Background
Epithelial ovarian cancer (EOC) is a common gynecological malignant tumor which seriously threatens women’s health [
1]. Despite high remission rates following extensive surgical resection of all visible tumors and subsequent postoperative adjuvant carboplatin and paclitaxel chemotherapy, the survival for patients with advanced EOC is still pessimistic because the disease-free interval is short and often transient [
2]. The primary obstacle in treatment of EOC remains as resistance to platinum-based chemotherapy [
3,
4]. Consequently, with the perspective of personalized treatment, there is an apparent need to introduce novel, strategic tools for the selection of patients. The prediction of patient with EOC who is resistant to first-line chemotherapy can allow making the correct selection of drugs that function via other mechanisms, and discovery of novel therapeutic strategies.
Serum cancer antigen 125 (CA125) is well known and has been used for a long time, representing the “gold standard” biomarker for EOC [
5,
6], and has played an important role in the clinical diagnosis, optimal surgery indication and chemotherapy response assessment of patients with EOC. However, there are some limitations, such as elevated CA125 which is only found in about 50% of stage I-II patients with EOC. Further, a consensus has not yet been reached in predicting chemotherapy sensitivity and survival by preoperative serum CA125 level in patients with EOC [
7‐
10]. As the tumor biomarkers for predicting chemo-resistance remains limited, the determination of novel biomarkers will be beneficial in strengthening surveillance of the disease and make reasonable clinical decisions, presenting the best opportunity for successful treatment and improved outcome.
Growth differentiation factor 15 (GDF15), which is also commonly called macrophage inhibitory cytokine-1 (MIC-1), is a secreted cytokine of TGF-β superfamily [
11‐
13]. Numerous studies have shown that serum GDF15 was significantly increased and correlated with clinical stage, lymphonodus involvement, and poor prognosis in a variety of epithelial malignancies, such as, esophageal squamous cell carcinoma, colorectal cancer, non-small cell lung cancer and gynecological malignancy [
14‐
27]. Furthermore, several studies have implied that elevated presurgerical serum GDF15 may be valuable, serving as an indicator in assessment of the treatment response [
24,
27]. However, the potential roles of serum GDF15 as a candidate predictor of chemo-resistance and clinical outcome have not yet been investigated comprehensively. Thus in our present research, we attempted to identify the relationship of presurgerical serum GDF15 with clinicopathological characters, as well as, the response to first-line chemotherapy and progression-free survival (PFS) outcomes in patients with EOC, and compare the results with those of CA125 level measurements.
Discussion
The identification of non-invasive biomarkers for predicting the response of EOC patients to chemotherapy will improve prognosis and represent an important direction for reduction of mortality resulting from EOC [
29,
30]. To date, CA125 is the most established serologic marker in advanced ovarian cancer [
5,
8,
31]. Therefore, in our present study, CA125 was used as a benchmark biomarker. Additionally, we evaluated GDF15 as a serum tumor biomarker regarding the detection of EOC, especially, as a prognostic marker for prediction of chemotherapy resistance. To our present knowledge, our research stands as the first to investigate potential clinical performance of serum GDF15 in EOC patients in both diagnosis and during post-operative chemotherapy, and its capacity is complemented when offered to CA125 negative individuals and early-stage patients.
First, we found that patients with EOC have a much higher serum GDF15 compared with normal controls as previously reported [
23], suggesting that serum GDF15 could be applied as a potential seromarker for differential detection of EOC. Previously study also indicated that median plasma GDF-15 concentration was elevated in ovarian cancer as compared to women with benign ovarian tumors (
p < 0.001); additionally, GDF-15 plasma concentration correlated inversely with survival time and was an independent predictor of survival, after correction for FIGO stage and age (
p = 0.01). In this study, we also found that serum GDF15 in early-stage EOC patients (FIGO stage I/II) were dramatically higher than healthy subjects, and the serum GDF15 remained elevated with the clinical stage of the tumor, compared to controls. These results indicate that highly elevated levels of serum GDF15 might have occurred at the early stage and corresponded to the progression of EOC. In addition to ovarian cancer, a study reported that high plasma GDF15 was significantly associated with FIGO stage III/IV disease and lymph node metastases (
p < 0.001) in large validation cohort of endometrial carcinomas, further suggested that GDF15 may be significantly associated with lymph node metastasis, possibly applicable for all gynecological tumors which worth further exploration [
32]. Nevertheless, the consequence or biological roles of serum GDF15 elevation in EOC remain to be elucidated [
33]. Until now, the research hotspots have been directed to explore the strength of GDF15 as a potential diagnostic molecular biomarker for EOC. A few publications have demonstrated the performance of serum GDF15 in distinguishing malignant and benign ovarian tumors [
24,
34‐
36]. In our research, GDF15 indicated high sensitivity of 85.3% at 88.3% specificity, consistent with reports from previous studies regarding serum GDF15 in EOC, with slightly differences in detective sensitivity, possibly associated to patient characteristics [
23]. Moreover, we demonstrate parallel diagnostic specificity and sensitivity. Compared with CA125, GDF15 showed a higher diagnostic sensitivity in early-stage EOC; while the median preoperative serum GDF15 level of the 13 patients with stage I-II EOC was only 604.6 pg/mL, with a range of 222.1–2681.8 pg/mL. This reveals a possible limitation in early detection of the disease because, at these stages, there are fluctuated values in low-level positivity, making it difficult to determine significance.
In our present study, we also found that CA125 and GDF15 levels were mutually complementary and their composite determination could significantly enhance the sensitivity which obtained from individual biomarker alone. The panel with the GDF15 and CA125 (Fig.
2a) acquired by the model of logistic regression demonstrated better diagnostic accuracy (AUC = 0.957) in differencing EOC from normal population. GDF15 improves the utilization of CA125 as a serum biomarker in EOC, and simultaneously using these two biomarkers enhances the sensitivity in EOC. Therefore, it is favorable to combine GDF15 with clinical proven biomarker CA125 to discriminate EOC from normal with high accuracy. We conclude that GDF15 is a promising, noninvasive seromarker and may be a valuable supplement to serum biomarkers available in use.
Monitoring optimal debulking would facilitate a more precise prognostic stratification, and contribute in improving the efficacy of multimodal therapy [
37,
38]. Our results revealed that presurgerical elevated GDF15 level was related to clinical advanced FIGO stages and lymphonodus metastasis in EOC patients. In present study, the serum level of GDF15 and CA125 in patients with EOC was detected and compared before and after surgical treatment. We found that CA125 serum concentrations were significantly lower following cytoreductive therapy, while GDF15 did not show this pattern, suggesting that CA125 is better than GDF15, relative to in-vivo tumor load in ovarian cancer, that is, CA125 levels were significantly associated with postoperative residual tumor size. These give us warnings that patients with elevated serum GDF15 levels may have the possibility of metastases, while CA125 predict worse debulking outcome, which may help gynecologist to determine eligible surgical patients.
It is well-established that biomarker used to predict clinical chemotherapeutic response has great significance in helping EOC patient’s management [
29,
30]. Adjuvant chemotherapy has improved PFS in various malignancies [
39,
40]. Currently, there is no efficient method for predetecting the clinical response of EOC cases to chemotherapy. Therefore, all EOC patients are customarily treated with first-line chemotherapy without regard for the adverse events, which do not represent a rational and reasonable approach [
41]. It would be favorable to hold an advance opportunity to better tailor individual treatment to EOC patient, especially, those with resistance to platinum, and it would be a milestone for successful treatment of women with EOC. In this study, we found that presurgerical serum GDF15 levels significantly correlated with chemosensitivity of EOC, that is, a high level of GDF15 was an indication that there is a high risk of resistance, which could not be discovered in the previous study [
23]. The prediction of chemosensitivity before treatment could help the clinician select the appropriate chemotherapy to customize individualized treatment strategy, which can cast light on the failures and successes of treatment in EOC cases and also supply an important basis for individualized therapy, reckoning the primary aim of our study. In addition, the result from follow-up data showed GDF15 levels in patients negatively correlated with PFS (
P = 0.0004). Multivariate Cox’s analysis indicated that serum GDF15 is an independent predictor for EOC (
P = 0.026). We will continue to expand and extend the period for obtaining follow-up samples, to explore the relationship between GDF15 and prognosis thoroughly.
There are several limitations concerning our present study. Firstly, we did not include the serum GDF15 level data during or after chemotherapy, hence, we were unable to analyze their connection with progression and remission of the disease. Secondly, this study is limited in its retrospective feature and the fact that it was not possible to obtain samples in strictly consecutive patients, which might have led to a potential selection bias. Therefore, further prospective research is needed to achieve the association between serial detections of GDF15 during or following chemotherapy and sound clinical outcomes in EOC patients. Notwithstanding its limitations, high numbers of patients with optimal surgical debulking and the thorough clinical staging followed by platinum-based chemotherapy was included in this mono-institutional retrospective research, which could be considered as strengths of our study. Additionally, present study involved homogeneous patients who underwent similar extensive surgical procedures and consistent platinum-based chemotherapy regimens, thus could heighten the generalizability of the findings.