Is serum survivin expression a predictive biomarker in locally advanced gastric cancer patients treated with neoadjuvant chemotherapy?

Abstract

BACKGROUND:

The potential prognostic value of survivin is variably reported depending on the gastric cancer.

OBJECTIVE:

Evaluation of the prognostic and predictive significance of serum survivin and its relation with survival and treatment response rates in patients with locally advanced gastric cancer (LAGC).

METHODS:

Serum samples were prospectively collected from 50 patients with newly diagnosed LAGC. Serum samples of 32 healthy subjects were also collected as control groups for survivin levels. Serum survivin levels were evaluated at baseline and after three cycles of neoadjuvant chemotherapy in LAGC patients.

RESULTS

: Median survivin level was 147 IU/L (range $=$ 4.4–4936) at baseline and was 27 IU/L (range $=$ 4.2–4737) after neoadjuvant chemotherapy. The difference between survivin levels of the control group (26 IU/L, range $=$ 3.8–1430) and pre-treatment patient group was statistically significant ( 0.001). Clinical response to mDCF regimen was classified as progressive (progressive disease) and non-progressive groups (partial response $+$ stable disease). Baseline survivin levels were similar between patients in progressive and non-progressive groups ($p=$ 0.55). Survivin levels were significantly reduced after chemotherapy in non-progressive group ( 0.001). In contrast, serum survivin levels increased in a stepwise fashion from baseline to post-chemotherapy in patients with progressive disease ($p=$ 0.06). Patients were divided into low and high survivin groups according to baseline median survivin levels. Median DFS was 12.4 and 14.6 months for low and high groups, respectively ($p=$ 0.18). Moreover, median OS was 14.4 and 24.9 months for low and high group, respectively ($p=$ 0.14).

CONCLUSION:

It can be suggested that serum survivin can be used as a predictor of response to chemotherapy- but not survival- in LAGC patients receiving neoadjuvant mDCF chemotherapy. However, large multicenter prospective studies are required to confirm these results.

1. Introduction

The only approved curative therapy for locally advanced gastric cancer patients is radical surgical resection. Cure rates remain low even after wide surgical resection and 5-year survival is less than 25% [1]. Improved outcomes were obtained in patients treated with multimodality regimens including neoadjuvant, adjuvant chemotherapy, radiotherapy or chemo-radiotherapy in addition to surgical resection. In previous studies, neoadjuvant mDCF (modified Docetaxel- Cisplatin-Fluorouracil) chemotherapy regimen was shown to be efficient in locally advanced gastric cancer patients [2]. However, some patient groups show unfavorable disease course despite these therapies and thus there is a need for various predictive and prognostic factors. In this context, establishing predictive factors associated with treatment would allow optimization of patient selection and hence prolong expected survival.

Survivin is the smallest member of the inhibitor of apoptosis protein family (IAPs). Survivin gene which is located on the 17q25 chromosome consists of 3 introns and 4 exons and encodes a 142-amino acid protein [3, 4]. Survivin protein is involved in cell proliferation and angiogenesis in addition to anti-apoptotic activity [5]. It is expressed in G2/M phase of cell cycle to promote cell division [6]. Survivin is expressed in fetal tissues and in majority of tumors while it is expressed in negligible quantity in differentiated adult tissues [7, 8, 9]. Survivin was suggested to promote tumor growth and progression by inhibiting apoptosis and increasing cell proliferation [10]. Studies have reported that survivin is prognostic for various cancer [3, 11, 12, 13]. In a meta-analysis of gastric cancer, a relation between survivin expression levels and overall survival have been emphasized [14].

The aim of this study was to evaluate the relation of pre- and post-chemotherapy tumor survivin expression to therapeutic response and prognosis in patients with locally advanced gastric cancer who underwent surgery following neoadjuvant mDCF.

2. Patients and methods

Locally advanced gastric cancer patients who were admitted to SBU Ankara Numune Training and Research Hospital between September 2013 and June 2015 were included in this prospective study. Patients who had histopathologically proven gastric adenocarcinoma, those at clinically and radiologically confirmed locally advanced stage (regional lymph nodes measuring short axis $⩾$ 15 mm in the computed tomography), receiving mDCF chemotherapy for neoadjuvant therapy and aged 18 years and older were included. Patients ineligible for surgery after neoadjuvant chemotherapy, those with a history of previous chemotherapy, patients who did not complete 3 cycles of neoadjuvant chemotherapy for any reason, patients with Eastern Cooperative Oncology Group (ECOG) scores of 3–4 and patients with impaired hematological, renal or hepatic functions were excluded from the study. Plasma samples for survivin levels were obtained from locally advanced gastric cancer patients at baseline and after 3 cycles of chemotherapy. Plasma samples were also obtained from the healthy subjects in the control group.

All patients received neoadjuvant chemotherapy with mDCF regimen [mDCF; docetaxel 60 mg/m${}^2$/day (day 1), cisplatin 60 mg/m${}^2$/day (day 1), 5-fluorouracil 600 mg/m${}^2$/day (days 1–5), every 3 weeks]. Patients underwent computerized tomography imaging before and after 3 cycles of neoadjuvant chemotherapy. Response to neoadjuvant chemotherapy was evaluated according to RESİST 1.1 criteria [15]. Patients eligible for surgery following neoadjuvant therapy were operated approximately 4–6 weeks after therapy. The study was approved by the Institutional Ethics Committee and all participants provided written informed consent.

All blood samples were obtained by venipuncture from the antecubital vein, between 8:00 AM and 9:00 AM, and following overnight fasting after 8–10 hours. The plasma was separated by centrifuging at 2500 rpm for 20 min at room temperature, within 30 minutes of blood collection. The human survivin enzyme linked immunosorbent assay (ELISA) kit (Elabscience Biotechnology Co., Ltd, PRC; Catalog No: E-EL-H1584) was used for the quantitative determination of survivin in serum. The kit uses Sandwich-ELISA as the method. The micro ELISA plate provided in this kit has been pre-coated with an antibody specific to survivin. Standards or samples are added to appropriate micro ELISA plate wells and combined with the specific antibody. Then a biotinylated detection antibody specific for Survand Avidin- Horseradish Peroxidase (HRP) conjugate is added to each microplate well and incubated. Free components are washed away. And substrate solution is added. After a short incubation, the enzyme reaction is stopped and the color generated is read at 450 nm by an automed ELISA reader (BioTek, Elx800 Chemistry Analyzer). The measured optical density is directly proportional to the concentration of survivin in either standards or samples.

2.1. Statistical analysis

Statistical analyses were performed using SPSS software version 22. Inc., Chicago, IL, USA). Data were presented as mean $\pm$ standard deviation or median and interquartile ranges, as appropriate. Categorical variables were reported as frequencies and group percentages. Differences in survivin values were analyzed by Mann-Whitney U-test. The Wilcoxon signed-rank test was used to compare the change in survivin values between baseline and 3 months after neoadjuvant chemotherapy. OS and PFS were estimated and plotted using the Kaplan-Meier product limit method. The mean and median OS with the corresponding two-sided 95% CIs was computed. A p value less than 0.05 was considered as statistically significant.

3. Results

A total of 50 locally advanced gastric cancer patients (one patient was excluded due to early operation) and 32 healthy subjects were included. Baseline demographic and tumor characteristics of the patients are presented in Table 1. Mean age was 55 (range $=$ 26–79) for locally advanced gastric cancer group and 50 (range $=$ 29–72) for healthy group. Male/Female ratios were 2.57 (36/14) and 1.28 (18/14) for patient and healthy groups, respectively. At diagnosis, all patients except one had ECOG performance scores of 0–1 and 60% ($n=$ 30) complained of weight loss (was defined as loss of more than 5 percent of usual body weight over six months). Histopathological diagnosis was poorly differentiated adenocarcinoma subtype in 70% ($n=$ 35) of patients and tumor location was proximal gastric in 56% ($n=$ 28).

Table 1.

Baseline characteristics of the study subjects

Age, years (mean, range)	55 (26–79)
Gender
Female	14 (28)
Male	36 (72)
ECOG, $n$ (%)
0	19 (38)
1–2	31 (62)
Smoking, $n$ (%)
Yes	38 (76)
No	12 (24)
Comorbidity, $n$ (%)
Yes	17 (34)
No	33 (66)
Weight loss, $n$ (%)	30 (60)
Histopathologic differentiation, $n$ (%)
Low grade (G1)	5 (10)
Medium grade (G2)	10 (20)
High grade (G3)	35 (70)
Tumor localization, $n$ (%)
Cardia or fundus	28 (56)
Corpus	14 (28)
Antrum	8 (16)
CEA (ng/mL)
High	17 (34)
Normal	33 (66)
CA 19-9 (U/ml)
Normal	39 (78)
High	11 (22)
PNI
Yes	36 (72)
No	14 (28)
LVI
Yes	33 (66)
No	17 (34)

ECOG: Eastern Cooperative Oncology Group, CEA: Carcinoembryonic Antigen, CA19-9: Carbohydrate Antigen 19-9, PNI: Perineural invasion, LVI: Lymphovascular invasion.

All patients received 3 cycles of neoadjuvant mDCF chemotherapy. According to response evaluation performed after 3 cycles of therapy, complete response was not obtained in any of the gastric cancer patients; 42% showed partial response (PR), 38% had stable disease (SD) and 20% showed progression. Total gastrectomy was performed in 70% of patients, and 30% underwent subtotal gastrectomy after neoadjuvant chemotherapy. Only 4 (16%) patients had R1 resection. Adjuvant chemotherapy and chemo-radiotherapy was given to 48% and 62% of patients, respectively.

Median serum survivin levels were 147 (minimum $=$ 4.4, maximum $=$ 4936) IU/L) measured at baseline for patients with neoadjuvant chemotherapy and was 27 (minimum $=$ 2.2, maximum $=$ 4737) IU/L) after neoadjuvant chemotherapy. The difference between survivin levels of the control group (26; minimum $=$ 3.8, maximum $=$ 1430 IU/L) and pre-treatment patient group was statistically significant ( 0.001). Patients were divided into low and high survivin groups according to baseline median survivin levels. The pathological characteristics of the patients with low and high survivin are presented in Table 2.

Table 2.

Association among baseline plasma survivin level and pathological features

	Low survivin	High survivin	$p$ value
	( 147)	($⩾$ 147)
	($n=$ 25)	($n=$ 25)
Primary tumor
T1	–	2 (8)	0.35
T2	3 (12)	3 (12)
T3	18 (72)	16 (64)
T4a	4 (16)	3 (12)
T4b	–	1 (4)
Regional lymph
node
N0	4 (16)	4 (16)	0.38
N1	1 (4)	5 (20)
N2	6 (24)	6 (24)
N3a	12 (48)	7 (28)
N3b	2 (8)	3 (12)
Resection margin
R0	21 (84)	22 (88)	0.68
R1	4 (16)	3 (12)
TNM stage
IA	–	2 (8)	0.51
IB	2 (8)	1 (4)
IIA	2 (8)	2 (8)
IIB	2 (8)	5 (10)
IIIA	4 (16)	4 (16)
IIIB	11 (44)	7 (28)
IIIC	4 (16)	4 (16)
Histopathological
grade
G1	1 (4)	4 (16)	0.08
G2	3 (12)	7 (28)
G3	21 (84)	14 (56)
Lymphovascular invasion	17 (68)	16 (56)	0.76
Perineural invasion	18 (72)	18 (72)	0.99

Clinical response to mDCF regimen was classified as progressive (progressive disease) and non-progressive groups (partial response $+$ stable disease). Table 3 represents the association between baseline and post-treatment survivin levels in the groups. Baseline survivin levels were similar between patients in progressive and non-progressive groups ($p=$ 0.55). However, survivin levels after chemotherapy were significantly lower in patients with low survivin levels ( 0.001). Accordingly, survivin levels were significantly reduced after chemotherapy in non-progressive group ( 0.001, Fig. 1). In contrast, serum survivin levels increased in a stepwise fashion from baseline to post-chemotherapy in patients with progressive disease ($p=$ 0.06).

Table 3.

Relation between overall clinical response and survivin after neoadjuvant chemotherapy

	PR $+$ SD	PD	$p^{*}$ value
	($n=$ 40)	($n=$ 10)
Baseline survivin
Median	163	92	0.55
Interquartile range	63–1171	26–395
Survivin after chemotherapy
Median	21	776	0.001
Interquartile range	13–64	116–4693
$p^{+}$value	0.001	0.06

PR: Partial response, SD: Stabil disease, PD: Progressive disease, (${}^{*}$) $p$ for Mann U Whitney, (${}^{+}$) $p$ for Wilcoxon signed rank test.

Figure 1.

Serum survivin levels changes in patients with or without clinical response.

On the other hand, survivin levels were significantly reduced after chemotherapy in patients who had partial response to neoadjuvant chemotherapy ($p=$ 0.02); but not in stabile and progressive groups ($p=$ 0.67; Table 4).

Table 4.

Relation between overall response rate and survivin after neoadjuvant chemotherapy

	PR	PD $+$ SD	$p^{*}$ value
	($n=$ 21)	($n=$ 29)
Baseline survivin
Median	255	100	0.36
Interquartile range	71–1427	58–395
Survivin after chemotherapy
Median	22	67	0.04
Interquartile range	11–43	20–295
$p^{+}$ value	0.002	0.67

PR: Partial response, SD: Stabil disease, PD: Progressive disease, (${}^{*}$) $p$ for Mann u Whitney, (${}^{+}$) $p$ for Wilcoxon signed rank test.

Median DFS was 12.4 and 14.6 months for low and high groups, respectively (log rank $p=$ 0.18). Moreover, median OS was 14.4 and 24.9 months for low and high group, respectively (log rank $p=$ 0.14). Figure 2 shows the overall and disease-free survival rates for patients with high and low survivin.

Figure 2.

Kaplan-Meier survival curves regarding to plasma survivin.

4. Discussion

In this study, as a remarkable finding, declining serum survivin levels after neoadjuvant DCF chemo- therapy regimen was related to favorable response in patients with locally advanced gastric cancer. We also observed that patients who progressed after neoadjuvant therapy significantly had higher serum survivin levels compared to baseline serum survivin levels. No significant differences in any demographic or clinic-pathological characteristic were observed between patients with high as compared to low tumor survivin expression.

Survivin inhibits apoptosis directly or indirectly through inhibition of inducing (caspase-9) and effector (caspase-3 and caspase-7) caspases. Overexpression of survivin has been reported in almost all malignancies in humans including liver, pancreas, stomach, breast, esophagus, colon, over, lungs and hematological malignancies [16, 17, 18, 19]. Increased survivin mRNA and protein levels in gastric cancer have been shown to play a role in the development of gastric cancer [20]. Survivin protein cannot be detected in normal gastric mucosa and its mRNA is low. In our study, concordant with tissue survivin levels, increased serum survivin levels were found in gastric cancer patients compared to healthy subjects ( 0.001). Thus analysis of serum survivin levels, apart from tumor tissue levels, can be used as a fast and simple alternative method.

Many studies and meta-analyses have been published associated with prognostic value of survivin in gastric cancer patients. In their study, Wang et al. have reported that survivin values were associated with more aggressive tumor biology, higher disease recurrence and shorter survival [21]. According to the meta-analysis by Krieg et al., survivin expression did not show statistically significant association with gender, depth of invasion, stage, grade of differentiation, histological type, distant metastases, lymphatic or blood vessel invasion and perineural invasion, while it was found significantly associated with lymph node metastasis and overall poor survival [14]. The meta-analysis of 16 studies, conducted by Liu et al., has reported that survivin expression (particularly cytoplasmic expression) was associated with poor overall survival [22]. In contrast to the above mentioned studies indicating poor prognostic features, there are studies showing survivin expression is not a negative prognostic factor [23, 24]. Vallböhmer et al. have found that cytoplasmic overexpression of survivin in locally advanced gastric cancer patients receiving neoadjuvant chemotherapy (cisplatin, 5-fluorouracil, leucovorin) was associated with longer overall survival [23]. Based on multivariate analysis, Okada et al. have reported that nuclear survivin expression of tumor tissue was associated with favorable prognosis while cytoplasmic survivin expression was not associated with prognosis [24]. Survivin expression was also shown not to have negative prognostic effect in breast, bladder, esophagus cancers or osteosarcoma besides gastric cancers [25, 26, 27, 28]. Similarly, in our study no statistical association was found between serum survivin values and recurrence or survival. Different results in our study and other studies might be related to lack of subcellular located survivin evaluation (due to different prognostic values of nuclear vs cytoplasmic survivin), statistical bias resulting from small sample size of the studies, different methods of evaluation (immune-histochemical, RT-PCR), heterogeneity of patient groups (early stage, locally advanced), methodological differences and evaluation based on different specimens (blood vs tissue).

Approximately 15 years ago, Ikeguchi and colleagues have detected a relation between survivin expression of gastric cancer cells and chemotherapy. These studies have reported that survivin expression could be an indicator for chemo-resistance of malignant cells [29, 30]. The study by Meng et al. have investigated the relation between survivin variant expression in gastric cancer tissue and chemo-resistance (gemcitabine, fluorouracil, cisplatin and docetaxel) and found chemo-resistance relation only between docetaxel and survivin variant (wild-type variant) [31]. A subsequent study by Meng et al. have reported chemotherapy response relation between wild type survivin variant mRNA levels and docetaxel [32]. In a study conducted with esophagus cancer patients, Beardsmore et al. have detected significant relation between survivin expression and proliferation index in locally advanced esophagus cancer patients receiving neoadjuvant chemo-radiotherapy and suggested that survivin expression can be associated with pathological complete response to neoadjuvant chemo-radiotherapy [33]. Additionally, nuclear survivin overexpression was found associated with favorable clinical response in advanced stage non-small cell lung cancer patients receiving taxan/platin [34]. The majority of the above mentioned studies were conducted with single chemotherapy agent, evaluated survivin expression of cancer tissue cells and primarily described survivin variant or survivin subclassification (mitochondrial or nuclear). Our study is unique compared to previous studies, being the first to evaluate serum survivin in patients receiving triple chemotherapy regimen and in our study, survivin levels were significantly reduced ( 0.001) in patients who had high serum survivin levels before neoadjuvant mDCF chemotherapy and showed clinical and radiological response to therapy. A reasonable explanation for better response to neoadjuvant chemotherapy in patients with high survivin levels might be predisposition of tumor cells to chemotherapeutic cell damage due to increased proliferative activity of tumor cells caused by survivin overexpression.

Limitations of our study were being a single-center study, inability to perform multivariate analysis due to limited number of patients and lack of information about the correlation between positive serum survivin and survivin expression of tumor tissue. Another limitation is evaluation of survivin in patients received only 3 cycles (not 1–2 cycles) of neoadjuvant mDCF chemotherapy.

5. Conclusion

It can be suggested that survivin can be used as a predictor of response to chemotherapy- but not survival- in locally advanced gastric cancer patients receiving neoadjuvant mDCF chemotherapy. However, large multicenter prospective studies are required to confirm these results.