Apolipoprotein D Expression in Primary Brain Tumors: Analysis by Quantitative RT-PCR in Formalin-fixed,Paraffin-embedded Tissue

Abstract

Apolipoprotein D (apoD) expression has been shown to correlate both with cell cycle arrest and with prognosis in several types of malignancy, including central nervous system astrocytomas and medulloblastomas. ApoD expression was investigated by real-time quantitative RT-PCR using RNA extracted from 68 formalin-fixed, paraffin-embedded brain specimens. Glyceraldehyde phosphate dehydrogenase was used as an internal control. Quantitation was achieved on all specimens. Sixteen poorly infiltrating WHO grade I glial neoplasms (i.e., pilocytic astrocytomas and gangliogliomas) showed an average 20-fold higher apoD expression level compared with the 20 diffusely infiltrating glial neoplasms (i.e., glioblastoma, anaplastic astrocytoma, oligodendrogliomas; p=0.00004). A small number of exceptions (i.e., two high-expressing glioblastomas and three low-expressing gangliogliomas) were identified. Analyzed as individual tumor groups, poorly infiltrating grade I pilocytic astrocytomas and gangliogliomas differed significantly from each tumor type within the diffusely infiltrating higher-grade category (p<0.05 for each comparison) but not from each other (p>0.05). Conversely, each individual tumor type within the diffusely infiltrating category differed significantly from both pilocytic astrocytomas and gangliogliomas (p<0.05) but did not vary from other infiltrating tumors (p>0.05). Ependymomas, non-infiltrating grade II neoplasms, expressed levels of apoD similar to or lower than levels expressed by the diffusely infiltrating gliomas. Ten medulloblastomas with survival longer than 3 years averaged slightly higher apoD expression than four fatal medulloblastomas; however, this result was not statistically significant and individual exceptions were notable. In 17 of the medulloblastomas, MIB-1 proliferation rates quantitated by image cytometry did not correlate with apoD expression. In addition, apoD expression was 5-fold higher in the slowly proliferating grade I glial neoplasms compared with non-proliferating normal brain tissue (p=0.01), suggesting that apoD expression is not simply an inverse measure of proliferation. ApoD expression measured by quantitative RT-PCR may be useful in the differential diagnosis of primary brain tumors, particularly pilocytic astrocytomas and gangliogliomas.

Keywords

brain tumors apolipoprotein D PCR astrocytoma medulloblastoma

Apolipoprotein D (apoD) is a member of the lipocalin family of genes involved in the transport of small hydrophobic molecules (Flower 1994; Rassart et al. 2000). Its production by many organs other than the liver and intestine makes it unique among the lipoproteins. It is the most abundant protein found within the cyst fluid of benign fibrocystic disease of the breast (Balbin et al. 1990), and it is found in high concentration in high-density lipoproteins in serum. In senescent fibroblasts, as well as hormonally treated breast and prostate carcinoma cell lines, high apoD levels correlate with cell cycle arrest (Provost et al. 1991a; Simard et al. 1991; Lopez-Boado et al. 1994; Sugimoto et al. 1994). By immunohistochemistry, apoD protein has been reported to correlate with prognosis in several different types of carcinoma (Diez-Itza et al. 1994; Vazquez et al. 2000).

Some of the highest levels of apoD expression are found within the brain, where apoD is present within astrocytes, oligodendrocytes, and certain populations of neurons (Smith et al. 1990; Provost et al. 1991b; Seguin et al. 1995; Navarro et al. 1998; del Valle et al. 2003). In a recently published expression microarray study involving 60 medulloblastomas, apoD was one of the four markers most useful in predicting the prognosis of medulloblastomas (Pomeroy et al. 2002). We recently reported that apoD expression, measured by immunohistochemistry and in situ hybridization, differed in different categories of brain tumors (Hunter et al. 2002). However, immunohistochemistry was associated with both false-positive and negative results, almost certainly because apoD is a small secreted protein that leaves the cell and is found in high concentrations in the extracellular space and plasma. In situ hybridization was subjective and more difficult to interpret.

Recently, real-time RT-PCR techniques have been developed that allow quantitation of RNA levels using formalin-fixed, paraffin-embedded tissue samples (Lehmann and Kreipe 2001; Specht et al. 2001). Because apoD RNA, unlike apoD protein, remains localized within cells, we used quantitative RT-PCR to investigate apoD expression in 60 primary human brain tumors of different histologic types and eight samples of non-neoplastic brain tissue. MIB-1 proliferative rates were simultaneously quantitated on 16 medulloblastomas by image cytometry. According to our experience, quantitative apoD RT-PCR was easier to interpret and less subjective than either immunohistochemistry or in situ hybridization.

Materials and Methods

Tissue

Sixty primary brain tumors and eight samples of non-neoplastic brain tissues were selected from the archives of the Emory University Department of Pathology. The diagnoses, as listed in Table 1, were confirmed by review of hematoxylin and eosin (H and E)-stained sections.

RNA Extraction

Areas for study were selected in H and E-stained formalin-fixed, paraffin-embedded tissue sections. Cases with significant admixtures of non-neoplastic tissue, such as low-grade infiltrating astrocytomas, were avoided. Selected areas in adjacent non-stained sections that had been deparaffinized were then scraped into 1.5 ml microcentrifuge tubes. Two hundred μl of lysis buffer (10 mM Tris HCL, pH 8.0; 100 mM EDTA, and 2% SDS) and 20 μl of 20 mg/ml proteinase K (Invitrogen; Carlsbad, CA) were added. The mixture was incubated at 60C and 50 rpm in a thermomixer (Model R; Eppendorf) for 16 hr. Total RNA was then extracted using Qiagen's total RNA kit (Qiagen; Valencia, CA). Briefly, 600 μl of RLT solution was added to the proteinase-digested sample and centrifuged at 14,000 rpm in an Eppendorf centrifuge for 5 min to pellet white precipitates. The supernatant was passed through a Qiashredder (Qiagen) by a centrifugation of 14,000 rpm for 2 min. An equal volume (750 μl) of 70% alcohol was added to the flow-through to precipitate nucleic acids and the mixture was applied onto an RNeasy column. Total RNA bound to the column was washed once with 700 μl of RW1 and twice with 500 μl of PRE and then eluted in 100 μl of RNAase/DNAase-free water.

Table 1

Apolipoprotein D expression in primary brain tumors

Diagnosis	Number of cases	Average ΔCta	Rangea	Difference with normal tissuec
Normal brain	8	+0.8 ± 1.8	-0.1 to 3.9	0.0
Poorly infiltrating gliomasb	16	+3.3 ± 1.7	-0.2 to 5.6	+5.0-fold (p=0.001)
Pilocytic astrocytoma	8	+3.7 ± 1.2	2.3 to 5.5	+6.4-fold (p=0.01)
Ganglioglioma	8	+2.8 ± 2.1	-0.2 to 5.6	+3.6-fold (p=0.04)
Diffusely infiltrating gliomasb	20	-1.0 ± 2.6	-9.1 to 2.7	-3.2-fold (p=0.03)
Oligodendroglioma	3	-0.5 ± 1.5	-1.8 to 1.1	Non-significant
Anaplastic astrocytoma	6	+0.1 ± 0.7	-0.9 to 0.7	Non-significant
Anaplastic oligodendroglioma	1	-1.9
Glioblastoma	10	-1.7 ± 3.6	-9.1 to 2.7	-5.0-fold (p=0.05)
SEGA	2	-3.3 ± 1.1	-2.5 to −4.0	-13.9-fold (p=0.01)
Ependymoma	4	-2.4 ± 0.7	-1.6 to −3.0	-7.8-fold (p=0.001)
Medulloblastoma	19	-0.4 ± 2.2	-7.0 to 3.1	Non-significant
Medulloblastomasb (dead)	4	-1.8 ± 4.0	-7.0 to 2.1
Medulloblastomasb (alive)	15	+0.3 ± 1.7	-2.1 to 3.1

Cycles above GAPDH control.

No significant difference between fatal and non-fatal medulloblastomas

Normal brain consists of a mixture of neurons, astrocytes, oligodendrocytes, microglia, and ependymal cells. Consequently, comparisons with normal brain are not comparisons with the non-neoplastic cell of origin of the gliomas. Student's t-test assuming unequal variance.

GAPDH, glyceraldehyde phosphate dehydrogenase; SEGA, subependymal giant cell astrocytoma.

Real-time RT-PCR Assay

The real-time assay was performed using Qiagen's one-step RT-PCR kit. Twenty μl of the RT-PCR reagent was added to 5 μl of tissue RNA. The 25 μl assay contained 1x one-step RT-PCR buffer (Qiagen), 200 nM dNTP, and 200 nM each of apoD forward primer (5’ gcctgccaagctggaagtt), apoD reverse primer (5’ ggccaggatccagtacggt), apoD probe (5’ Hex agttttcctggtttatgccatcgg Qsy-7), 1.25 μl of a housekeeping gene, glyceraldehyde phosphate dehydrogenase (GAPDH) primers and probe mixture (Roche; Indianapolis, IN), and 200 nM of Rox standard II dye (Synthegen; Houston, TX). The GAPDH probe was labeled with Fam (5') and Tamra (3'). Both of the apoD and GAPDH detection systems were mRNA specific because their forward and reverse primers were derived from two different exons. Rox was used as a passive dye for assay volume normalization. RT-PCR was carried out at 50C for 30 min, 95C for 15 min, and 50 cycles of 95C for 15 sec and 60C for 30 sec. The amplification efficiencies of the apoD and GAPDH were about the same (90%). Threshold cycle (Ct) values were obtained from the ABI 7700 Sequence Detection software (Applied Biosystems; Foster City, CA) by adjusting the baseline level 3-fold, the value recommended by the software. A large majority of the data points in this study were in the high twenties cycle range. Some data points were in the low thirties cycle range and none was greater than 40. Linearity over the 20- to 40-cycle range was established by standard dilution curves (data not shown). Selected data points, including all high apoD-expressing glioblastomas and medulloblastomas, were repeated without a significant change in value. In previous experiments, coefficients of variation for triplicate data points using the ApoD RT-PCR assay ranged between 2 and 20%.

ApoD and GAPDH RNA levels were measured in the same reaction vessel with the Hex and Fam probes, respectively. The index (ΔCt) of ApoD RNA was determined by the difference of the dCts of GAPDH and ApoD (ΔCt = Ct of GAPDH - Ct of ApoD). When expressed this way, a sample having a higher ΔCt index expressed a higher level of ApoD RNA than another sample with a lower ΔCt index by 1.9ΔΔCt fold (ΔΔCt is the difference of the ΔCt indexes of the two samples, and 1.9 reflected the amount of amplicon produced per amplification cycle with an amplification efficiency of 90%). For instance, if the ΔCts of two samples are 3 and −3, then the ΔΔCt is 6 and the first sample should express 1.9² or 47-fold higher apoD than the second sample (Livak and Schmittgen 2001).

Immunohistochemistry

Sections of formalin-fixed, paraffin-embedded tissue (5 μm) were tested for the presence of MIB-1 (1:50) (Immunotech/Colter; Westbrook, ME) using an avidin-biotin-complex technique and steam-induced antigen retrieval. An avidinbiotinylated enzyme complex kit (DAKO LSAB2; DAKO, Carpinteria, CA) was used in combination with the automated DAKO autostainer. Sectons were deparaffinized and rehydrated, then steamed in citrate buffer (pH 6) for 20 min, and cooled for 10 min before immunostaining. All tissues were exposed to 3% hydrogen peroxide for 5 min, streptavidin enzyme complex for 25 min, biotinylated secondary linking antibody for 5 min, and Richard Allen Scientific hematoxylin counterstain for 1 min. Between incubations, sections were washed with Tris-buffered saline.

Image cytometric nuclear MIB-1 quantitation was performed using the Automated Cellular Imaging System (ACIS; ChromaVision Medical Systems, San Juan Capistrano, CA). The ACIS uses automated bright field microscopy imaging to detect, classify, and count stained cellular objects based on predetermined color and morphology. Ten regions of each specimen were randomly selected for analysis.

Statistical Methods

Significant differences for apoD expression were calculated with Student's t-test assuming unequal variance. The levels of apoD and MIB-1 for each sample were tabulated and one-way ANOVA tests were performed with a Type I error rate of 0.05 set a priori. Tukey-Kramer HSD pairwise tests were performed between groups to investigate the diagnostic potential for apoD levels and its relationship with MIB-1 levels. Statistical analyses were performed using the SAS JMP software, version 5.0 (SAS Institute Inc; Cary, NC).

Results

All samples tested, even very small tissue aspirates (CUSA washings), yielded quantitative results presented in Table 1, Figure 1, and Figure 2. Normal brain specimens (including four samples of cerebellum and four samples of cerebral cortex with white matter) averaged 0.8 cycles more than the GAPDH control. Although the cerebellar samples averaged ∼1 cycle more than the supratentorial samples, this difference was not statistically significant. It should be kept in mind that normal brain tissue includes a large variety of different cell types, only ∼10% of which are astrocytes. Hence, a comparison between astrocytomas and normal brain is not a comparison between non-neoplastic and neoplastic astrocytes, but rather largely a comparison between different cell types. In this study, normal brain tissue was used strictly as a reference level for purposes of comparison.

ApoD expression was significantly increased in poorly infiltrating WHO grade I brain tumors, i.e., pilocytic astrocytomas and gangliogliomas, compared with the normal brain tissue reference. The pilocytic astrocytomas averaged a ΔCt of 3.7, 6.4-fold higher than normal brain (p = 0.01), and all eight of the pilocytic astrocytomas were at least 2.6-fold above normal brain. Gangliogliomas averaged a ΔCt of 2.8, 3.6-fold higher than normal brain (p = 0.04); however, three of the gangliogliomas showed levels of expression similar to or lower than normal brain. Low apoD levels in one of the gangliogliomas (apoD = 0.3) may have been the result of dilution by a significant lymphocytic infiltration in this tumor; however, low apoD levels in two other gangliogliomas (apo D = −0.2 and 1.2) could not be explained by this mechanism.

Figure 1.

Black squares represent individual data points. Horizontal bar within each rectangle represents the mean of each group. Horizontal bars at the upper and lower ends of each rectangle represent 25% confidence intervals (1 SD). Horizontal bars outside the rectangles represent 75% confidence intervals (2 SD). EP, ependymoma; GBM, glioblastoma; MB, medulloblastoma; GG, ganglioglioma; PA, pilocytic astrocytoma.

In contrast with the poorly infiltrating grade I tumors, diffusely infiltrating primary brain tumors showed decreased apoD levels compared with normal brain. ApoD levels in 10 glioblastomas (grade IV infiltrating astrocytomas) averaged a ΔCt of −1.7, 5.0-fold less than normal brain (p = 0.05). Taken together, all infiltrating glial neoplasms (i.e., GBM, AA, and oligodendrogliomas) averaged a ΔCt of −1.0, 3.2-fold less than normal brain (p = 0.03).

Two individual cases of glioblastomas showed higher apoD expression levels than normal brain. Interestingly, one of these glioblastomas (ΔCt = 2.72) was being radiated at the time of biopsy, and the other glioblastoma (ΔCt = 2.05) had finished a course of radiation and 1,3-bis(2-chloroethyl)-1-nitrosourea (BNCU) chemotherapy 4 months prior to biopsy. The first patient survived 5 months after onset of symptoms, whereas the second patient survived two and one half years following the onset of symptoms. Only two other tumors in this series had received therapy, an anaplastic astrocytoma (ΔCt = −0.94) and an anaplastic mixed oliogastrocytoma (ΔCt = −2.46). These tumors had been radiated 4 and 5 years prior to biopsy, respectively.

Figure 2.

ApoD expression in primary brain tumors. Each individual diagnostic group differed statistically in mean apoD value from all diagnostic groups in the opposite oval (p<0.05) but did not differ from diagnostic groups in the same oval (p>0.05). PA, pilocytic astrocytoma; GG, ganglioglioma; MB, medulloblastoma; AA, anaplastic astrocytoma; GBM, glioblastoma; EP, ependymoma.

Comparison of apoD expression between the 8 pilocytic astrocytomas and the 20 infiltrating glial tumors showed an average 20-fold difference that was highly significant (p = 0.00001). Only a single glioblastoma (ΔCt = 2.72, in the process of being radiated at the time of the biopsy) showed a higher level of apoD expression than the two pilocytic astrocytomas showing the lowest levels of expression.

Although relatively few cases were studied, apoD expression levels significantly less than normal brain were seen in four grade II ependymomas (p = 0.001) and two grade I subependymal giant cell astrocytomas (p = 0.01). These results suggest that apoD expression may be low in these cell types. Because both of these are non-infiltrating tumor types, apoD levels do not appear to correlate strictly with infiltrative behavior.

ApoD expression levels were quantitated in 19 medulloblastomas, all of which were untreated at the time of biopsy but subsequently received radiation and/or chemotherapy (Table 2). A broad spectrum of apoD ΔCt levels ranging between −7.0 and +3.1 was obtained. The average apoD expression level (-0.4) was slightly less than normal brain, a result that did not reach statistical significance (p=0.11). Four medulloblastomas were fatal at intervals ranging from 1 to 4 years following diagnosis. Fourteen patients were still alive at the time of this study, 10 of these having survived for at least 3 years. Surviving patients averaged 1.7-fold (0.8 cycles) higher than dead patients; however, this result did not reach statistical significance (p = 0.36). In addition, two of the dead medulloblastoma patients had apoD expression levels greater than the average medulloblastoma expression level (ΔCt = 1.1 and 2.1), and half, i.e., seven, of the surviving patients had apoD levels lower than the average expression level. These results suggest that, on an individual case basis, apoD expression is not a reliable marker of prognosis in medulloblastomas.

Table 2

Apolipoprotein D expression in medulloblastomas

Age/Sex	ApoD ΔCt	MIB-1	Survival status (post-biopsy)
Neuroblastoma
2F	3.1	7%	Alive (6 years)
11F	2.6	26%	Alive (8 years)
Desmoplastic
2M	2.1	Not done	Dead (8 months)
8F	1.2	34%	Alive (1 year)
9F	1.1	18%	Unknown
25M	1.1	36%	Dead (3 years)
3M	0.7	44%	Alive (4 years)
14M	0.5	34%	Alive (1 year)
15M	0.4	21%	Alive (1.5 years)
11M	0.2	Not done	Alive (3 years)
11M	-0.7	26%	Alive (1.5 years)
Medullomyoblastoma
2M	-0.9	17%	Dead (3 years)
7M	-1.1	26%	Alive (3.5 years)
11M	-1.3	28%	Alive (4 years)
8M	-1.5	16%	Alive (5 years)
9M	-1.9	37%	Alive (4.5 years)
8M	-2.1	18%	Alive (5 years)
4M	-2.1	40%	Alive (3 years)
2F	-6.95	25%	Dead (1 year)

Because we encountered a spectrum of apoD expression levels in medulloblastomas, we investigated the relationship between apoD expression and proliferation in this group of tumors. MIB-1 proliferation rates were measured in 16 of the medulloblastomas, and the results were correlated with apoD expression. No statistically significant correlation was found. This result supports the conclusion that apoD expression levels are not simply an inverse marker of cell proliferation.

Discussion

In the past, formalin-fixed, paraffin-embedded specimens have been considered a poor source of RNA because the RNA in these specimens is degraded (i.e., cut it into smaller fragments), and proteins are extensively cross-linked. In our RT-PCR study, quantitative results were obtained for all 68 formalin-fixed, paraffin-embedded specimens, amply demonstrating the capacity of optimized RT-PCR techniques for the study of formalin-fixed brain tumors. The RT-PCR protocol used in this study overcomes the problems of formalin fixation primarily by 1. use of small amplicons (i.e., <100 base pairs) and 2. use of a prolonged (i.e., 16 hr) proteinase digestion.

Small, secreted molecules, such as apoD or albumen, may be difficult to assess by immunohistochemistry. High background and false positive results may occur because of high levels of protein in plasma and extracellular fluid. Conversely, false negative results may occur because the protein is rapidly secreted out of the cell in which it is produced. Albumen RNA, for example, is considered to be an excellent marker for hepatocytes; however, albumen protein immunohistochemistry is not useful for the identification of hepatocytes.

The quantitative RT-PCR method used here for apoD shows several advantages compared with our previous results using immunohistochemistry and in situ hybridization to assess apoD (Hunter et al. 2002). In the current study, no false negative results were obtained in the pilocytic astrocytoma group, and false positive tests related to the presence of apoD protein in serum and extracellular fluid were avoided. Quantitative PCR was successful even on the smallest tissue samples, such as aspirates, needle biopsies, or CUSA washings, i.e., the tissue samples most likely to give false negative results by immunohistochemistry (Hunter et al. 2002). Because the PCR method is quantitative, subjective interpretations, which we found to be a problem with both apoD immunohistochemistry and in situ hybridization, are avoided. Broader experience with both apoD immunohistochemistry and quantitative PCR will be needed to fully evaluate the utility of each of these techniques.

This study confirms the potential utility of apoD in the differential diagnosis of primary brain tumors, particularly in the distinction of pilocytic astrocytomas and some cases of ganglioglioma from higher-grade infiltrating gliomas. Distinction of WHO grade I pilocytic astrocytomas and gangliogliomas from the biologically distinct diffusely infiltrating glial neoplasms (i.e., WHO grades II-IV anaplastic astrocytoma, glioblastoma, oligodendroglioma) is of critical importance. Poorly infiltrating grade I tumors are curable in many cases by surgical removal and rarely progress to a higher grade. In contrast, the diffusely infiltrating gliomas are virtually always fatal and commonly progress to a higher grade. Furthermore, in some cases histologic features in poorly infiltrating gliomas, such as degenerative nuclear atypia, high cellularity, and microvascular proliferation, may closely mimic high-grade infiltrating gliomas.

The prognosis for medulloblastoma, an embryonal tumor of the cerebellum, is variable and poorly predictable (Packer et al. 1999). With current treatment modalities, ∼20% to 30% of these malignant neoplasms are fatal within several years. The 5-year survival of these patients is ∼50% to 70%, and long-term survivals are not uncommon (Chojnacka and Skowronska-Gardas 2004). In a recent expression microarray analysis of 60 medulloblastomas, apoD was identified as one of the four best predictors of outcome in this tumor type (Pomeroy et al. 2002). Consequently, we were particularly interested in studying the relationship between apoD expression and prognosis in medulloblastomas.

Our four fatal cases of medulloblastoma showed 1.7-fold less apoD expression compared with 10 surviving cases, a difference that was not statistically significant. Because only 18 cases were studied, it is difficult to draw firm conclusions from this result. However, our data do suggest that on an individual case basis apoD expression may not be a reliable prognosis marker in medulloblastoma. One half of the surviving cases showed apoD levels lower than average, and one half of the fatal cases showed apoD levels higher than average. It is possible that apoD expression will prove useful in predicting prognosis of medulloblastomas if used in conjunction with other markers or if larger series of cases are studied.

Our results are consistent with the premise that apoD levels correlate with cell cycle arrest, but suggest that other factors also contribute to apoD expression levels. Pilocytic astrocytomas and gangliogliomas (known to have low MIB-1 proliferative rates) showed high apoD expression. Conversely, high-grade astrocytomas (known to be associated with high MIB-1 proliferative rates) showed lower levels of apoD expression. Interestingly, the two glioblastomas with aberrantly high apoD expression were either in the process of being treated or recently treated with radiation or chemotherapy, treatments known to inhibit cell proliferation. Nevertheless, when we compared MIB-1 proliferation rates with apoD expression in medulloblastomas, no correlation could be demonstrated. Furthermore, pilocytic astrocytomas and gangliogliomas (tumors showing low but definite proliferation) showed significantly higher apoD expression than normal brain, which is non-proliferative. Surprisingly low apoD expression levels were found in two subependymal giant cell astrocytomas (SEGAs), hamartomatous neoplasms known to show low MIB-1 proliferation rates. Taken together, these results suggest that apoD is not simply an inverse measure of proliferation.

In summary, the results of this quantitative RT-PCR study demonstrate that 1. apoD expression levels can be quantitated in formalin-fixed, paraffin-embedded tissue samples; 2. apoD expression levels are elevated in pilocytic astrocytomas and gangliogliomas and may prove to be useful in the differential diagnosis of these tumors; 3. on an individual case basis, apoD expression levels may not be a very useful marker of therapeutic response in medulloblastomas; and 4. apoD expression levels are not simply an inverse measure of proliferation.

References

Balbin

Freije

Fueyo

Sanchez

Lopez-Otin

(1990) Apolipoprotein D is the major protein component in cyst fluid from women with human breast gross cystic disease. Biochem J 271:803–807.

Chojnacka

Skowronska-Gardas

(2004) Medulloblastoma in childhood: impact of radiation technique upon the outcome of treatment. Pediatr Blood Cancer 42:155–160.

del Valle

Navarro

Astudillo

Tolivia

(2003) Apolipoprotein D expression in human brain reactive astrocytes. J Histochem Cytochem 51:1285–1290.

Diez-Itza

Vizoso

Merino

Sanchez

Tolivia

Fernandez

Ruibal

(1994) Expression and prognostic significance of apolipoprotein D in breast cancer. Am J Pathol 144:310–320.

Flower

(1994) The lipocalin protein family: a role in cell regulation. FEBS Lett 354:7–11.

Hunter

Young

Olson

Brat

Bowers

Wilcox

Jaye

(2002) Differential expression between pilocytic and anaplastic astrocytomas: identification of apolipoprotein D as a marker for low-grade, non-infiltrating primary CNS neoplasms. J Neuropathol Exp Neurol 61:275–281.

Lehmann

Kreipe

(2001) Real-time PCR analysis of DNA and RNA extracted from formalin-fixed and paraffin-embedded biopsies. Methods 25:409–418.

Livak

Schmittgen

(2001) Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 25:402–408.

Lopez-Boado

Tolivia

Lopez-Otin

(1994) Apolipoprotein D gene induction by retinoic acid is concomitant with growth arrest and cell differentiation in human breast cancer cells. J Biol Chem 269:26871–26878.

10.

Navarro

Tolivia

Astudillo

del Valle

(1998) Pattern of apolipoprotein D immunoreactivity in human brain. Neurosci Lett 254:17–20.

11.

Packer

Goldwein

Nicholson

Vezina

Allen

Ris

Muraszko

(1999) Treatment of children with medulloblastomas with reduced-dose craniospinal radiation therapy and adjuvant chemotherapy: A Children's Cancer Group Study. J Clin Oncol 17:2127–2136.

12.

Pomeroy

Tamayo

Gaasenbeek

Sturla

Angelo

McLaughlin

Kim

(2002) Prediction of central nervous system embryonal tumour outcome based on gene expression. Nature 415:436–442.

13.

Provost

Marcel

Milne

Weech

Rassart

(1991a) Apolipoprotein D transcription occurs specifically in nonproliferating quiescent and senescent fibroblast cultures. FEBS Lett 290:139–141.

14.

Provost

Villeneuve

Weech

Milne

Marcel

Rassart

(1991b) Localization of the major sites of rabbit apolipoprotein D gene transcription by in situ hybridization. J Lipid Res 32:1959–1970.

15.

Rassart

Bedirian

Do Carmo

Guinard

Sirois

Terrisse

Milne

(2000) Apolipoprotein D. Biochim Biophys Acta 1482:185–198.

16.

Seguin

Desforges

Rassart

(1995) Molecular characterization and differential mRNA tissue distribution of mouse apolipoprotein D. Brain Res Mol Brain Res 30:242–250.

17.

Simard

Veilleux

de Launoit

Haagensen

Labrie

(1991) Stimulation of apolipoprotein D secretion by steroids coincides with inhibition of cell proliferation in human LNCaP prostate cancer cells. Cancer Res 51:4336–4341.

18.

Smith

Lawn

Wilcox

(1990) Cellular localization of apolipoprotein D and lecithin:cholesterol acyltransferase mRNA in rhesus monkey tissues by in situ hybridization. J Lipid Res 31:995–1004.

19.

Specht

Richter

Muller

Walch

Werner

Hofler

(2001) Quantitative gene expression analysis in microdissected archival formalin-fixed and paraffin-embedded tumor tissue. Am J Pathol 158:419–429.

20.

Sugimoto

Simard

Haagensen

Labrie

(1994) Inverse relationships between cell proliferation and basal or androgen-stimulated apolipoprotein D secretion in LNCaP human prostate cancer cells. J Steroid Biochem Mol Biol 51:167–174.

21.

Vazquez

Gonzalez

Merino

Vizoso

(2000) Expression and clinical significance of apolipoprotein D in epithelial ovarian carcinomas. Gynecol Oncol 76:340–347.