Pattern of placental alkaline phosphatase (PLAP) expression in human tumors: a tissue microarray study on 12,381 tumors

Abstract Placental alkaline phosphatase (PLAP) is commonly expressed at high levels in testicular germ cell tumors. PLAP immunohistochemistry (IHC) is thus often used to confirm this diagnosis, especially in cases of putative metastasis. However, other tumors can also express PLAP. To comprehensively determine PLAP expression in normal and tumor tissue, a tissue microarray containing 16,166 samples from 131 different tumor types and subtypes as well as 608 samples from 76 different normal tissue types was analyzed by IHC. Moderate to strong PLAP positivity was found in 27 (21%) of 131 different tumor types including seminoma (96%), embryonal carcinoma (85%), and yolk sac tumors of the testis (56%); endometrioid carcinoma of the endometrium (28%) and the ovary (20%); gastric adenocarcinoma (22%); serous carcinoma (not otherwise specified) of the ovary (17%) and the uterus (11%); adenocarcinoma of the ampulla of Vater (15%); carcinosarcoma of the ovary (11%) and the uterus (8%); esophageal adenocarcinoma (10%); invasive urothelial carcinoma (4%); cholangiocarcinoma (2%); and adenocarcinoma of the lung (1%). Low‐level PLAP immunostaining, often involving only a small fraction of tumor cells, was seen in 21 additional tumor entities. The clinical significance of PLAP expression may vary between tumor types as high PLAP expression was linked to advanced pathological tumor stage (p = 0.0086), nodal metastasis (p = 0.0085), and lymphatic (p = 0.0007) and blood vessel invasion (p = 0.0222) in colorectal cancer, but to low pathological tumor stage in endometrial cancer (p = 0.0043). In conclusion, our data identify several tumor entities that can show PLAP expression at comparable levels to testicular germ cell tumors. These tumor entities need to be considered in cases of PLAP‐positive metastasis. Low‐level PLAP expression can be found in various other tumor entities and should generally not be viewed as a strong argument for germ cell neoplasia.


Introduction
Placental alkaline phosphatase (PLAP), also known as alkaline phosphatase, placental type (ALPP), is encoded by the ALPP gene at chromosome 2q37.1 [1]. PLAP is a dimer of 65 kDa consisting of 535 amino acids and is thought to play a role in guiding migratory cells and transport specific molecules over the plasma membrane [1,2]. PLAP is expressed in the placenta from the ninth week of gestation and its concentration increases continually throughout pregnancy [2]. PLAP can be separated into three distinct isoenzymes corresponding to early, mid, and term placenta [3]. In normal human tissues, the expression of PLAP is largely restricted to the placenta but low-level RNA expression has also been reported for uterine cervix, fallopian tube, andto a lower levelthe lung [4,5].
To better understand the prevalence of PLAP immunostaining in different tumor types, a comprehensive study analyzing a large number of neoplastic and non-neoplastic tissues under highly standardized conditions is needed. We thus analyzed PLAP expression in more than 16,000 tumor tissue samples from 131 different tumor types and subtypes as well as 76 non-neoplastic tissue types by immunohistochemistry (IHC) in a tissue microarray (TMA) format.

Tissue microarrays
TMAs composed of normal and tumor tissues were employed for this study. The normal TMA contained eight samples from eight different donors from each of 76 different normal tissue types. The cancer TMAs contained a total of 16,166 primary tumors from 131 tumor types and subtypes. Histopathological data including grade, pathological tumor (pT) stage, and pathological lymph node (pN) status were available from 583 ovarian cancers, 259 endometrial cancers, and 1,784 colorectal cancers. The dataset on colorectal cancer also included molecular information on mismatch repair protein deficiency. The composition of both normal and cancer TMAs is described in the Results section. All samples were from the archives of the Institutes of Pathology, University Hospital of Hamburg, Germany; the Institute of Pathology, Clinical Center Osnabrueck, Germany; and Department of Pathology, Academic Hospital Fuerth, Germany. Tissues were fixed in 4% buffered formalin and then embedded in paraffin. TMA tissue spot diameter was 0.6 mm. The use of archived remnants of diagnostic tissues for manufacturing of TMAs and their analysis for research purposes as well as patient data analysis has been approved by local laws (HmbKHG, §12) and by the local ethics committee (Ethics commission Hamburg, WF-049/09). All work was carried out in compliance with the Helsinki Declaration.

Immunohistochemistry
Freshly cut TMA sections were immunostained under the same experimental conditions. Two different primary antibodies were used for PLAP detection: MSVA-350R (rabbit recombinant; MS Validated Antibodies, Hamburg, Germany) and IR779 (mouse monoclonal 8A9, Agilent DAKO, Santa Clara, CA, USA). The normal tissue array was analyzed with both MSVA-350R and IR779, while the multitumor array was analyzed with MSVA-350R only. For MSVA-350R, slides were deparaffinized with xylol, rehydrated through a graded alcohol series, and exposed to heat-induced antigen retrieval for 5 min in an autoclave at 121 C in pH 9.0 Target Retrieval Solution (Agilent). Endogenous peroxidase activity was blocked with Peroxidase Blocking Solution (Agilent) for 10 min. The primary antibody was diluted 1:150 and applied for 60 min at 37 C. For IR779, the slides were deparaffinized and rehydrated as described previously, and exposed to heat-induced antigen retrieval for 15 min in Agilent's PT Link pretreatment module at 95 C in pH 9.0 retrieval buffer. Slides were transferred to an Autostainer Link 48 device (Agilent) for peroxidase blocking (5 min) and incubation of the primary antibody (ready to use prediluted for 20 min at room temperature). Both antibodies were visualized using the respective EnVision reagents (Agilent) for manual and automated staining according to the manufacturer's directions. One pathologist (NG) analyzed all immunostainings. For tumor tissues, the percentage of positive neoplastic cells was estimated, and the staining intensity was semiquantitatively recorded (0, 1+, 2+, and 3+). For statistical analyses, the staining results were categorized into four groups. Tumors without any staining were considered negative. Tumors with 1+ staining intensity in ≤70% of cells and 2+ intensity in ≤30% of cells were considered weakly positive. Tumors with 1+ staining intensity in >70% of cells, 2+ intensity in 31-70%, or 3+ intensity in ≤30% were considered moderately positive. Tumors with 2+ intensity in

Statistics
Statistical calculations were performed with JMP 14 software (SAS Institute Inc., Cary, NC, USA). Contingency tables and the chi-square test were performed to search for associations between PLAP and tumor phenotype. Survival curves were calculated according to Kaplan-Meier. The log-rank test was applied to detect significant differences between groups. A P value of ≥0.05 was considered as statistically significant.

PLAP in normal tissues
With two different antibodies (MSVA-350R and IR779), particularly strong PLAP immunostaining was found in the placenta. Here, strong PLAP positivity was seen in chorion cells as well as in cyto-and syncytiotrophoblast of mature placenta ( Figure 1A,D). Staining was only moderate and limited to the surface cell membrane in the trophoblast of early placenta, and only weak in amnion cells. Also, for both antibodies, weak PLAP staining was seen at the apical membrane of epithelial cells in the endocervix ( Figure 1B,E), endometrium, and the fallopian tube, although this did not occur in all samples analyzed. PLAP immunostaining was lacking for both antibodies in most other tissues including all epithelial

PLAP in human tumors
cells of the gastrointestinal and the genitourinary tract, gallbladder, liver, pancreas, salivary and bronchial glands, breast glands, Brunner glands, thyroid, pituitary gland, adrenal gland, parathyroid gland, testis, epididymis, seminal vesicle, prostate, nonkeratinizing and keratinizing squamous epithelium of various different sites, skin appendages, hematopoietic and immune cells, and the brain. Staining of muscular tissues revealed complete absence of staining by MSVA-350R ( Figure 1C) while Agilent Dako IR779 showed moderate to strong staining of smooth muscle ( Figure 1F) and weak to moderate staining of skeletal muscle. These latter findings were considered to be due to cross-reactivity.  (Table 1). Representative images of PLAP-positive tumors are shown in Figure 2. The highest rate of positive staining was found in testicular tumors, followed by tumors of the female genital tract, gastroesophageal, and pancreaticobiliary cancers. It is of note that only weak PLAP immunostaining was occasionally found in 21 different tumor entities. In most of these tumors, PLAP immunostaining was limited to a small fraction of tumor cells ( Figure 2D,E). None of the 48 leiomyomas, 84 leiomyosarcomas, 7 rhabdomyosarcomas, or 91 angiomyolipomas showed any PLAP staining. A graphical representation of the rank order of PLAP positive and strongly positive cancers is shown in Figure 3.

PLAP expression and histopathological parameters
The relationship between PLAP expression and histopathological data in ovarian, endometrial, and colorectal cancers is summarized in Table 2. The data show that high PLAP expression is linked to advanced pT stage (p = 0.0086), nodal metastasis (p = 0.0085),

V Reiswich et al
and lymphatic (p = 0.007) and blood vessel invasion (p = 0.0222) in colorectal cancer, while low PLAP expression was found to be associated with advanced pT stage in endometroid carcinoma of the endometrium (p = 0.0043). Associations between PLAP expression and tumor phenotype were not found in serous (not otherwise specified) and endometrioid ovarian cancer.

Discussion
In an immunohistochemical analysis of more than 10,000 tumors analyzed by IHC, it is important to use suitable reagents and protocols. The International Working Group for Antibody Validation (IWGAV) has proposed that antibody validation for IHC on formalin-fixed tissues should include either a comparison of the findings obtained by two different independent antibodies or a comparison with expression data obtained by another independent method [62]. Here, 76 different normal tissue categories were included in the antibody comparison experiment to ensure that any antibody cross-reactivity would be detected in our validation experiment. The fact that the antibodies MSVA-350R and Agilent IR779 both showed strong PLAP staining in chorion and trophoblastic cells of the placenta and weak staining of amnion cells and apical membranes of endocervical, endometrial, and fallopian tube epithelium confirms that these findings are PLAP specific. Most of these results are also consistent with data from three independent RNA screening studies, including the Human Protein Atlas (HPA) RNA-seq tissue dataset [63], the FANTOM5 project [64,65], and the Genotype-Tissue Expression (GTEx) project [5], which also suggest that the uterine cervix is the organ with the second highest PLAP expression following placenta. PLAP RNA expression was not described for endometrium and fallopian tube, but this may be due to the small fraction of the total cells of these organs expressing PLAP. RNAs derived

584
V Reiswich et al from small structures or rare cell types are largely underrepresented and thus potentially missed in RNA analyses. Lung was also described to produce very limited amounts of PLAP RNA but this is not supported by the findings in our IHC analysis. It is of note that the strong immunostaining of smooth muscle derived from various organs seen with clone 8A9 was not seen with MSVA-350R and is thus considered to reflect cross-reactivity. In line with this interpretation, PLAP RNA expression has previously not been described in smooth muscle cells [5,64,65]. Based on these data, the antibody MSVA-350R was solely used for our tumor tissue analyses. Studies using clone 8A9 have previously described PLAP expression in leiomyoma [49], leiomyosarcoma [49], and angiomyolipoma of the kidney [49,57]. As we did not find any PLAP staining in a total of 223 tumors of these categories, it appears certain that earlier results were caused by antibody cross-reactivity and not by true PLAP expression.
The successful analysis of PLAP expression in 12,381 cancers of 131 different tumor types and subtypes confirmed a high frequency of PLAP expression in testicular tumors but also showed that frequent and high-level PLAP immunostaining occurs in various other tumor types, most commonly derived from the female genital tract, the gastroesophageal, and the pancreaticobiliary system. Our findings observed for seminomas (99%), embryonal carcinoma (97%), and yolk sac tumors (75%) of the testis are largely consistent with the literature [4,[6][7][8][9]12,16,[19][20][21][22][23]26,30,33,35,36,38,41,42]. That the highest PLAP positivity rates in extra-testicular cancers were found in tumors of the female genital tract  . Graphical comparison of PLAP data from this study (x) in comparison with the previous literature (dots). Red: n = 1-9, orange: n = 10-50, green: n > 51. For comparison purposes, studies that did not differentiate between different tumor subtypes were marked with blue dots and the overall positivity rate was applied to the different tumor subtypes present in our tumor microarrays. All studies are referred to in the reference list.

586
V Reiswich et al fits well with the distribution of PLAP expression in normal tissues and also with earlier studies. Several authors have previously described variable levels of PLAP expression in high-grade serous carcinomas [35,43,45,52,54,59], endometroid carcinomas [4,35,45,52], and other variants [4,35,52,58] of ovarian cancer as well as in endometrial cancer [4,35]. Adenocarcinomas of the stomach and of the esophagus were also among the commonly PLAP-positive tumors. Previous studies have reported 67% PLAP positivity in a study on 6 adenocarcinomas of the esophagus [4] and in 38% of 8 [4], 23% of 107 [48], and in 0 of 2 gastric adenocarcinomas [35]. Moreover, the TCGA database described elevated PLAP expression in 60% of 354 gastric adenocarcinomas [1]. From a diagnostic point of view, it is important to keep in mind that very high PLAP expression levels, which are often considered characteristic for germ cell tumors, predominated in germ cell tumors but also occurred in multiple additional tumor entities. These includedin addition to those mentioned abovefurther clinically important and frequent cancer types such as adenocarcinoma of the lung, urothelial cancer, colorectal adenocarcinoma as well as mucoepidermoid carcinoma of salivary glands. It is also noteworthy that weak PLAP expression limited to a small subset of tumor cells can occur in a wide variety of tumor entities and must not be viewed as a strong argument for the germ cell origin of a cancer. It was not within the scope of our study to analyze molecular mechanisms and functional consequences of PLAP expression in these cancers. However, the availability of clinicopathological data for some of the tumor entities that expressed PLAP in a significant fraction of cases enabled an analysis of the potential clinical significance of PLAP expression. Finding a link between PLAP upregulation and colon cancer aggressiveness supports the concept of targeting PLAP in colon cancers [66]. That the respective findings were inverse between endometrial and colorectal cancer might suggest that the tumor biologic role of PLAP expression can vary between tumor entities.
The data from this study provide a comprehensive ranking list of tumors according to their PLAP expression across a large variety of tumor entities. It is a strongpoint of our study that all tissues were stained in 1 day under exactly the same experimental conditions and that one expert pathologist interpreted all immunostains, resulting in as much standardization as possible. It is almost certain that the use of different protocols, antibodies, interpretation criteria, and thresholds used to define 'positivity' have jointly caused the high diversity of literature data on PLAP expression in cancer (summarized in Figure 4). The frequencies described in this study are thus specific to the reagents and protocols used in our laboratory. It is expected that different experimental conditions would have changed the PLAP positivity ratesespecially in the cancers with low expression levelsbut would have little impact on the tumor ranking based on the PLAP positivity rates.
In summary, our data show that PLAP can be highly expressed in a variety of tumor types. Besides germ cell tumors, which show the highest PLAP expression prevalence, high-level PLAP expression can be found in cancers from the female genital tract, the gastroesophageal, and the pancreaticobiliary system as well as in a few other tumor types. Low-level PLAP expression can be found in various other tumor entities and should generally not be viewed as a strong argument for germ cell neoplasia.