been shown to be effective in some patients with PanNETs, and these therapies are likely to be effective only in neoplasms, such as PanNETs, in which the mTOR pathway has been activated8,13,32,56,59. Although a few studies have suggested that a proportion of SPNs behave in an aggressive fashion22,23,29,39,44,50,52, several studies have demonstrated that they have an excellent prognosis11,16,18,37. PanNETs can also exhibit a spectrum of behaviors5,7,9,9,12,19,20,38,43,45,55, and their malignant potential may not be readily ascertained histologically.
In conclusion, the morphologic features of SPNs and PanNETs overlap. Small SPNs with minimal degenerative changes are particularly hard to recognize. In these cases, hyaline globules should raise SPNs in the differential diagnosis, and the diagnosis can be established with immunolabeling for β-catenin and CD10. Additionally, the presence of hyaline globules should not be used as a sole diagnostic criterion for SPNs, as 5% of PanNETs contain hyaline globules. Features supporting the diagnosis of SPN include nuclear β-catenin labeling and CD10 expression, an insidious pattern of invasion, clear cells, and nuclear grooves.
This work was funded by a GI SPORE grant (NIH grant P50-CA62924).
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.