Mcl-1 expression in gestational trophoblastic disease correlates with clinical outcome

A differential expression study

Authors

  • Pui-Yee Fong M.Phil.,

    1. Department of Pathology, Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Queen Mary Hospital, Hong Kong, China
    Search for more papers by this author
  • Wei-Cheng Xue M.D,

    1. Department of Pathology, Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Queen Mary Hospital, Hong Kong, China
    Search for more papers by this author
  • Hextan Y. S. Ngan M.D.,

    1. Department of Obstetrics and Gynaecology, Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Queen Mary Hospital, Hong Kong, China
    Search for more papers by this author
  • Kelvin Y. K. Chan Ph.D.,

    1. Department of Pathology, Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Queen Mary Hospital, Hong Kong, China
    2. Department of Obstetrics and Gynaecology, Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Queen Mary Hospital, Hong Kong, China
    Search for more papers by this author
  • Ui-Soon Khoo M.D.,

    1. Department of Pathology, Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Queen Mary Hospital, Hong Kong, China
    Search for more papers by this author
  • Siu-Wah Tsao Ph.D.,

    1. Department of Anatomy, Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Queen Mary Hospital, Hong Kong, China
    Search for more papers by this author
  • Pui-Man Chiu M.Med.Sc.,

    1. Department of Pathology, Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Queen Mary Hospital, Hong Kong, China
    Search for more papers by this author
  • Lai-Shan Man M.Phil.,

    1. Department of Pathology, Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Queen Mary Hospital, Hong Kong, China
    Search for more papers by this author
  • Annie N. Y. Cheung M.D.

    Corresponding author
    1. Department of Pathology, Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Queen Mary Hospital, Hong Kong, China
    • Department of Pathology, The University of Hong Kong, Queen Mary Hospital, Pokfulam Road, Hong Kong, China
    Search for more papers by this author
    • Fax: (011)852 28725197


Abstract

BACKGROUND

Hydatidiform moles (HMs) are abnormal pregnancies with a propensity for developing persistent disease in the form of gestational trophoblastic neoplasia (GTN), which requires chemotherapy. In previous studies, the authors demonstrated that low apoptotic activity was correlated with the progression of HM to GTN, and they hypothesized that some apoptosis-related genes may determine this progression.

METHODS

The differential expression of apoptotic genes in HMs that subsequently developed into GTN was compared with the same expression in HMs that spontaneously regressed using a human apoptosis array; then, the expression was evaluated with real-time quantitative polymerase chain reaction analysis and immunohistochemistry using 54 clinical samples from patients with HMs who had follow-up data available.

RESULTS

Using an apoptosis array, greater expression of Mcl-1, which is an antiapoptotic gene, was detected in HMs that subsequently developed into GTN. It was confirmed that the levels of Mcl-1 RNA expression (P = 0.017) and Mcl-1 protein expression (P < 0.001) in HMs that developed into persistent disease and required chemotherapy were significantly greater compared with the levels in HMs that regressed. Moreover, Mcl-1 immunoreactivity, which was detected predominantly in cytotrophoblasts, was correlated with the apoptotic index, as assessed with M30 cytoDeath immunohistochemistry, which is a good indicator of apoptotic events in the early-stage disease.

CONCLUSIONS

The current results demonstrated that Mcl-1, as identified by a cyclic DNA array, may play a role in the pathogenesis of HMs and may have potential as a useful marker for predicting the clinical behavior of HMs. Cancer 2005. © 2004 American Cancer Society.

Gestational trophoblastic diseases encompasses a heterogeneous set of diseases that arise from abnormal trophoblast tissue, including hydatidiform mole (HM), invasive mole, choriocarcinoma, placental site trophoblastic tumor, and epithelioid trophoblastic tumor.1–3 HMs can be classified further into complete HMs (CMs) and partial HMs (PMs). It is a unique disease that arises from allograft of conceptus with potential for local invasion and widespread metastasis, and the degree of aggressiveness varies in different types of moles. Although most HMs spontaneously regress after suction evacuation, approximately 8–30% of patients with HM will develop persistent gestational trophoblastic neoplasia (GTN) requiring chemotherapy.1–4 Except for the serial assays of serum and urine human chorionic gonadotrophin (hCG) levels, there is no reliable predictor for the early detection of GTN.

Cyclic DNA (cDNA) arrays, which are available from commercial companies and human genome centers, allow the simultaneous analysis of various genes involved in human carcinogenesis, cell cycle and transcription regulation, DNA repair and synthesis, apoptosis, and signaling of growth factors and receptors. The differential expression profiles of multiple genes can then be identified, leading to a better understanding of the disease.5–10

In our previous studies, we demonstrated that telomerase and apoptotic activities in HM were associated with the development of GTN.11–13 In 2 independent studies on apoptotic activity, as assessed by the transferase (TdT)-mediated deoxyuridine triphosphate (dUTP) nick-end labeling (TUNEL) approach and the M30 cytoDeath antibody, patients who had HM that spontaneously regressed had significantly higher apoptotic indices compared with patients who had HM that developed into persistent disease requiring chemotherapy.12, 13

We hypothesized that some apoptosis-related genes may determine such disease progression. In the current study, we attempted to identify differentially expressed, apoptosis-related genes in HM that subsequently developed into GTN compared with HM that spontaneously regressed using the Atlas™ Apoptosis Array (Clontech, Palo Alto, CA). Confirmatory studies using relatively large numbers of clinical samples and correlations with clinicopathologic and biologic parameters were carried out to assess the biologic significance of the identified genes. To our knowledge, this is the first differential expression array study in which patients who had HMs that progressed favorably were compared directly with patients who had HMs that progressed unfavorably.

MATERIALS AND METHODS

Selection of Clinical Samples

Patients with HM usually were treated by suction evacuation after ultrasound diagnosis. Tissues were selected routinely for routine formalin fixation and paraffin embedding for pathologic diagnosis. Some tissues were snap frozen and stored at − 70 °C until RNA extraction. Approval was obtained from the Ethics Committee at the University of Hong Kong for the collection of tissues from patients with gestational trophoblastic disease.

Clinical follow-up data were retrieved in patients who had a diagnosis of HM. GTN was diagnosed if there was a plateau in the hCG level for 4 weeks or if there was a further increase in the hCG level for 3 consecutive weeks.4 Three persistent moles and three regressed moles were used for the cDNA array experiments. Thirteen HMs that regressed and 6 HMs that developed into GTN were used for confirmatory real-time reverse-transcriptase polymerase chain reaction (RT-PCR) experiments. Of the latter six patients, two patients developed metastasis to the lungs, and one patient had additional metastasis to the liver.

Archival, formalin fixed, paraffin embedded blocks of HM also were retrieved. The pathology reports from patients with a diagnosis of HM who were treated at Queen Mary Hospital, the University of Hong Kong, were reviewed. Thirteen PMs (gestational age, 8–17 weeks) and 41 CMs (gestational age, 8–28 weeks) were selected randomly for this study. The histologic features of these moles in hematoxylin and eosin-stained sections were reviewed.1–3 The diagnoses in most of these moles were confirmed further with fluorescent microsatellite genotyping after microdissection and also with chromosome in situ hybridization to analyze ploidy.14, 15 According to our follow-up data, 39 HMs spontaneously regressed after suction evacuation, whereas 15 HMs developed into GTN that required chemotherapy. The clinical and demographic data on the patients involved in the array study, real-time RT-PCR analyses, and immunohistochemical assay are summarized in Table 1.

Table 1. Clinical and Pathologic Parameters of the Patients with Hydatidiform Moles Involved in the Current Study
Patient no.Test(s)Age (yrs)DXGestational weeksFUaMetsMcl-1Mcl (%)Mcl scoreUterine size gestational ageTheca-lutein cystsPulmonary symptomsPostevacuation uterine bleeding
  • DX: diagnosis; FU: follow-up; Mets: metastasis; A: apoptosis array; CM: complete mole; ND: not determined; NA: not available; R: real-time polymerase chain reaction analysis; I: immunohistochemistry; Unk: unknown; PM: partial mole.

  • a

    Regressed (1) or developed gestational trophoblastic neoplasia (2).

1A25CM101NoNDNDNDYesNoNoYes
2A34CM92YesNDNDNDNANANANA
3A44CM62YesNDNDNDNANANANA
4A,R51CM81NoNDNDNDYesNoNoYes
5A,R28PM62NoNDNDNDNANANANA
6A,I24CM171No224NoNoNoNo
7R,I35CM151No235YesYesNoNo
8R,I45CM62Yes235YesNoUnkYes
9R,I29CM172Yes336YesNoNoNo
10R31CM91NoNDNDNDNANANANA
11R18CM111NoNDNDNDYesNoNoNo
12R16CM72NoNDNDNDYesNoNoNo
13R37CM381NoNDNDNDNoNoNoNo
14R33CM111NoNDNDNDNoNoNoNo
15R34CM131NoNDNDNDNoNoNoNo
16R33CM131NoNDNDNDNoNoNoNo
17R18CM61NoNDNDNDYesNoNoNo
18R48CM81NoNDNDNDYesNoNoNo
19R31CM101NoNDNDNDNoNoNoYes
20R43CM62NoNDNDNDNANANANA
21R48CM202NoNDNDNDNoNoNoYes
22R38CM122NNDNDNDNoYesNoNo
23R35CM192NoNDNDNDNANANANA
24I28PM141No134NANANANA
25I30PM101No134NANANANA
26I24PM151No235NANANANA
27I34CM201No112NANANANA
28I42CM101No134NANANANA
29I29CM81No112NANANANA
30I23CM121No224NANANANA
31I27PM131No213NANANANA
32I32PM141No224NANANANA
33I38PM181No224NoNoNoNo
34I39PM81No134NANANANA
35I23PM101No112NANANANA
36I25PM171No213NANANANA
37I23PM111No235NANANANA
38I24CM141No224NANANANA
39I37CM101No134NANANANA
40I25CM121No112NoNoNoYes
41I26CM141No112NANANANA
42I40CM101No112NoNoNoNo
43I18CM121No325NANANANA
44I24CM171No235NANANANA
45I27CM141No213NANANANA
46I25CM131No325NANANANA
47I17CM251No213NANANANA
48I25CM101No224NANANANA
49I28CM111No112NANANANA
50I32CM111No213NANANANA
51I30CM91No224NANANANA
52I28CM141No224NoNoNoNo
53I24PM101No235NANANANA
54I29CM121No112NoNoNoYes
55I26CM111No112NoYesNoNo
56I29CM121No235NANANANA
57I26CM101No224NoNoNoNo
58I34CM101No224NANANANA
59I24CM121No213NANANANA
60I25CM101No224NANANANA
61I26PM112No236NANANANA
62I35PM142Yes336NANANANA
63I25CM142No235NANANANA
64I25CM102No235YesNoNoNo
65I27CM262Yes235NoNoNoNo
66I45CM92No325YesNoNoNo
67I29CM102No336YesNoNoYes
68I38CM262Yes235NoNoNoYes
69I34CM82N336YesNoNAYes
70I27CM112Y336NANANANA
71I30CM122No336NANANANA
72I32CM122No235YesNoNoNo
73I31CM112No235NoNoNoYes

RNA Extraction and cDNA Array

Frozen sections were assessed histologically and microdissected to ensure that > 90% of tissues were trophoblasts before RNA extraction. Total RNA from the frozen tissues from the three regressive moles and the three persistent moles were extracted for the cDNA array experiment. Frozen tissue samples of HMs were pulverized mechanically under liquid nitrogen using Tissue Pulverizer (Bessman) to ensure no thawing before extraction. mRNA was isolated and purified using QuickPrep Micro mRNA Purification Kit (Amersham Pharmacia Biotech, Buckinghamshire, England). The differential gene expressions of mRNA from regressive and persistent moles were then compared using the commercially available Atlas™ Apoptosis Array (Clontech), which analyzed 205 cDNA fragments related to apoptosis.

Equal amounts of pooled mRNA from three regressive moles and three persistent moles were used for probe synthesis.16 Briefly, 1.5 μg of mRNA were combined with 1 μL of CDS Primer Mix (Clontech). After denaturation, Master Mix containing 2 μL 5 × reaction buffer, 1 μL 10 × 2′dioxynucleotide-S-triphosphate mix; 3.5 μL [α-33P] deoxyadinosine triphosphate (2500 Ci/mmol, 10 μCi/μL), 0.5 μL dithiothreitol (100 mM), and 1 μL Moloney murine leukemia virus reverse transcriptase were added and were then incubated in the PCR thermal cycler at 50 °C for 25 minutes. To quench the reaction, 1 μL of 10 × Termination Mix was added to each tube. The synthesized, 33P-labeled cDNAs were purified by column chromatography. The denatured probe mixture was then hybridized on the membranes overnight at 68 °C. The membranes were then washed, wrapped in plastic wrap, and exposed to X-ray film (BioMax MS film; Kodak, New York, NY) in intensifying screen at − 70 °C. Film was developed after 3 days of exposure. The images of the arrays were captured, and the intensities of the dotted genes were measured with AtlasImage™ (Clontech). Array experiments were duplicated using arrays of the same lot. The gene expression was normalized with respect to housekeeping genes: ubiquitin, hypoxanthine-guanine phosphoribosyltransferase (HPRT), liver glyceraldehyde 3-phosphate dehydrogenase (GAPDH), brain-specific tubulin α 1 subunit (TUBA1), cytoplasmic β-actin (ACTB), and 23-kDa highly basic protein; 60S ribosomal protein L13A (RPL13A) located at positions 1A, 1C, 1D, 1E, 1G, and 1H, respectively.

Real-Time RT-PCR

After histologic assessment and microdissection, if necessary, to ensure the purity of samples, total RNAs from tissues were extracted by the TRIzol® reagent (GibcoBRL) and were purified using the RNeasy® Mini Kit (Qiagen, Valencia, CA) and the RNase-free DNase Set (Qiagen). Quantification of RNA was determined by both gel electrophoresis and ultraviolet absorbance at 260 nm. All RNA samples were stored at − 70 °C until use.

According to the results of apoptotic array, real-time RT-PCR was performed to compare the RNA expression of Mcl-1 and insulin-like growth factor 2 (IGF-2) in HM that regressed and HM that developed into GTN. For each sample, 3 μg of total RNA were reverse transcribed to cDNA by Superscript™ II RNase H reverse transcriptase (GIBCO) with random hexamers and RNasin ribonuclease inhibitor (Promega, Madison, WI). The cDNA was stored at − 20 °C and was used as a template for amplification in PCR.

Primers and TaqMan probes were designed that harbored two exons, such that they were specific to the target transcript, and any detection of amplicon from genomic DNA was eliminated. Mcl-1 has two reported transcript variants resulting from alternative splicing.17 The distinct encoding isoforms have been identified. The longer gene product, Mcl-1L (isoform 1), enhances cell survival by inhibiting apoptosis, whereas the alternatively spliced, shorter gene product, Mcl-1S (isoform 2), promotes apoptosis and is death-inducing.17 The isoform that was investigated in this study was Mcl-1L (isoform 1). The assay was designed to detect the alternative splicing, longer gene product (isoform 1), Mcl-1L, and to eliminate detection of MCL-1S17: Mcl-1L (exon 2) forward primer, 5′-CAC AGA CGT TCT CGT AAG GAC AA-3′; (exon 3) reverse primer, 5′-ACA TTC CTG ATG CCA CCT TCT AG-3′; and Mcl-1L TaqMan probe, 5′-FAM-ACG GGA CTG GCT AGT TAA ACA AAG AGG CTG-TAMRA-3′; IGF-2 (exon 2) forward primer, 5′-CTT CTA CTT CAG CAG GCC CG-3′; (exon 3) reverse primer, 5′-TAG CAC AGT ACG TCT CCA GGA GG-3′; and IGF-2 TaqMan probe, 5′-FAM-CAG CCG TGG CAT CGT TGA GGA GT-TAMRA-3′. The 25-μL quantitative RT-PCR reaction included 1 × Universal Master Mix (Applied Biosystems, Foster City, CA), 340 nM of each primer, 120 nM of probe, and 2 μL of synthesized cDNA. The expression level of the housekeeping gene, TBP, as described previously,18 was measured to normalize MCL-1L and IGF-2 on the same sample. Standard curve was created to quantify the expression of Mcl-1L, IGF-2, and TBP using known concentration of serial diluted and linearized plasmid containing the corresponding gene insert. The quantitative real-time reaction was carried out and analyzed using ABI7700 Sequence Detection System (Applied Biosystems).

Immunohistochemistry

Paraffin sections (5 mm thick) were mounted on 3-aminopropyltriethoxysilane-coated slides, dewaxed, and rehydrated. Antigen retrieval was performed by microwave pretreatment in 10 mM citrate buffer, pH 6.0, for approximately 15 minutes in a Bio-Rad H2500 microwave oven at 95 °C. The sections were incubated with primary antirabbit polyclonal Mcl-1 (dilution, 1:50; S-19; Santa Cruz Biotech, Santa Cruz, CA) at 4 °C overnight. This antibody detected both the longer gene product, Mcl-1L, and the shorter gene product, Mcl-1S. The sections were then incubated sequentially with 1:100 biotinylated donkey antirabbit antibody (Amersham, Piscataway, NJ) followed by 1:100 streptavidin-biotin peroxidase complex (Dako, Glostrup, Denmark). Freshly prepared 3,3-diaminobenzidine (Sigma, St. Louis, MO) was used for color development. The sections were then counterstained with Mayer hematoxylin, dehydrated, and mounted. A first-trimester placenta known to be immunoreactive for the antibody studied was used as a positive control, and the negative control was achieved by substituting the primary antibody with 1 × phosphate buffered saline.

Assessment of Mcl-1 Expression

Assessment of the immunostaining of Mcl-1 was semiquantitative and was performed by two individual pathologists (W.-C.X. and A.N.Y.C.), as described previously.19 The immunostaining was scored with respect to relative intensity (Mcl-I) on an arbitrary scale (1, weak; 2, moderate; and 3, intense). Percentages of positive cells (Mcl-%) were estimated and graded as follows: 1, 1–25% positive cells; 2, 26–50% positive cells; and 3, > 50% positive cells. The immunoreactive score (Mcl-score) for each case was the sum of the values for the 2 parameters, which ranged from 0 to 6. Immunoreactive scores were then divided further into 2 groups: a low-score group (0–4) and a high-score group (5–6).

Statistical analyses were performed using the Statistical Package for Social Science (version 8 for Windows). Correlations between numerical data were determined using the Pearson chi-square test. Numerical data that were not in normal distribution were analyzed by the Mann–Whitney test. Logistic regression analysis was performed to analyze the relations between the experimental results and the clinical parameters. P values < 0.05 were considered statistically significant.

RESULTS

Up-Regulation of Mcl-1 mRNA Expression in Persistent Moles Was Identified by the cDNA Array

The two hybridized membranes of the Atlas™ Apoptosis Array showed similar gene expression patterns with the exception of two genes. Two-fold up-regulation of IGF-2 (code 16B; GenBank accession no. M29645) (results not shown) and induced myeloid leukemia cell differentiation protein (MCL-1; code 10J; GenBank accession no. 08246) (Fig. 1) were noted in persistent moles. Additional, confirmatory, real-time RT-PCR studies were performed for these two genes.

Figure 1.

A representative section from the Atlas™ Human Apoptosis Array showing gene expression in regressive moles (A) and persistent moles (B). Up-regulation of mRNA expression of Mcl-1 (arrowheads) was found in B.

Increased Expression of Mcl-1 in Progressive Moles Was Confirmed by Real-Time RT-PCR and Immunohistochemical Studies

The RNA expression levels of the full-length MCL-1 long (MCL-1L) in 19 cases of HM were shown in Figure 2. The expression of Mcl-1L was significantly greater in HMs that subsequently developed persistent GTN, than in HMs with spontaneous remission (P = 0.017, Mann-Whitney test). Conversely, there was no statistically significant difference in IGF-2 RNA expression between the 2 groups (P = 0.293).

Figure 2.

Quantitative real-time polymerase chain reaction analysis of Mcl-1 expression in regressive moles (RM) and in moles that developed gestational trophoblastic neoplasia (GTN), as illustrated by the ratio of signals from p16 and from the TBP housekeeping gene, measured by threshold cycle value (ΔCT). The expression of Mcl-1 was significantly greater in hydatidiform moles that developed into GTN compared with Mcl-1 expression in regressive moles.

Based on the confirmed, quantitative RNA analyses, we also investigated the protein expression of Mcl-1 in paraffin embedded clinical samples of CMs and PMs. Immunoreactivity for the Mcl-1 antibody (Fig. 3) was found predominantly in the cytoplasm of cytotrophoblasts. Relatively weak expression of Mcl-1 was found in syncytiotrophoblasts and in intermediate trophoblasts.

Figure 3.

Immunoreactivity for Mcl-1 was found predominantly in the cytotrophoblasts (CT) of the hydatidiform mole, which subsequently (A) regressed or (B) developed persistent gestational trophoblastic neoplasia requiring chemotherapy. The expression in syncytiotrophoblasts (ST) was relatively weak.

Data from the immunohistochemical assay for Mcl-1 are presented in Table 1. The Mcl-1 expression levels in HMs that subsequently developed into GTN with or without metastasis (mean, 5.47) was significantly greater (P < 0.001, Mann–Whitney test) compared with HMs that subsequently regressed (mean, 3.56). Further comparison of the low-score and high-score groups showed that increased expression of Mcl-1 was associated significantly with HMs that subsequently developed into GTN (P < 0.0001; Pearson chi-square test; odds ratio, 111.2; 95% confidence interval, 6.013–2058). Because interpretation of the intensity of staining may carry a risk of technical bias, the extent of staining was assessed again independently. The percentages of cells stained with Mcl-1 were statistically significantly greater in HMs that subsequently developed into GTN compared with HMs that regressed (P < 0.0001; Pearson chi-square test; odds ratio, 35.64; 95% confidence interval, 4.168–304.7). The results confirmed the cDNA array findings. Thus, there was a correlation between the expression of Mcl-1 and the progression of HM.

Logistic regression analysis also showed that the expression of Mcl-1 was associated significantly with the progression of HM (P < 0.0001); whereas no association was observed for Mcl-1 expression with patient age, gestational age, uterine size at diagnosis, theca-lutein cysts, pulmonary symptoms, or postevacuation uterine bleeding (all P > 0.05).

Although the S19 antibody used in this study detected both gene products (Mcl-1L and McL-1S), findings from the immunohistochemical analysis concurred with results of the real-time PCR analysis, with greater Mcl-1 expression detected in HMs that persisted and developed into GTN. The results were double blinded for the real-time PCR analysis and the immunohistochemical assay. However, the current immunohistochemical experiments were limited by the fact that distinction between MCL-1L and Mcl-1S protein expression was difficult and likely impossible. Because of the small number of samples with both RNA expression and protein expression analyzed, a correlation between RNA transcription and protein translation was not suitable. In other words, although the RNA and the protein expression patterns were in agreement, it could not be concluded definitely that Mcl-1L was detected immunohistochemically.

These results also were compared with our previous findings in the studies of apoptotic activities and related factors.9–12 Mcl-1 immunoreactivity was correlated with the apoptotic indexes assessed by the M30 cytoDeath antibody, which is a good indicator of apoptotic events at early stages (P < 0.001). However, it was not correlated to the apoptotic index as assessed with the TUNEL approach (P > 0.05).

DISCUSSION

HM is the most common type of gestational trophoblastic disease and is characterized by significant hydropic enlargement and variable trophoblastic hyperplasia. Although the majority of HMs spontaneously regress after suction evacuation, some may develop into GTN and will require chemotherapy.1–3 Similar to other human malignancies, neoplastic transformation of trophoblasts is likely to be a multistep process involving multiple genetic alterations.20 Potential biologic markers related to malignant transformation of trophoblasts include c-erbB-2,21 cyclin E,22 DOC-2/hDab2,23 Ras GTPase-activating protein,24 telomerase activity,11 and the apoptotic index.12, 13 However, no molecular parameter currently can replace serial hCG assays as the mainstay for predicting GTN.

Based on our previous studies, the apoptotic index has been documented as a likely association factor in the development of GTN in HM.12, 13 Therefore, we employed an apoptosis cDNA array to identify the genes related to apoptosis that may be expressed differentially in aggressive HM. The cDNA array is useful in the screening of differentially expressed genes. However, cDNA studies on human GTD have been sparse.25–27 To the best of our knowledge, this is the first cDNA array study that directly compares the differential expression of genes in HMs that progressed favorably and HMs that progressed unfavorably.

Apoptosis is controlled mainly by two different but interacting activation pathways.28 The extrinsic pathway is initiated by transmembrane receptors (CD95, tumor necrosis factor [TNF], and TNF-related apoptosis-inducing ligand receptor) that activate the caspase cascade sequence. The intrinsic pathway requires the release of mitochondrial proteins and cytochrome c and involves the proapoptotic and antiapoptotic Bcl-2 family proteins.29 Most of these apoptosis-related proteins are included in the apoptosis cDNA array that was used in the current study.

It is noteworthy that the pattern of expression regarding apoptosis-related genes in HMs that regressed and in HMs that persisted, in fact, were very similar with the exception of two genes, Mcl-1 and IGF-2. It was found that these two genes were overexpressed in HMs that developed into GTN. Only the increased expression of Mcl-1 (and not IGF-2) was confirmed by quantitative real-time PCR analysis. Mcl-1 protein expression was studied further and confirmed the findings of the cDNA and RNA expression studies. This lack of a complete correlation between the cDNA array and subsequent RT-PCR or immunohistochemical analysis has been observed previously and may be due to variation in gene expression profiles in individual patients.16 We also noted that there was no significant difference in the Bcl-2 and Bax expression in this apoptosis cDNA experiment; in fact, such results concurred with our earlier findings using a semiquantitative immunohistochemical approach, with which no statistically significant correlation was found between Bcl-2 or Bax expression and the clinical progression of patients with HM.12 The observation that so few differentially expressed, apoptosis-related genes could be identified may be related to the possibility that it is the interaction of genes, and not altered expression of individual genes, that contributes to the difference in apoptotic activity in HMs with distinct clinical behavior. Conversely, other genes may be operating in affecting the apoptotic activity of HMs with different levels of clinical progression, and their identification remains to be explored further.

The Mcl-1 gene shares significant amino-acid sequence homology with Bcl-2 and, thus, is believed to be within the Bcl-2 family that regulates apoptosis through the formation of homodimers and heterodimers, which maintain the homeostasis that controls the progression to apoptosis.30–33 Earlier immunohistochemical studies demonstrated Mcl-1 expression in a variety of human epithelial tissues and malignancies.34–40 It also was found that the proportion of Mcl-1-positive cells was greater in high-grade tumors and metastases.34–36

Trophoblasts can be divided into cytotrophoblasts, syncytiotrophoblasts, and intermediate trophoblasts. Cytotrophoblasts are proliferating germinal cells41–43 and subsequently will differentiate and fuse to form syncytiotrophoblasts. It has been found that functional genes display spatial differences in expression in trophoblastic tissues.12, 44–48 In the current study, Mcl-1 was found in the cytoplasm of cytotrophoblasts and villous stromal cells, whereas relatively weak expression was found in syncytiotrophoblasts and in intermediate trophoblasts. This immunostaining pattern of Mcl-1 was the opposite of the Bcl-2 pattern that we observed in our previous studies.12 In fact, it has been found that Mcl-1 expression and Bcl-2 expression occur in gradients with opposing directions in complex epithelia.37

Such spatial distribution in Mcl-1 expression also concurred with the patterns we observed in our previous studies on apoptotic activity in trophoblastic disease.11–13 The M30 cytoDeath antibody recognizes caspase-related cleavage sites before the disruption of membrane asymmetry and DNA strand-breaks occur and can identify apoptotic events earlier than the TUNEL approach. Our results showed a significant correlation between Mcl-1 and M30 expression, but not with the TUNEL index in GTD, in agreement with the possibility that Mcl-1 may be responsible for the antiapoptotic activity in the early stage of apoptosis located in the less differentiated cytotrophoblasts.49

In summary, the results of the current study demonstrate the differential expression of Mcl-1 in patients who had HM that subsequently developed into persistent disease and required chemotherapy compared with Mcl-1 expression in patients who had HM that spontaneously regressed. Mcl-1 may be a useful marker for predicting the progression of HM. If such findings are confirmed in larger studies, then Mcl-1 may be employed as alternative marker and may reduce the burden on both patients and community resources regarding the follow-up of patients with HM.

Acknowledgements

The authors thank Ms. Stefanie Liu for technical advices and Mr. S. K. Lau for photographic assistance.

Ancillary