神經母細胞瘤生物預後因子之研究 —臨床重要性及對外科術式之影響

The Study of Biologic Prognostic Factors of Neuroblastoma—Clinical Significance and Impact on Surgical Decision

許文明(Wen-Ming Hsu)

指導教授 ： 郭明良 ； 陳維昭 ； 謝豐舟

國立臺灣大學/醫學院/臨床醫學研究所/博士(2004年)

https://doi.org/10.6342/NTU.2004.00864

摘要

中文摘要

　　神經母細胞瘤係源自於胚胎神經脊細胞的惡性腫瘤。它是兒童在中樞神經以外最常見的硬性腫瘤(solid tumor)，約佔所有兒童癌症的5~10%。在新生兒的年盛行率約七千分之一，在台灣每年約新增30名病例。這些病人在被發現時，約5成為第4期病患，在積極治療下，其五年存活率也僅約30%。因此，神經母細胞瘤實為兒童癌症的一項重要課題。然而儘管多數的神經母細胞瘤均十分惡性，卻也有相當比例的神經母細胞瘤可自行分化為良性腫瘤，甚至自行消失，由此可見神經母細胞瘤是一種生物行為十分複雜的腫瘤。

　　目前對於神經母細胞瘤的治療方式主要包括：外科手術、化學治療、放射線治療、及自體骨髓移植。以外科手術而言，由於神經母細胞瘤均原發於腎上腺或交感神經節，因此其腫瘤大多位於後腹腔或脊柱兩側，常會包住主動脈、下腔大靜脈或其他人體重要血管及器官，甚至鑽入脊椎，造成手術切除極度困難及危險，在耗費極大的體力及時間之後也未必能將腫瘤完全切除。然而由於神經母細胞瘤的生物行為十分複雜，有些腫瘤即使切不乾淨，甚至不必切除，預後仍然很好，而有些腫瘤即使切的很乾淨，仍然很快就會復發及惡化。在過去的研究中，針對高期別的神經母細胞瘤，手術切除乾淨與否對預後是否有幫助存在著極大的爭議。這些研究大多只考慮手術本身，而忽略的腫瘤生物行為本身的重要性。

　　為了了解神經母細胞瘤複雜的生物行為，到目前為止，至少有195種生物預後因子被提出 ，雖然其中僅約10種被認為有臨床的重要性，但是例如MYCN（一種基因轉譯因子）及Trk-A（一種神經生長因子接受器）對於神經母細胞瘤的預後有極大的影響。神經母細胞瘤若帶有多倍數MYCN其預後極差，反之若表現陽性Trk-A則預後較好。忽略了這些生物因子，很可能會誤導了外科治療在神經母細胞瘤的角色。而由於在高期別的神經母細胞瘤中僅約3成帶有多倍數MYCN，另僅約4成表現陽性Trk-A，可見多數的高期別神經母細胞瘤，其腫瘤生物行為無法由MYCN及Trk-A來預測。因此，尋找新的生物預後因子亦是重要的課題。

　　目前對於神經母細胞瘤的病理成因仍然不明，但大多認知和胚胎神經母細胞無法分化或凋亡有關，因此，可能的生物預後因子也應該和神經母細胞瘤的分化或凋亡有關。我們找到兩個可能的目標蛋白質，一個是Calreticulin，另一個是GRP78。

　　Calreticulin是一種細胞內質網蛋白，主要是負責新生蛋白質的摺疊、運送，以及鈣離子平衡的調控。在細胞實驗中，Calreticulin曾被發現當神經母細胞分化時會大量增加，且會移至細胞膜表面，而當細胞膜表面的Calreticulin被抑制時，細胞的分化也會被抑制。另外Calreticulin也曾被發現當在細胞中大量表現時會促進細胞的凋亡。由此可見Calreticulin可能和神經母細胞瘤的分化及凋亡有關，有可能可以成為神經母細胞瘤的預後因子。

　　另外GRP78也是一種細胞的內質網蛋白，和Calreticulin有類似的功能，即負責新生蛋白質的摺疊、運送及鈣離子的調控。GRP78曾被發現在胚胎神經管的神經外皮細胞會大量表現，顯示它和神經的發育可能有關。另外在細胞實驗發現當細胞分化時GRP78會大量增加，而如果GRP78被大量表現時會促進神經細胞的分化，但當GRP78的表現被抑制時，神經細胞的分化也會被抑制。此外，當神經細胞凋亡時，GRP78的mRNA也會特異性的增加，因此，顯然GRP78也和神經母細胞瘤的分化及凋亡有密切的關聯，GRP78也可能成為神經母細胞瘤的預後因子。

　　基於上述討論，我們進行了以下三個研究。

　　第一部分，我們想探討MYCN及Trk-A這兩種已知神經母細胞瘤的重要生物預後因子，對於外科在高期別神經母細胞瘤的角色的影響。我們一共收集了10位第3期及35位第4期的神經母細胞瘤病患。依其手術結果分為腫瘤切除乾淨及切除不乾淨兩組。另外，並使用螢光原位接合的方法來檢測腫瘤細胞中MYCN的倍數，以及使用組織免疫染色法來偵測Trk-A的表現。同時將病患的預後依MYCN及Trk-A的結果分為3群，即（1）低危險群： Trk-A陽性且MYCN為單倍數，（2）中危險群：Trk-A陰性且MYCN為單倍數，（3）高危險群：MYCN為多倍數。結果顯示45位病人中有21位腫瘤切除乾淨，另外帶有MYCN多倍數的有15個（33.3%），Trk-A陽性的有17個（37.8%）。腫瘤切除乾淨的這群病人有較高的比率產生手術合併症（33.3%比8%，p=0.036）。低危險群的腫瘤較中危險群或高危險群的腫瘤容易切除乾淨（分別為12/13，4/17及5/15，p<0.001）。在低危險群病人不論腫瘤是否切除乾淨病人均存活，而在高危險群病人，不論腫瘤是否切除乾淨其五年存活率為0%。因此，在這兩群病人其預後與手術方式無關。但在中危險群病人，若腫瘤切除乾淨則五年存活率為66.7%，反之切除不乾淨者則五年存活率僅為23.1%，顯然腫瘤切除乾淨對於這群病患的預後有很大的幫助（p=0.037）。這個研究顯示要將神經母細胞瘤切除乾淨是一種具高度危險的手術。對於具有低危險生物因子的病患，其腫瘤通常很容易切除乾淨且預後極佳。而具有中危險生物因子的病人，儘管手術風險會增加，但盡可能把腫瘤切除乾淨是有助於改善預後。至於具有高危險生物因子的病患，應儘量採取低風險的手術方式，更有效的內科療法才能改善他們的預後。生物預後因子顯然對於高期別神經母細胞瘤的手術方式有重大影響，因此，在進行重大手術之前，能夠越充分了解其生物預後因子，將有助於降低手術風險並改善病患的預後。

　　第二部分的研究是為了了解Calreticulin的臨床重要性，以及Calreticulin是否可成為神經母細胞瘤的生物預後因子。本研究共收集了68位神經母細胞瘤病患，包括6位第1期，13位第2期，10位第3期，34位第4期，及5位4S期。其中8位沒有臨床症狀而是經由尿液篩檢發現。有22位病患的年齡小於1歲，有37位病患的腫瘤原發於腎上腺，組織病理型態依其細胞分化的程度及比率分為未分化型神經母細胞瘤（35位），正在分化型神經母細胞瘤（20位）及神經節神經母細胞瘤（13位）。我們用組織免疫染色法檢測Calreticulin在腫瘤組織中的表現。另外以西方點墨法來驗證抗體的專一性，並以定量real-time PCR來印證組織免疫染色結果的可靠度。結果顯示68位病人中有32位（47.1%）的腫瘤細胞呈現陽性Calreticulin染色，陽性Calreticulin染色只表現於神經母細胞瘤細胞或神經節細胞，但Schwann氏細胞則不表現，而在18個腫瘤中血管內皮細胞也同時呈現陽性Calreticulin。
這個組織染色可被抗體專一性的抗原所抑制，同時抗體的專一性也進一步地由西方點墨法驗證。另外定量real-time PCR所檢驗腫瘤組織中的Calreticulin的mRNA含量也和組織免疫法的結果相當一致。Calreticulin的陽性率和組織的分化程度成正比，同時在大量篩檢年齡小於1歲，及期別較早（1，2，4S）的病患中也比較高，但在帶有多倍數MYCN者則較低。陽性Calreticulin染色不論在單變項或多變項分析均可預測較好的預後。而即使在高期別（3，4）及單倍數MYCN腫瘤仍具有預測預後能力。本研究證實了Calreticulin的表現和腫瘤組織的分化有密切的正相關，同時可作為神經母細胞瘤病患的生物預後指標。而Calreticulin在高期別及單位數MYCN腫瘤仍具有預測預後能力，更顯示其臨床上的重要性。

　　第三部分的研究是為了了解GRP78臨床重要性，以及GRP78是否可成為神經母細胞瘤的生物預後指標。延續第二部分的研究，我們同樣收集68位神經母細胞瘤病患，並用相同的研究方法即組織免疫染色，西方墨點法及real-time PCR去檢測GRP78在腫瘤組織的表現。結果和Calreticulin呈現類似的結果。GRP78的染色強度及陽性率和腫瘤組織的分化程度成正相關。同時在期別較早的腫瘤GRP78的陽性率較高，而在帶有多倍數MYCN的腫瘤則陽性率較低。單變項及多變項分析均顯示GRP78之表現可預測良好的預後。而同樣的，GRP78在高期別及單倍數MYCN腫瘤仍具有區別預後之能力。

　　經由我們一系列的研究，我們發現神經母細胞瘤的生物預後因子對於治療方式尤其是手術方法的選擇有十分重大的影響，在手術前徹底分析其生物因子有助減少合併症並改善預後。而我們也確立了Calreticulin及GRP78這兩種內質網蛋白可作為神經母細胞瘤新的生物預後指標。而由於這兩種蛋白質均和神經母細胞瘤的分化和凋亡有關，因此進一步地探討他們在神經母細胞中的調控，將有助於了解神經母細胞瘤的病理成因，並可作為治療的標的。

關鍵字

神經母細胞瘤 ； Calreticulin ； 生物預後因子 ； GRP78 ； 外科術式

並列摘要

1. Backgrounds

1.1 Neuroblatoma is a clinically and biologically heterogenous tumor

Neuroblastoma (NB) is an embryonic cancer of the postganglionic sympathetic nervous system, which most commonly arises in the adrenal gland. It is one of the most common pediatric cancer with an incidence of 8.0 per million per year; and 96% of cases occur before the age of 10 years. The estimated prevalence is about 1 in 7000 live births, therefore there are about 30 new cases diagnosed in Taiwan. Neuroblastoma is quiet a heterogeneous tumor and presents a broad clinical and biologic spectrum ranging from highly undifferentiating tumors with very poor outcomes to the most differentiated benign ganglioneuroma or neuroblastomas with high probability of spontaneous regression and hence favorable prognosis. To explain the heterogeneous behavior of neuroblastoma, more than 195 prognostic biologic markers have been reported, and at least two distinct types of neuroblastoma have been suggested to exist. One, favorable NB, is characterized genetically by near-triploid karyotypes with whole chromosome gains. These tumors are usually localized and frequently seen in patients less than 1 year of age or detected by mass screening. The favorable NBs either regress spontaneously or differentiate into benign ganglioneuromas, and hence leading to a good outcome with minimal or even no treatment. The other, unfavorable NB, is characterized by diploid karyotypes with structural changes, including deletion of 1p or 11q, unbalanced gain of 17q and/or amplification of the MYCN protooncogene. Patients with unfavorable NBs are usually older than 1 year of age and have advanced stages of disease and a very poor prognosis, even with aggressive treatment. These lines of evidence suggest that more detail understanding the clinical and biologic characteristics of NB may allow more appropriate intensity of therapy on these tumors. Indeed, recent progress in biologic and genetic understanding of NB has led to a risk-related therapy to improve therapeutic outcome and avoid unnecessary treatment.

1.2 Pathogenesis of NB

The pathogenesis of NB remains obscure. NB cells have a great potential to differentiate into mature cells or to spontaneously regress. A large proportion of NBs identified from mass screening, as well as stage 4S tumors, later differentiate into mature histologies or spontaneously regress. In addition, many normally expressed molecular markers in embryonic neuroblastic cells, such as HNK-1, tyrosine hydroxylase, Trk-A and CD44 are found in NBs, suggesting that NB arises during the developmental stages of the embryonic sympathetic system. Furthermore, expression of apoptosis-related genes has been demonstrated in NB, and patients with NB tumors that showed more apoptosis had a better prognosis. These studies suggest that failure of either differentiation or regression by apoptotic death of NB cells is critical for the development of clinical NB.

1.3 The role of surgery in NB

For decades, there has been debate regarding the role of surgery in NB, especially in advanced NB. Surgical resection alone can provide adequate treatment for early-stage disease. However, the importance of tumor resection as a therapeutic maneuver in the treatment of advanced NB remains controversial. Some investigators have reported that gross total resection (GTR) of tumors improves survival rate in advanced stages, while others have not found any improvement with GTR. Most of these studies did not take into consideration the effect of biologic factors of NB on the outcome of aggressive surgery but only clinical factors of disease stage and patient’s age. Biologic factors, such as MYCN (a transcription factor) and Trk-A (a nerve growth factor receptor), have great influence on the tumor behavior of NB and the patient’s outcome. NBs with MYCN amplification usually progress rapidly and lead to a very poor outcome, while those tumors with high levels of Trk-A expression are susceptible to tumor differentiation or regression and lead to a favorable outcome. Neglecting these biologic factors in the treatment of NB may mislead the true role of surgery in the therapeutic strategies.

1.4 The role of calreticulin in NB

Calreticulin (CRT) is an endoplasmic reticulum protein with two major functions: molecular chaperoning and regulation of Ca2+ homeostasis. Furthermore, CRT can also modulate cell adhesion, integrin-dependent Ca2+ signaling and steroid sensitive gene expression outside the endoplasmic reticulum. Although CRT has many physiologic functions in the cell, its role in pathologic conditions has been studied infrequently. Evidence suggests that CRT is linked to the biology of NB. CRT has been found on the surface of NB cells and is essential for neurite formation when the cells are induced to differentiate. In a NB cell line study, Johnson et al showed that CRT protein levels increased markedly when the cells were induced to differentiate with dibutyryl c-AMP. CRT has also been found to affect cell susceptibility to apoptosis and to be over-expressed in highly apoptotic regions of the embryo. In addition, CRT has been shown to be essential for neural development in mice. These lines of evidence give rise to the intriguing possibility that CRT may affect the differentiation and apoptosis of NB, and thus may have a role in the tumor behavior of this cancer.

1.5 The role of GRP78 in NB

Glucose-regulated protein 78 (GRP 78) is the endoplasmic reticulum (ER) located member of the family of heat shock protein 70 (HSP70) molecular chaperone. GRP78 is essential for protein folding in the ER lumen, for translocation of newly-synthesized secretory precursors across the ER membrane, and for the transport back across the membrane of aberrant polypeptides destined for degradation by the proteosome. GRP78 may also contribute to the Ca2+ homeostasis in the cell. It has been shown that GRP78 and other HSP70 family members are constitutively expressed at high levels in neuroepithelial cells of the neural tube, suggesting that these proteins may play significant roles in the normal differentiation and development of neural tissue. In vitro studies with PC12 cell line showed: GRP78 protein levels increased after cells were induced to differentiate; overexpression of GRP78 protein by adenoviral gene transfer promoted the neurite outgrowth by nerve growth factor (NGF); and antisense nucleotides for the GRP78 gene blocked the NGF induced neurite outgrowth. In addition, it has been shown that selective induction of GRP78 mRNA plays an important role in the apoptotic death of neurons deprived of NGF and could be a landmark of the neuronal apoptotic death. These lines of evidence give rise to the intriguing possibility that GRP78 may affect the differentiation and apoptosis of NB, and thus may have a role in the tumor behavior of this cancer.

2. Studies

2.1 The influence of biologic factors on the surgical decision of advanced NB

2.1.1 Aim of study

GTR of advanced NB is a high risk procedure and carries a great challenge to the Pediatric surgeons, which usually required dissection of the tumor from great vessels and other vital organs such as kidneys, liver, and spinal cord. The surgical complication rate may be as high as 30%, including 20% of patients having their normal organs removed (in particular, their kidneys). However, it is not clear for which subtype of advanced NB a surgeon should carry out such a risky radical resection in order to achieve the best long-term result with permanent remission. One important reason for such a dilemma is probably the marked biologic heterogeneity of NB. More than 195 different tumor markers in NB have been studied in the literature. In addition, several reports have shown that although GTR contribute to the control of local recurrence, the overall survival of NB patients depends largely on the resolution of distant metastasis. This evidence suggests that the biologic nature of the NB has great impact on the tumor behavior, and hence the strategy to treat this tumor. In this study, we evaluated the impact of biologic factors of MYCN and Trk-A on the outcome after gross total resection (GTR) in advanced NB with the aim of determining the role of GTR in different biologic groups of NB.

2.1.2 Study design and methods

The present study was designed to use two important biologic markers, MYCN and Trk-A, by which the patients were classified into 3 risk groups, to define the impact of biologic makers on the role of surgery in NB. The 3 risk groups were (1) low risk, positive Trk-A expression and normal MYCN copy number, (2) intermediate risk, negative Trk-A expression and normal MYCN copy number, (3) high risk, MYCN amplification with positive or negative Trk-A expression.

From December 1990 to December 2002, 45 advanced-stage NB patients treated at National Taiwan University Hospital and having complete follow-up were included in this study. Thirty-five patients had stage 4 tumors and the other 10 had stage 3 tumors according to the International NB Staging System. There were 27 males and 18 females. The median age at diagnosis was 3.5 years (range, 0.5-11.5 years) with 4 patients were of 1 year of age or younger and 41 were older than 1 year. The pathology was favorable in 17 tumors and unfavorable in 28 according to Shimada’s classification. In 26 patients the primary tumors were located in the adrenal gland; 9 were paraspinal, 5 mediastinal, 3 cervical, and 2 pelvic.

After initial diagnostic biopsy surgery, all patients received induction chemotherapy. GTR of the tumor was attempted after 3 to 6 cycles of chemotherapy. GTR was defined as macroscopic primary tumor and regional lymph nodes were all removed. There should be no grossly visible tumors detected by the immediate follow-up image studies (Fig. 1, A-D). Patients receiving lesser procedures (partial excision and biopsy) were grouped as incomplete resection (IR) to compare with the GTR group. After the definitive surgical procedure, patients subsequently received either 3 to 5 cycles of maintenance chemotherapy or autologous bone marrow transplantation. The median follow-up after the diagnosis was 30 months (range 5 to 144 months).

The expression of Trk-A was assessed on formalin fixed, paraffin embedded tissue by an avidin-biotin complex immunoperoxidase staining technique as described previously. The immunoreactivity of Trk-A was recorded as follows: “negative” indicated staining was absent throughout the specimen, and “positive” indicated that brownish granular staining was present in cytoplasm of the NB or ganglion cells.

The MYCN status in NB tumors was evaluated by FISH analysis of formalin-fixed paraffin-embedded tissues or fresh tumor single cells. Fluorescent signals representing the presence of MYCN gene were counted for 100 cells on each slide. Cells with clusters or more than 6 grains of red fluorescence were considered evidence of MYCN gene amplification.

2.1.3 GTR carries a higher complication rate

After induction chemotherapy, GTR was achieved in 21 patients. Among 24 patients with incomplete tumor excision, 5 patients did not receive further tumor excision except biopsy due to dying from rapid tumor progression or chemotherapy complications. In the GTR group, there was no surgical mortality, but 1 chylothorax, 2 adhesion ileus, and 4 nephrectomies, with a total complication rate of 33% (7/21). In the IR group, there was one mortality (dying from respiratory failure after biopsy of mediasitnal tumor), and 1 nephrectomy, with a total complication rate of 8% (2/24). The complication rate was significantly higher in the GTR group (P = 0.036). This difference was mainly due to that there were more nephrectomies in the GTR group than the IR group (4 and 1 respectively). Nephrectomy may be needed in selective patients whose tumors directly invade the kidney or the renal vessels to achieve GTR. However, the risk of nephrectomy should be minimized by preoperative intensified chemotherapy and skillful surgical technique.

2.1.4 Association between other clinicopathologic and biologic factors

Fifteen (33.3%) tumors had variable percentage (12-93%) of cells with positive signals of fluorescence of MYCN in FISH analysis, and were recognized as MYCN amplification (Fig. 2, A, B). Seventeen (37.8%) tumors had positive Trk-A imuunostaining on NB or ganglion cells (Fig. 2, C, D).

Further analysis of the association between GTR and various clinicopathologic and biologic factors showed that GTR was significantly easier to be achieved in tumors with low biologic risk than with intermediate or high risk (12/13, 4/17, and 5/15 respectively, P < 0.001, Table 1). In addition, GTR also tended to be achieved in tumors with favorable histology (P = 0.059).

The degree of resection may be confounded by the tumor biology. It is likely that tumors with a better prognosis lend themselves to more complete resection. Here, we showed that GTR was significantly easier to be achieved in low risk NBs (P < 0.001). In a cohort of non-stage 4 NB treated without cytotoxic therapy, Cheung et al showed that 42 of 43 patients were alive and well at a median follow-up of 50+ months. Thirty-one of 33 (94%) tested tumors in this cohort showed positive Trk-A expression. On the contrary, there were only 37.8% of advanced NBs in our study showed positive Trk-A expression. These evidences suggest that NB tumors with positive Trk-A expression tend to be more localized and less invasive, which makes complete resection easier.

2.1.5 Survival analysis

2.1.5.1 Univariate analysis

Kaplan-Meier analysis showed that patients with GTR or IR of tumors had a 5-year survival rate of 52.3% and 16.7% respectively (Fig. 3). GTR predicted a significantly better survival (P = 0.001). Further univariate analysis showed that Shimada’s classification of histology and positive Trk-A expression strongly correlated with better survival, whereas MYCN amplification predicted a very poor outcome (Table 2). In addition, the prognosis of all patients could be clearly discriminated by the biologic risk grouping: (1) low risk, 13 patients, 5-year survival rate 65.6%, (2) intermediate risk, 17 patients, 5-year survival rate 32.1%, (3) high risk, 15 patients, 5-year survival rate 0% (Fig. 4, P < 0.001, log-rank test).

2.1.5.2 Impact of biologic factors on surgical outcome and decision

The survival of NB patients was stratified according to the risk grouping and surgical procedures (Fig. 5). For each risk group, GTR could only provide prognostic value in the intermediate risk patients (Fig. 6, P = 0.037), but not in the low risk or high risk patients. In 17 patients of intermediate risk group, 4 received GTR of their tumors with a 5-year survival rate of 66.7%, and the other 13 had IR of their tumors with a 5-year survival rate of 23.1%. All 13 patients of low risk group survived, and all of them except 1 received GTR of their tumors. There were 5 patients received GTR and 10 patients received IR of their tumors in the high risk group. All the 15 patients of high risk group died within 5 years except 2 patients with GTR of primary tumors showed bone and brain metastasis at 11 and 13 months follow-up respectively. On the other hand, in 24 patients who had IR of their tumors, only 1 patient belonged to the low risk group and survived, while the prognosis of the 13 patients of intermediate risk group was as poor as the 10 patients of the high risk group (5-year survival rate, 23.1% and 0% respectively, Fig. 7A, P = 0.214). In 21 patients who had GTR of their tumors, the prognosis of the 5 patients of intermediate risk group was as favorable as the 12 patients of low risk group (5-year survival rate, 66.7% and 62.3% respectively), and was significantly favorable than the 4 patients of high risk group (5-year survival rate 0%, Fig. 7B, P = 0.004).

We showed that GTR only provided prognostic value in the intermediate risk group (negative Trk-A expression and normal MYCN copy number). However, GTR did not carry a better prognosis in the low risk patients (positive Trk-A expression and normal MYCN copy number) due to the favorable prognosis of this group. Also GTR did not provide prognostic value in the high risk group (MYCN amplification with positive or negative Trk-A expression) due to the extremely poor prognosis of these patients. Matthay et al showed that in advanced NBs with favorable biologic features (determined by MYCN copy number, Shimada’s histopathology, and serum ferritin level), GTR did not improve survival rate due to the overall favorable outcome of this group of patients. Nevertheless, GTR did improve the outcome in patients of advanced NB with unfavorable features. These evidences suggest that biologic markers have great impact on surgical outcome of NB. On the other hand, our analysis further showed that in the IR group, patients with intermediate biologic risk had a poor prognosis as the high risk patients, whereas in GTR group the prognosis of the intermediate risk patients was as favorable as the low risk patients. This result denotes that in patients with selective biologic risks, surgical procedures do alter their natural outcomes.

It has been shown that GTR or partial resection of advanced NB with MYCN amplification carried similar outcomes, but both procedures did carry a better survival than simple biopsy did. This may be due to the very poor biologic nature of some NBs, which progressed rapidly after a biopsy and never underwent delayed surgery. In addition, there is only one third (15/45) of advanced NB showing MYCN amplification, which denotes that advanced NB without MYCN amplification (5-year survival rate 46.8%) is a large group of heterogeneity. Therefore, additional biologic markers, such as Trk-A, are needed to further discriminate these heterogenous tumors. As shown by our results, a more detailed biologic classification of patients would provide a more precise guide to the treatment strategy. We suggest that only patients of intermediate risk would benefit from the risky procedure of GTR. For patients of low risk or high risk, GTR should only be attempted when the surgical risk is not so high to provide a better local control. Low risk patients with IR of their tumors would still have a favorable outcome under a less aggressive chemotherapy regimen or even no cytotoxic therapy. High risk patients with either GTR or IR of their tumors still have a very poor outcome, and intensified chemotherapy is required to improve their survival.

2.1.6 Summary

GTR is a serious and often life-threatening operation, nevertheless, it only carries prognostic benefit in a selective intermediate risk biologic group of NB patients. GTR can be easily achieved in the low risk patients with favorable outcome. Nevertheless, for high risk NB patients, GTR should only be attempted after the metastatic tumor is well-controlled by intensified chemotherapy to aid the control of local recurrence. In addition, it is not justified to sacrifice vital organs such as kidney to achieve GTR for high risk patients. Finally, it should be stressed that biopsy of primary or metastatic tumors before chemotherapy for thorough study of biologic markers is critical to the guide of treatment of advanced NB, although most diagnosis can be made by biochemical studies and bone marrow aspirations.

2.2 Clinical Significance of calreticulin expression in NB

2.2.1 Aim of study

Recent advances in understanding the biology and genetics of NB have allowed a risk-group based therapy, suggesting that detailed biologic studies of tumor behavior are critical to the treatment of NB. The current study was designed to determine whether immunohistochemical expression of CRT in NB was associated with clinicopathologic and biologic parameters and whether the expression was useful for predicting outcome. In addition, the mRNA and protein levels of CRT in tumor tissues were also quantified by real-time PCR and Western blot respectively to compare with the immunohistochemical findings.

2.2.2 Methods of study

Sixty-eight NB patients treated at the National Taiwan University Hospital from January 1991 to December 2002 were included in this study. Eight of these NB patients were identified by mass screening of urinary vanillyl-mandelic acid for infants. The eligible criteria for patients to be enrolled in the study were the availability of sufficient tumor tissues for thorough studies and having complete follow-up. There were 36 males and 32 females. The median age at diagnosis was 2.5 years (range, 0 - 11.5 years). Adrenal gland (37 cases) was the most common primary tumor site, followed by retroperitoneum (18 cases), mediastinum (6 cases), neck (4 cases) and pelvis (3 cases). The histologic features of NB were classified into undifferentiated NB (UNB, 35 cases), differentiating NB (DNB, including poorly differentiated subtype, 20 cases), and ganglioneuroblastoma (GNB, 13 cases) according to the percentage and degree of differentiation of the NB cells using the criteria of the International NB Pathology Classification [13, 14]. Six patients had stage 1 tumors, 13 had stage 2 tumors, 10 had stage 3 tumors, 34 had stage 4 tumors and 5 had stage 4S tumors according to the International NB Staging System (INSS). Fifteen of the 68 NBs were demonstrated to have MYCN amplification by fluorescence in situ hybridization analysis of formalin-fixed paraffin-embedded tissues or fresh tumor single cells. All stage 1 tumors were treated by surgery alone, whereas the other tumors were treated with a combination of surgery and chemotherapy with or without autologous bone marrow transplantation according to the patient’s risk grouping. The median follow-up after diagnosis was 38 months (range, 1-144 months). The overall 5-year survival was 52.6%.

CRT expression was assayed using an avidin-biotin complex immunoperoxidase staining technique on archival paraffin-embedded tissue specimens obtained before chemotherapy. The immunoreactivity of CRT was recorded as follows: “negative” indicated staining was absent throughout the specimen, and “positive” indicated that brownish granular staining was present in cytoplasm of the NB or ganglion cells. To verify the specificity of the CRT antibody, a blocking peptide corresponding to the carboxy-terminal 17 amino acids (401 – 417) (5x of antibody, Santa Cruz, California, USA) was added along with the CRT antibody when performing the positive control staining to see if the immunostaining could be blocked specifically.

The mRNAs of CRT in 7 NB tumor tissues were examined by real-time PCR for comparison with the immunohistochemical expression of CRT.

CRT proteins in NB tumors were also examined by immunoblot analysis to compare with the results obtained by immunohistochemical study. To test the specificity of anti-CRT antibody used in immunohistochemical study, competition assay was performed in which the specific blocking peptide (5x of antibody) was added along with the CRT antibody during immunoblotting. Cell lysates from un-treated and all-trans retinoic acid-treated NB cell line (Neuro-2A, American Type Culture Collection CCl-131) were used as positive control for immunoblot analysis of CRT.

2.2.3 Expression profile of CRT

In the positive control staining of a ganglioneuroma, positive CRT staining was seen in ganglion cells, and localized in the cytoplasm consistent with ER location of CRT (Figure 8, A). When the specific blocking peptide was added along with the anti-CRT antibody, the immunostaining was blocked completely (Figure 8, B). Among 68 NB tumors, 32 (47.1%) showed positive expression of CRT in NB or ganglion cells. The CRT positive staining was mainly seen in differentiating NB cells or ganglion cells (Figure 8, C-E). Eighteen of the 32 NB tumors (56.3%) also had positive CRT expression in the endothelial cells. However, no tumor had positive staining of Schwann cells. Three tumors having CRT expression only in endothelial cells but not in neuroblastic cells were designated as having negative CRT expression in the analysis. One undifferentiated NB with brain metastasis showed negative CRT staining in the tumor, whereas the adjacent normal neurons did express CRT (Figure 8, F). Furthermore, normal sympathetic ganglions and the adrenal medulla also showed positive CRT staining in immunohistochemical studies.

CRT mRNA and protein levels in 7 NB tumors were evaluated by real-time PCR and Western blot to compare with the results of immunostaining (Figure 9A, lanes a-g). The results of real-time PCR were quite compatible with those of immunostaining. Cases with differentiated histologies (DNB or GNB) showed high expression of CRT mRNA and positive immunostaining (Figure 9A, lanes a, b, c, f). On the contrary, cases with UNB had low expression of CRT mRNA and negative immunostaining (Figure 2A, lanes d, e, g). In Western blot study using the same antibody as used in immunohistochemistry, a single band of approximately 55 kDa corresponding to the reported size of CRT could be detected in positive control Neuro-2A cells as well as in cases with high mRNA expression and positive immmunostaining (Figure 9A, D0, D5, and lanes a-c). However, there were a few cases showing a discrepancy between immunoblot and immunostaining results (Figure 9A, lanes e-g). Since the stromal cells of differentiated histologies had either no or very low expression of CRT, they could result in a dilution effect of proteins in the tumor lysates (Figure 2A, lane f). On the other hand, since cases with undifferentiated histology usually had a quite compact NB cell population, despite that individual NB cell had a low CRT protein level, the total protein levels might be high in the whole tumor lysate (Figure 9A, lanes e, g). In the cell line studies, differentiated Neuro-2A cells induced by retinoic acid treatment for 5 days (Figure 9A, D5) had higher CRT protein levels than untreated cells (Figure 9A, D0). However, untreated cell lysates still had a significant level of CRT protein. These results suggested that in studying such a heterogenous tumor as NB, immunohistochemistry had the advantage to identify the specific cells of interest and to evaluate the overall tumor quality simultaneously when comparing with PCR and Western blot. The specificity of anti-CRT antibody used in immunohistochemistry was again confirmed by a competition assay, in which the CRT immunoblotting was blocked completely by the specific antigenic peptide (Figure 9B).

2.2.4 Association between CRT expression and other clinicopathologic and biologic variables

The relationship between CRT expression and clinicopathologic and biologic variables of NB is summarized in Table 3. The percentage of positive CRT immunostaining increased as the tumor histology became differentiated with 12 of 20 (60%) differentiating NBs and 12 of 13 (92.3%) GNBs showing positive staining, whereas only 8 of 35 (22.9%) undifferentiated NBs having positive staining (P < 0.001). Positive CRT immunostaining was also frequently seen in tumors detected from mass screening (P = 0.022) as well as in tumors of infants (P = 0.016) and early clinical stages (stage 1, 2, 4S) (P < 0.001). However, there was an inverse correlation between CRT expression and MYCN amplification (P = 0.003) with only two tumors showing both MYCN amplification and positive CRT expression.

NB tumors with clinical characteristics of younger age (≤ 1 year), early clinical stages, and detected by mass screening have a strong tendency to differentiate or regress spontaneously. This evidence suggests that NB tumors with positive CRT expression may be more likely to differentiate or regress spontaneously. In fact, there was a strong correlation between CRT expression and histologic grade of differentiation, yet only CRT expression but not histologic grade of differentiation was demonstrated to be an independent prognostic factor. This result supports the notion that CRT negatively regulates the growth of NB cells by affecting more than cell differentiation alone. In addition, positive CRT immunostaining was seen in normal brain neurons, adrenal medulla, and sympathetic ganglia. This finding suggests not only that CRT expression is involved in the differentiation and regression of NB, but also that up-regulation of CRT is required for the normal development of neuronal cells.

2.2.5 Survival analysis

2.2.5.1 Univariate and multivariate analysis

Kaplan-Meier analysis revealed that 5-year survival rate in patients with positive or negative CRT expression was 81.9% and 28.4% respectively (Figure 10). Positive CRT expression predicted a significantly better survival (P < 0.001, log-rank test). Univariate analysis further showed that younger age (≤ 1 year) at diagnosis, early clinical stages, and differentiated tumor histology strongly correlated with better survival, whereas MYCN amplification predicted a very poor outcome (Table 4). In multivariate analysis, CRT expression, in addition to clinical stage and MYCN amplification, was demonstrated to be an independent prognostic factor of NB (Table 4).

2.2.5.2 Significance of CRT expression in the prognostic discrimination

For further understanding the significance of CRT expression in the prognostic discrimination, the survival of NB patients was stratified according to the CRT expression, clinical stage and MYCN amplification (Figure 11). In 24 NBs of early stages (stage 1, 2, or 4S), 20 had positive CRT expression and 4 had negative CRT expression. All of them survived except 2 surgical mortalities. In 44 patients with advanced-stage disease (stage 3 or 4), the prognosis could be clearly distinguished by CRT expression. While in the category of advanced-stages and positive CRT expression, only 5 of 12 patients (42%) died of their disease. In the category of advanced-stages and negative CRT expression, 27 of 32 patients died, indicating an 84% risk of an unfavorable outcome. Thereby positive CRT expression predicted a favorable chance of survival in the patients with advanced-stage disease (Figure 12A, P = 0.007, log-rank test). Among 15 NBs with MYCN amplification, only 2 had positive CRT expression. All 15 patients with MYCN amplification died of NB except 2 were alive with residual tumor at 5 and 7 months follow-up. On the other hand, among 53 NBs without MYCN amplification, positive CRT expression distinguished the patients with favorable outcomes from those with unfavorable outcomes (Figure 12B, P < 0.001, log-rank test). In the category of normal MYCN copy number and positive CRT expression, only 4 of 30 patients (13%) died, whereas in the category of normal MYCN copy number and negative CRT expression, 17 of 23 patients (74%) died of their disease.
Clinical stage and MYCN amplification are two well-known prognostic factors of NB. Nevertheless only 35% (24/68) of NB patients were detected at early stages and fewer patients (22%, 15/68) were found to have tumors with MYCN amplification. In addition, since not all patients with advanced-stage disease had very poor prognoses (5-year survival rate 32%, Table 2) and not all patients whose tumors had no MYCN amplification had very good prognoses (5-year survival rate 67%, Table 2), these two categories of patients were actually large groups of patients with clinical heterogeneity. Additional factors were required to further distinguish the prognoses of these patients. Our results showed that among patients with advanced-stage disease and those with tumors without MYCN amplification, positive CRT expression distinguished the patients with favorable prognoses from those with unfavorable prognoses (Figure 12). These results indicate that immunohistochemical study of CRT may provide complementary prognostic information, which in turn may be helpful to the determination of the most appropriate intensity of therapy.

2.2.6 Association of CRT and Trk-A expression
Expression of Trk-A, a nerve growth factor (NGF) receptor, has been shown to be associated with the differentiation and death of the NB cells, as well as favorable prognosis of NB patients. Trk-A, after activation by NGF, may promote intracellular signaling cascades, including the Ras/ERK protein kinase pathway, the PI3K/Akt kinase pathway, and PLC-γ1. Activated PLC-γ1 acts to hydrolyze phosphatidylinositides to generate diacylglycerol, which may further activate the PKC-δ. PKC-δ in turn is required for activation of the ERK cascade and for neurite outgrowth. Interestingly, it has been shown that CRT is a substrate and binding protein for all PKC isoforms, suggesting that CRT plays an important role in the common PKC activated signaling pathway. Thus, it is conceivable that CRT may participate in the process of Trk-A mediated NB cell differentiation and death. This notion is further supported by the evidence that CRT and Trk-A expression showed a combination effect on the NB patient survival in our series.
2.2.7 CRT expression in endothelial cells

It was very interesting to find a specific expression of CRT in the endothelial cells by our immunohistochemistry study. A parallel expression of CRT in both endothelial cells and neuroblastic cells was also observed. More than half of the tumor samples that had positive CRT staining in their neuroblastic cells also had positive staining in their endothelial cells. Three tumors had CRT expression only in endothelial cells but not in neuroblastic cells and were designated as having negative CRT expression in the analysis. Two of these 3 tumors were GNB, and the remaining one was an undifferentiated NB detected by mass screening. These findings suggest that CRT expression in the endothelial cells may also be relevant to the differentiation and regression of NB. However, the relationship between the expression of CRT in endothelial cells and neuroblastic cells is not known. CRT has been shown to be an anti-angiogenic factor that may inhibit the growth of endothelial cells, and has been used as a target of gene therapy for cancer. Interestingly, it has also been shown that inhibition of angiogenesis may induce differentiation and apoptosis in NB. The association of CRT expression in endothelial cells and neuroblastic cells in our studies may suggest that CRT is a potential target for the treatment of NB by both inhibiting endothelial cells and promoting differentiation and regression of NB cells.

2.2.8 Summary

CRT is a unique endoplasmic reticulum protein that affects many cellular functions; therefore it is possible that it is involved in many pathologic conditions, especially in cancers. CRT has been shown to exist in the nuclear matrix of human hepatocellular carcinoma and various carcinoma cell lines, and over-expressed in human breast carcinoma. However, the role and importance of CRT in these cancers are not known. Our study demonstrates for the first time that expression of CRT in NB correlates with a differentiated histology and better outcome. Detection of CRT expression in tumor tissues may be of potential use as a predictive marker for NB. However, due to the limited sample size from one single institute of this cohort, additional studies with larger patient populations are required to further elucidate the prognostic significance of CRT expression in NB.

2.3 Clinical significance of GRP78 expression in NB

2.3.1 Aim of study

To evaluate the clinical importance of GRP78 in NB, especially its role in the differentiation and regression of this tumor, and to analyze the prognostic relevance of GRP78 expression in this tumor, the expression of GRP78 in NB tumors was studied and correlated with clinicopathologic and biologic parameters.

2.3.2 Methods of study

From December 1990 to December 2003, 68 NB patients treated at the National Taiwan University Hospital and receiving complete follow-up were included in this study. Eight of the 68 patients were diagnosed without any symptom by mass screening of urinary vanillyl-mandelic acid. All diagnoses of tumors were confirmed by histologic assessment of a specimen obtained from the primary or metastatic tumor at surgery. There were 37 boys and 31 girls. The median age at diagnosis was 2.5 years (range, 0 - 11.5 years). Thirty-seven of the 68 NBs originated primarily from the adrenal gland. The histologic morphologies of NBs were classified into undifferentiated NB (Schwannian stroma poor), differentiating NB (Schwannian stroma poor, including poorly differentiated subtype), and ganglioneuroblastoma, intermixed (GNB, Schwannian stroma-rich) according to the percentage and degree of differentiation of the NB cells using the criteria of the International NB Pathology Classification.15,16 The nodular type GNB was classified into either undifferentiated or differentiating NB according to the morphology of their NB nodules, since the prognosis of this type of tumors depends mainly on their NB nodules. The differentiating NB and GNB were pooled as differentiated NB in the survival analysis. The tumors were staged according to the International NB Staging System. All stage 1 tumors were treated by surgery alone, whereas the remaining tumors were treated with a combination of surgery and chemotherapy with or without autologous bone marrow transplantation according to the patient’s risk grouping. The median follow-up after diagnosis was 43.5 months (range, 1-156 months). The overall 5-year survival in this cohort was 54.8%.

The 68 tumor specimens collected before chemotherapy were retrieved from the surgical pathology archives at National Taiwan University Hospital. Tissues were fixed in formalin and embedded in paraffin. GRP78 expression was evaluated by immunohistochemical staining with a standard streptavidin-biotin method. Immmunoreactivity of GRP78 was assessed by one pathologist who was blinded to the clinical backgrounds of the patients. To verify the specificity of anti-GRP78 antibody, a specific blocking peptide (5x of antibody, Santa Cruz, California) was added along with the primary antibody when performing positive control staining to see if the staining could be blocked specifically.

The mRNAs of GRP78 in seven NB tumor tissues were examined by real-time PCR for comparison with the immunohistochemical results.

The amount of the GRP78 proteins was examined by immunoblot analysis to compare with the immunohistochemical results. To test the specificity of the anti-GRP78 antibody used in immunohistochemistry, competition assay was performed in which the specific blocking peptide (5x of antibody) was added along with the GRP78 antibody during immunoblotting. Cell lysates from un-treated and all-trans retinoic acid-treated NB cell line (Neuro-2A, American Type Culture Collection CCl-131) were used as positive control for immunoblot analysis.

2.3.3 Expression profile of GRP78

2.3.3.1 Immunohistochemistry

We first examined the specificity of immunohistochemical staining for GRP78. In the positive control staining of a ganglioneuroma, positive GRP78 staining was seen in ganglion cells, and localized in the cytoplasm consistent with ER location of GRP78 (Fig. 13, A). No Schwannian stromal cell showed positive staining of GRP78. In the competition study, the immunostaining was blocked completely by the specific peptide (Fig. 13, B). The immunoreactivity of GRP78 was classified into four categories: “−” (no expression, no stained cells or only isolated single stained cells seen), “1+” (weak expression, around 10-35% of cells stained), “2+” (moderate expression, around 35-70% of cells stained) and “3+” (strong expression, more than 70% of cells stained) (Fig. 13, C-F). Tumors were assigned into negative GRP78 expression (“−” in immunoreactivity) and positive GRP78 expression (“1+”, “2+”, or “3+” in immunoreactivity) in the statistical analysis. Immunoreactivities (1+ to 3+) of GRP78 could be detected in 40 of the 68 NBs (58.8%).

2.3.3.2 Real-time PCR and Western blot

The results of immunohistochemistry were compared with the GRP78 mRNA and protein levels in 7 tumors as evaluated by real-time PCR and Western blot respectively (Fig. 14A, lanes a-g). Cases with high expression of GRP78 mRNA had positive immunostaining (Fig. 14A, lanes a-c,f,g) as well as definite 78-kd bands on the Western blot membrane using the same anti-GRP78 antibody (Fig. 14A, b,c,g). Conversely, cases with low expression of GRP78 mRNA showed negative immunostaining (Fig. 14A, lanes d,e) as well as low protein levels by Western blot analysis (Fig. 2A, lane d). However, there were a few cases with a discrepancy between immunohistochemical and immunoblot results, such as lanes a, e and f in Figure 2A. The stromal cells of differentiated tumors had either no or very low expression of GRP78, they might thus result in a dilution effect of proteins during immunoblotting (Fig. 14A, lanes a, f). Conversely, although NB cells of undifferentiated histology had low GRP78 protein levels, they usually had a very compact distribution which might result in higher GRP78 protein levels in the total tumor lysates (Fig. 14A, lane e). In the cell line studies, the GRP78 protein levels increased after the Neuro-2A cells induced to differentiate by retinoic acid treatment for 5 days (Figure 14A, D5). Nevertheless, untreated cell lysates still had a significant level of GRP78 protein (Figure 14A, D0). Thus, immunohistochemistry had the advantage to identify the specific cells of interest and to evaluate the overall tumor quality simultaneously when comparing with PCR and Western blot in studying such a heterogenous tumor as NB. The specificity of anti-GRP78 antibody used in immunohistochemistry was again confirmed by a competition assay, in which the GRP78 immunoblotting was blocked completely by the specific antigenic peptide (Figure 14B).

2.3.4 Association of GRP78 expression other clinicopathologic and biologic factors

The majority of NBs with positive GRP78 expression were of differentiated histologies with positive staining mainly on differentiating NB cells or ganglion cells. The intensity and percentage of positive GRP78 immunostaining correlated strongly with the differentiation of histology. As the tumor histology became differentiated, the intensity and percentage of positive GRP78 immunostaining increased simultaneously (Fig. 15, P = 0.001, chi-square test). The relationship between GRP78 expression and clinicopathologic and biologic variables of NBs is summarized in Table 5. In addition to histologic grade of differentiation, positive GRP78 immunostaining also correlated strongly with early clinical stages (stage 1, 2, 4S) (P = 0.002). However, there was an inverse correlation between GRP78 expression and MYCN amplification (P = 0.001) with only three tumors showing both MYCN amplification and positive GRP78 expression.

2.3.5 Survival analysis

2.3.5.1 Univariate and multivariate analysis

Kaplan-Meier analysis showed that patients with positive or negative GRP78 expression had a 5-year survival rate of 75.3% and 25.1% respectively (Fig. 16). Positive GRP78 expression predicted a significantly better survival (P < 0.001). Further univariate analysis showed that patient’s age ≤ 1 year, early clinical stages (stage 1, 2, or 4S), and differentiated histology (including differentiating NB and GNB) strongly correlated with better survival, whereas MYCN amplification predicted a very poor outcome (Table 6). Multivariate analysis by the Cox proportional hazard model showed that only clinical stage and GRP78 expression were independent prognostic factors (Table 7).

Biologic factors that predict a favorable outcome for NB patients are usually associated with differentiation or regression of NB cells and early clinical stages, such as HNK-1, Trk-A and H-ras. Here we demonstrated clearly that GRP78 expression in NBs strongly correlated with high grades of histologic differentiation as well as other favorable characteristics of early clinical stages and absence of MYCN amplification. Further survival analysis did show that GRP78 expression was an independent favorable prognostic factor for the NB patients. In addition, although there was a strong correlation between GRP78 expression and histologic grade of differentiation, GRP78 expression predicted a favorable prognosis in patients with either undifferentiated or differentiated histologies of NBs (Fig. 5). This finding suggests that GRP78 negatively regulates the growth of NB cells by affecting more than cell differentiation alone.

2.3.5.2 Significance of GRP78 expression in the prognostic discrimination

For further understanding the significance of GRP78 expression in the prognostic discrimination, the impact of GRP78 expression on patient survival was further analyzed according to the factors of tumor histology, clinical stage, and MYCN status. In patients with either undifferentiated or differentiated histologies of NBs, positive GRP78 expression clearly predicted a better survival (Fig. 17, A and B, P < 0.001 and P = 0.003 respectively). The prognostic categories of GRP78 expression and clinical stage are shown in Fig. 18. In 24 patients with NBs of early stages, 20 showed positive GRP78 expression. All the 24 patients survived except 2 surgical mortalities. In the patients with advanced-stage disease (stage 3 or 4), the prognosis could be clearly distinguished by GRP78 expression. While 11 of 20 patients (55%) died of their disease in the category of advanced stages and positive GRP78 expression, 21 of 24 patients (88%) died in the category of advanced stages and negative GRP78 expression. Positive GRP78 expression predicted a favorable outcome in patients with advanced-stage NBs (Fig. 19A, P = 0.001, log-rank test). Only 3 of the 14 NBs with MYCN amplification showed positive GRP78 expression. All 14 patients with MYCN amplification died of their disease except 2 were alive with residual tumor at 17 and 19 months follow-up. Among 54 patients without MYCN amplification, positive GRP78 expression distinguished the patients with favorable outcomes from those with unfavorable outcomes (Fig. 19B, P < 0.001, log-rank test). Only 8 of 37 (21.6%) patients with normal MYCN copy number and positive GRP78 expression died of their disease, however 14 of 17 (82.4%) patients with normal MYCN copy number and negative GRP78 expression died of their disease.

Clinical stage and MYCN status have great impact on the NB patient survival. Nevertheless, a great portion of patients (44/68, 64.7%) was detected only when they were in advanced stages (stage 3 and 4). This category of patients had a 5-year survival rate of 33.9%, indicating that not all patients in this category had a very poor prognosis. Furthermore, MYCN amplification could only be detected in 12 of 34 (35.3%) NBs that caused death (Fig. 18), indicating that NBs without MYCN amplification did not always carry a good prognosis. Additional factors are required to further distinguish the prognoses of these two large groups of patients with advanced stages and without MYCN amplification. Our results showed that among patients with advanced-stage disease and those with tumors without MYCN amplification, positive GRP78 expression distinguished the patients with favorable prognoses from those with unfavorable prognoses (Fig. 19). Thereby assessing GRP78 expression in NB may provide complementary prognostic information in addition to clinical stage and MYCN status, which in turn may be helpful to the determination of the most appropriate intensity of therapy.

2.3.6 Possible role of GRP78 in the differentiation and apoptosis of NB

The exact role of GRP78 in the differentiation or apoptosis of NB cells remains unknown. It has been shown that GRP78 may contribute to G1 cell cycle arrest by down-regulation of cyclin D3, a G1 cyclin.29 The G1 cell cycle arrest is a critical checkpoint for the neuronal precursor cells to further differentiate into neuronal cells if they are exposed to NGF or to go to apoptosis if they are deprived of NGF. In addition, since GRP78 plays a role in protein translocation and secretion, it might be required for the translocation of needed proteins to the outgrowing neurites during NB cell differentiation.

2.3.7 Summary

In summary, our study demonstrates that expression of GRP78 may be of potential use as a predictive marker, and may provide prognostic information in discriminating the biologic character of NTs. Therefore, evaluation of this molecular information at the time of diagnosis can help to determine the appropriate strategy of treatment. Importantly, we detect GRP78 expression on routine formalin-fixed, paraffin-embedded tumor tissues with immunohistochemistry using antibody with predetermined specificity. This method is simple and can be easily adopted by clinical laboratories in general hospitals to assist the evaluation of patients with NBs.
3. Conclusions
Our serial studies demonstrate that biologic prognostic factors have great impact on NB patient survival. In addition, there should be different surgical approaches according to different biologic risk groups. Therefore, a detail molecular analysis is essential before a major operation of advanced NB.

Due to the importance of biologic prognostic factors, we subsequently determine that two ER chaperone proteins, CRT and GRP78, are independent prognostic factors of NB. Importantly, these two factors may provide complementary prognostic value in addition to traditional prognostic factors, such as MYCN and clinical stage. CRT and GRP78 are of great potential in the clinical usage for the prognostic discrimination of NB patients.

Both CRT and GRP78 expression correlate with the differentiation of NB, and possibly contribute to the apoptosis of NB. Since the pathogenesis of NB is related to the differentiation and apoptosis of NB cells, in this sence, further investigation of the regulation of CRT and GRP78 expression in NB may shed a new light on the pathogenesis of NB. In addition, through the regulation of CRT and GRP78 expression, we may develop a new treatment for the dismal NB patients.

並列關鍵字

Neuroblastoma ； Biologic prognostic factor ； Surgery ； GRP78 ； Calreticulin

參考文獻

Aoki T, Koike T, Nakano T, Shibahara K, Kondo S, Kikuchi H, et al. Induction of Bip mRNA upon programmed cell death of differentiated PC12 cells as well as rat sympathetic neurons. J Biochem (Tokyo) 1997; 121: 122-7.

Barnes JA, Smoak IW. Glucose-regulated protein 78 (GRP78) is elevated in embryonic mouse heart and induced following hypoglycemic stress. Anat Embryol (Berl) 2000; 202: 67-74.

Beckwith JB, Martin RF. Observations on the histopathology of neuroblastomas. J Pediatr Surg 1968; 3: 106-10.

Beckwith JB, Perrin EV. In Situ Neuroblastomas: A Contribution To The Natural History Of Neural Crest Tumors. Am J Pathol 1963; 43: 1089-104.

Bernstein ML, Leclerc JM, Bunin G, Brisson L, Robison L, Shuster J, et al. A population-based study of neuroblastoma incidence, survival, and mortality in North America. J Clin Oncol 1992; 10: 323-9.