Polymorphous low-grade adenocarcinoma: an analysis of epidemiological studies and hints for pathologists

Background This study is an analysis of the prevalence of polymorphous low grade adenocarcinoma (PLGA) in epidemiological surveys of salivary tumors published in the English language from 1992 to 2012. Methods These surveys included studies from different researchers, countries and continents. The 57 surveys for which it was possible to calculate the percentage of PLGAs among all malignant minor salivary gland tumors (MMSGT) were included in this review. Results The statistical analyses show significant differences in the PLGA percentage by time period, country and continent in the studies included in this review. The percentage of PLGAs among MMSGTs varied among the studies, ranging from 0.0% to 46.8%. PLGA rates have varied over the period studied and have most recently increased. The frequency of reported PLGA cases also varied from 0.0% to 24.8% by the country in which the MMSGT studies were performed. The PLGA percentages also varied significantly by continent, with frequencies ranging from 3.9% in Asia to 20.0% in Oceania Conclusion Based on these results, we concluded that although the accuracy of PLGA diagnoses has improved, they remain a challenge for pathologists. To facilitate PLGA diagnoses, we have therefore made some suggestions for pathologists regarding tumors composed of single-layer strands of cells that form all of the histological patterns present in the tumor, consistency of the cytological appearance and uniformly positive CK7, vimentin and S100 immunohistochemistry, which indicate a single PLGA phenotype. Virtual slide The virtual slide(s) for this article can be found here: http://www.diagnosticpathology.diagnomx.eu/vs/1059098656858324


Introduction
Polymorphous low-grade adenocarcinoma (PLGA) is a malignant epithelial tumor characterized by cytological uniformity, morphological diversity, an infiltrative growth pattern and low metastatic potential [1]. This tumor was recognized as a distinct entity in 1983 by Freedman and Lumerman and Batsakis et al., and it was named polymorphous low-grade adenocarcinoma by Evans and Batsakis in 1984 [2][3][4].
Clinically PLGA presents as an indolent asymptomatic swelling but occasionally can be painful and even ulcerate. The most common location of PLGA is the palate, although other locations have been described. It occurs more frequently in women affecting mainly the sixth and seventh decade of life. For more details on clinical presentation, prognosis and treatment, we recommend the reviews by Pogodzinski et al., and Paleri, Robinson and Bradley [5,6]. In general these authors indicate a low grade malignancy and good prognosis of this tumor They also recommend a very careful and systematic follow-up,since recurrences and rare metastases can occur many years after the surgery. The tumor is characterized by single-layer strands of cells that can form lobular, tubular, cribriform, trabecular, papillary-cystic and cystic histological patterns, which can be illustrated by the presence of extracellular matrix between the strands of cells identified in lobular or solid patterns [7].
Most PLGA cells are cytologically uniform and range from small to medium in size, with vesicular oval nuclei and inconspicuous nucleoli. Their cytoplasm is ample and exhibits a variable appearance, including eosinophilic, basophilic and clear aspects. The cells have indistinct outlines that lend a syncytial pattern to the active cellular mass. Groups of cells with a coarsely eosinophilic granular cytoplasm, mimicking oncocytes, are occasionally observed, as are mucous cells [1,8].
The cells show a unique electron microscopy and immunohistochemical phenotype. All cells have microvilli apically and are attached to the basal lamina. The cells are positive for vimentin, CK 7 and S100, a pattern only shared by the mammary analogue secretory carcinoma, as recently described by Skalova et al. and rarely by focal plasmacytoid cells in pleomorphic adenoma [8][9][10]. A regular distribution of positive staining for β1, β2 and β3 integrins and striking bipolar staining in all of the neoplastic cells reinforces this unique phenotype [11].
Single cells, usually infiltrating surrounding structures, and clear cells in nests are also observed in the lobular PLGA subtype. The stroma appears either strongly eosinophilic and hyalinized, or muco-hyalinized with a bluish tint. Foci of residual salivary gland acini surrounded by neoplastic cells are occasionally found. Peri-neural invasion by groups of tumor cells is a frequent finding, and psammoma-like structures are occasionally observed. This tumor frequently presents with prominent vascularity [8].
Despite greater understanding of this tumor, PLGA remains a diagnostic challenge for pathologists. This conclusion is based on the variability of the epidemiological results obtained by several groups who have studied this tumor.
We have reviewed the epidemiological studies in an attempt to analyse the proportion of PLGAs in salivary gland tumors.

Materials and methods
This analysis included 57 epidemiological studies of salivary gland tumors published in the English language from 1992 to 2012. The year 1992 marked the inclusion of PLGA in the World Health Organization (WHO) classification of salivary gland tumors [12]. Studies were included in the analysis if they contained the data needed to calculate the fraction of PLGAs in the malignant minor salivary gland tumor (MMSGT) total.
The studies addressing only major salivary gland tumors were excluded because the few cases published on that topic do not significantly contribute to the understanding of PLGA; similarly, studies that included only children and adolescents were excluded from this analysis.
When data were available, we extracted the following information: total number of salivary gland tumors; number of minor salivary gland tumors and their fraction of the total number of tumors; number of MMSGTs and their fraction of the total number of minor salivary gland tumors; number of PLGAs and their fraction of the total number of MMSGTs; and the total number of minor salivary gland tumors.
In this study, we analyzed the fraction of PLGAs in the total number of MMSGTs. It was not possible to obtain the absolute or relative frequencies of malignant salivary tumors, either among minor salivary tumors alone or among major and minor tumors, from studies that reported only MMSGTs.

Statistical analysis
Data were tabulated and descriptive statistics were calculated using frequency tables. G tests were used to ascertain whether the PLGA fraction of all MMSGTs varied by the year, country and continent in which the studies were performed. We would like to emphasize that at no point was it presumed that these studies reflect the prevalence of this tumor with respect to the aforementioned variables (year, country and continent). The significance level was set at 5%. The statistical calculations were performed using the SPSS 20 software package (IBM corporation, Armonk, NY, USA).

Results
Fifty-seven surveys of salivary gland tumors were included in this review (Table 1)  . From 26,960 cases of salivary gland tumors, 431 (1,6%) were accepted by the authors as been PLGAs.
There has been a significant increase (p < 0.0001 for the G test) in the fraction of PLGA cases reported in the literature since 2007, as shown in Table 2 Table 2).
The frequency of PLGA also varied significantly (p < 0.0001 for the G test) by country, as shown in Table 3. Of the 431 PLGA cases included in this review (Table 1), 95 (22.0%) were from studies performed in the USA, 83 (19.3%) were from Chinese studies and 66 (15.3%) were from Danish studies. The percentage of PLGAs among MMSGTs varied among the studies, ranging from 0.0% to 24.8% ( Table 3).
The frequency of reported PLGA cases also varied significantly (p < 0.0001) by the continent in which the MMSGT studies were performed. The continent with   the highest reported frequency of PLGAs was Asia, with 3,702 of the 6,891 reported cases (53.7%), followed by America (24.2%) and Europe (19.1%), as shown in Table 4. The PLGA percentages also varied significantly by continent, with frequencies ranging from 3.9% in Asia to 20.0% in Oceania.

Discussion
Analysis of the data from 57 epidemiological studies reflects a variety of methodologies, some examined all (major and minor) salivary gland tumors, while others examined only tumors of the minor glands but included benign and malignant tumors or even MMSGTs alone. This variability most likely reflects differences between the institutions from where most of the data were collected, such as hospitals and medical or dental schools. In other words, it does not reflect the real epidemiology of this tumor in these countries or continents, since they are a few isolated reports. Nevertheless, it was possible to discern the PLGA percentages among the MMSGT cases, which was the aim of this study. We observed that PLGA rates have varied over the period studied and have most recently increased, most likely due to improved PLGA diagnostic accuracy. Over the last two study periods, the PLGA fraction has stabilized at a value that probably reflects a more accurate percentage of PLGAs among MMSGTs.
We also noted that the percentage varied by the continent where the studies were performed and by individual authors. Based on these results, we suggest that geographical differences alone cannot account for the varying incidence rates, such as occurs with Warthin tumor, which has a lower incidence in Africa, and with the lymphoepithelial carcinoma that has an evident predilection for Inuits (Eskimo), Chinese and Japanese [33,37,69,70]. Also based on these differences it is impossible to extracting other important data as the differences in ACC survival rates between Chinese and occidental data as recently demonstrated by Zhou et al. [71].
Despite our improved understanding of this entity over time, worldwide differences found amongst the studies indicate that diagnosing PLGA remains challenging, probably because histological and cytological criteria are not uniformly applied. Interestingly, this diagnosis does not appear in some of the series, which used the designation "adenocarcinoma" with no further definition, which raises the question of whether a tumor is actually an adenocarcinoma NOS, a PLGA or another entity.
Since the 1990s, many studies have attempted to develop a useful marker for PLGA or to differentiate it from other histologically similar tumors [72][73][74][75]. To date there has been no reliable molecular marker to distinguish Table 4 The numbers and percentages of polymorphous low-grade adenocarcinomas in malignant minor salivary gland tumors by continent as described in the studies included in this review as described in the studies included in this review  PLGA from other MMSGTs [76]. The major research focus is currently on finding immunohistochemical differences between PLGA and adenoid cystic carcinoma (ACC), mainly in the cribriform histology, common to both tumors, which has been tirelessly attempted [77][78][79][80][81][82][83][84][85][86][87]. Controversy on this subject persists in the literature. Some authors believe that immunohistochemistry does not have any proven diagnostic value for identifying PLGA [6,78,88,89]. However, we do not share this opinion as we have successfully used immunohistochemistry in difficult cases or to confirm a histological diagnosis.
For diagnostic purposes, it is essential to characterize the morphology of the cell, the diversity of the histological tumor patterns and to recall that the PLGA cellular population exhibits a constant cytological appearance, despite a variety of growth patterns.
In our experience it is important to note that tumor cytology and histology are usually sufficient for a final diagnosis. However, immunohistochemistry is valuable in unclear PLGA cases, however. Uniformly positive vimentin and CK 7 staining, except for the rare two-layer ducts, is sufficient for a final PLGA diagnosis (Figure 1). S100 is also positive in almost all of the cells, but this characteristic is only diagnostically supportive. When examining cytoskeleton filaments in salivary gland tumors, it is also important to observe which cells are positive for each protein, rather than simply indicating the percentage of tumors in a series that are positive for each marker. Using this information, the immunohistochemistry of the cytoskeleton filament contributes greatly to the diagnosis of salivary gland tumors, especially PLGAs.