Novel algorithm including CA-125, HE4 and body mass index in the diagnosis of endometrial cancer

YGYNO-976823; No. of pages: 7; 4C: Gynecologic Oncology xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Gynecologic Oncology journal homepage: www.elsevier.com/locate/ygyno

Novel algorithm including CA-125, HE4 and body mass index in the diagnosis of endometrial cancer☆ Tamara Knific a, Joško Osredkar b, Špela Smrkolj c, Irena Tonin c, Katja Vouk a, Andrej Blejec d, Snježana Frković Grazio e, Tea Lanišnik Rižner a,⁎ a

Institute of Biochemistry, Faculty of Medicine, University of Ljubljana, 1000 Ljubljana, Slovenia Clinical Institute of Clinical Chemistry and Biochemistry, University Medical Centre Ljubljana, 1000 Ljubljana, Slovenia Division of Obstetrics and Gynaecology, University Medical Centre, 1000 Ljubljana, Slovenia d National Institute of Biology, 1000 Ljubljana, Slovenia e Department of Pathology, Division of Obstetrics and Gynaecology, University Medical Centre, 1000 Ljubljana, Slovenia b c

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Serum CA-125 and HE4 levels are significantly higher in endometrial cancer patients. • Model to distinguish endometrial cancer from benign gynaecological diseases is proposed. • Diagnostic algorithm includes serum CA-125 and HE4 levels and BMI (AUC = 0.80). • Serum HE4 levels differentiate patients with lymphovascular invasion (AUC = 0.81). • Serum HE4 levels stratify patients with deep myometrial invasion (AUC = 0.78).

a r t i c l e

i n f o

Article history: Received 29 May 2017 Received in revised form 3 July 2017 Accepted 11 July 2017 Available online xxxx Keywords: Endometrial cancer Biomarkers CA-125 HE4 Algorithm

a b s t r a c t Objectives. To evaluate the diagnostic and prognostic potential of preoperative serum CA-125 and HE4 levels in patients with endometrial cancer. Methods. Prospective case–control study of 133 women who underwent surgical treatment at the University Medical Centre Ljubljana (64 patients with endometrial cancer, 69 control patients with prolapsed uterus or myoma). Serum CA-125 and HE4 levels were determined using electrochemiluminescent assays. Results. Serum CA-125 and HE4 levels were significantly higher in patients with endometrial cancer, compared to the controls (p = 2.67 × 10−4, 1.36 × 10−7, respectively). A diagnostic model that combines serum CA-125 and HE4 levels and body mass index separated patients with endometrial cancer from controls, with AUC of 0.804, sensitivity of 66.7%, and specificity of 84.6%. Serum HE4 levels showed good prognostic potential and stratified the patients according to presence/absence of deep myometrial invasion (p = 0.001) or lymphovascular invasion (p = 0.003), with AUCs of 0.78 and 0.81, respectively. In low-risk patients with grade

Abbreviations: AIC, Akaike Information Criteria; AUC, area under the curve; BMI, body mass index; ECLIA, electrochemiluminescent immunoassay; ROC, receiver operating characteristics. ☆ Contributions: TLR and JO designed the study; TLR contributed to the final version of the manuscript. TK performed the experiments and the statistical analysis, and drafted the manuscript. SFG evaluated the histology and tumour grading. ŠS and IT collected the patient clinical data. AB and KV supervised the statistical modelling and contributed to the final version of the manuscript. All of the authors reviewed the manuscript before submission.Funding: This study was supported by a J3-6799 Grant to T.L.R., and a Young Researcher Grant to T.K., both from the Slovenian Research Agency.Conflict of interest: The authors have nothing to disclose. ⁎ Corresponding author at: Institute of Biochemistry, Faculty of Medicine, University of Ljubljana, Vrazov trg 2, 1000 Ljubljana, Slovenia. E-mail address: [email protected] (T.L. Rižner).

http://dx.doi.org/10.1016/j.ygyno.2017.07.130 0090-8258/© 2017 Elsevier Inc. All rights reserved.

Please cite this article as: T. Knific, et al., Novel algorithm including CA-125, HE4 and body mass index in the diagnosis of endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.07.130

2

T. Knific et al. / Gynecologic Oncology xxx (2017) xxx–xxx

Prognosis Logistic regression

1 and 2 endometrioid cancer for whom lymphadenectomy can be avoided, HE4 allowed stratification according to deep myometrial invasion (p = 3.39 × 10−4), with AUC of 0.84. Although median HE4 levels were higher in patients with lymphovascular invasion, this difference did not reach significance (p = 0.06). Conclusions. A model based on preoperative serum CA-125 and HE4 levels and body mass index has good diagnostic accuracy for separation of patients with endometrial cancer and control patients. In patients with endometrial cancer, serum HE4 levels allow prediction of deep myometrial and lymphovascular invasion. © 2017 Elsevier Inc. All rights reserved.

1. Introduction Endometrial cancer is the fifth most common gynaecological malignancy in Slovenia, with the highest incidence in women between the ages of 50 and 64 years [1]. Worldwide, endometrial cancer accounts for 76,000 deaths and there are an estimated 320,000 new cases per year [2]. Risk factors for the development of endometrial cancer include obesity, diabetes mellitus, arterial hypertension, early menarche and/or late menopause, anovulation, nulliparity, hyperandrogenism, use of exogenous oestrogens unopposed by progestins, excessive endogenous production of oestrogens, and genetic predisposition [3]. Additionally, the high-risk population includes patients treated with tamoxifen, which is the standard therapy for the majority of the 1.6 million women with breast cancer who are identified yearly worldwide [2], and of the patients with Lynch syndrome, with over 1.0 million cases in Europe alone [4]. The standard treatments for endometrial cancer include hysterectomy and bilateral salpingo-oophorectomy [5] however, there is still no consensus regarding lymphadenectomy as a part of the complete surgical staging procedure, as it carries additional risk for developing lymphedema, deep vein thrombosis, and neurological and vascular injuries [6]. Blood biomarkers that can predict the presence of deep myometrial and lymphovascular invasion preoperatively would thus allow selection of patients who will need lymphadenectomy [7, 8]. Clinical assessment and imaging techniques have been exploited as prognostic markers of endometrial cancer [9], although they lack sensitivity and specificity for the detection of deep myometrial invasion and extra-uterine disease [10,11]. These shortcomings would be overcome with a non-invasive disease-specific biomarker. Although several prognostic biomarkers for endometrial cancer have been studied, with some of them indicated promising results, none of these have been accepted into clinical practice [9,12,13]. Cancer antigen 125 (CA-125) is an epithelial-cell surface antigen that is expressed in many different types of tumour cells, and it is usually used in the evaluation and follow-up processes for patients with ovarian cancer [14]. In 1984, Niloff reported elevated CA-125 levels in patients with recurrent and advanced endometrial cancer [15]. Indeed, in recent years there have been several studies that have evaluated CA-125 as a potential diagnostic and prognostic biomarker for endometrial cancer [13,16]. Considering its greater prognostic than diagnostic characteristics, CA-125 has been suggested as a prognostic marker for preoperative evaluation of patients with endometrial cancer [17,18]. There has also been a growing interest in the role of human epididymis protein 4 (HE4) in endometrial cancer. Serum HE4 levels are elevated in patients with serous and endometrioid ovarian carcinomas, and also in those with recurrent ovarian cancer [19–22]. Serum HE4 levels are also higher in patients with endometrial cancer, compared to healthy controls, and they show potential as a diagnostic biomarker for endometrial cancer [20–22]. Several studies have reported superior diagnostic characteristics of HE4 in combination with CA-125, as compared to CA-125 alone, for the detection of endometrial cancer, where addition of age further increased the diagnostic value [23]. The published studies thus support the potential of both of these biomarkers in preoperative evaluation of endometrial cancer patients. The aims of the present study were two-fold: (i) to evaluate preoperative serum CA-125 and HE4 levels in endometrial cancer and control patients, and to examine their diagnostic and/or prognostic potential as

single biomarkers and as the combination of both biomarkers with patient clinical characteristics; and (ii) to construct a model with the optimal diagnostic and/or prognostic characteristics. To the best of our knowledge there have been only four case–control studies that investigated the diagnostic and prognostic potentials of serum CA-125 and HE4 levels combined with statistical modelling [21–24]. 2. Methods 2.1. Patient enrolment Patient enrolment took place from June 2012 to December 2014 at the Department of Obstetrics and Gynaecology, University Medical Centre Ljubljana, Slovenia. In this case–control study, we included 133 women who underwent surgical treatment. Based on clinico-histopathological findings, the study participants were stratified as patients with endometrial cancer (n = 64) and as the control group of women with prolapsed uterus or myoma (n = 69). Within a week prior to surgery, morning blood samples were collected and additional information was obtained regarding their life-style and gynaecological and clinical status (Table 1). For sample collection and processing, strict and detailed standard operating procedures were followed, and serum samples were stored at −80 °C until further analysis. The study was approved by the National Medical Ethics Committee of the Republic of Slovenia (Nr. 0120-127/2016-2), and all of the participants signed written informed consent before being included in this study. The depth of myometrial invasion was assessed by two independent gynaecological pathologists (SFG, MB). The Cancer Registry of the Republic of Slovenia was searched for the vital status of the patients, with the cut-off point of 15 November 2016. 2.2. Measurement of CA-125 and HE4 Specimens were analysed using electrochemiluminescent immunoassays (ECLIAs) specific for CA-125 and HE4, on a Cobas e411 immunoassay analyser (Roche Diagnostics GmbH, Manheim, Germany). The measuring ranges for serum CA-125 and HE4 levels with these ECLIAs were 0.600–5000 U/mL and 15.0–1500 pmol/L, respectively, and the quality control data are present as Supplementary Data (Supplementary Table 1). CA-125 quantitative determination kits (REF: 11776223190, LOT: 139788-01) and HE4 detection kits (REF: 05950929190, LOT: 112732-01) were used, along with calibrators, such as CA125IICalSet (REF: 07030207190 LOT: 134830-01), HE4 CalSet (REF: 05950945190, LOT: 187553-01), ElecsysPreciControl HE4 (REF: 05950953190, LOT: 187791-02), and ElecsysPreciControlTumor Marker (REF: 11776452, LOT: 187283-02). 2.3. Statistical analysis The data was numerically anonymised and was collected in Microsoft Office Excel 2013 spreadsheets. All the data was first analysed descriptively. Serum levels were compared between the case and control groups using two-sided Wilcoxon rank-sum tests (Mann-Whitney U tests), as the data was not normally distributed. Non-parametric Kruskal-Wallis H tests were used for comparisons of more than two groups. Pairwise Wilcoxon tests with Holm corrections were used as post-hoc tests to determine the difference within each group. Fisher's

Please cite this article as: T. Knific, et al., Novel algorithm including CA-125, HE4 and body mass index in the diagnosis of endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.07.130

T. Knific et al. / Gynecologic Oncology xxx (2017) xxx–xxx

exact tests were used for comparisons of categorical variables. Statistical significance was set at p b 0.05. Receiver operating characteristic (ROC) curves were used to compare the separation of patients with endometrial cancer and patients with benign conditions using the biomarkers. A basic logistic regression model was used, where associations between each biomarker level and endometrial cancer were tested. Additionally we calculated the effects of the clinico-pathological characteristics as possible confounders. On the basis of differences shown in descriptive analysis we decided which potential confounding variables will be tested in the model. Subset analysis was performed to distinguish between patients without and with myometrial invasion and lymphovascular invasion. Inclusion of age, menopausal status, arterial hypertension, diabetes, smoking status and medication intake did not show any significant confounding effects and did not improve the overall fit of the model. On the basis of the Akaike Information Criteria (AIC), a diagnostic model was selected that included both serum markers and body mass index (BMI). Survival analysis was carried out by the Kaplan-Meier method and log-rank tests, to determine the differences between the survival curves. The survival model was fitted using the Cox proportional hazards regression model with Efron approximation. For statistical analysis, R Studio (RStudio Team 2015), R-package Epicalc (Chongsuvivatwong, 2010) and a survival package were used.

3

3. Results 3.1. Clinical characteristics of the patient cohort The cases group comprised 64 patients with endometrial cancer, with mean age of 64.8 ± 8.3 years (range, 46–88 years) and mean BMI of 31.9 ± 7.3 kg/m2 (range, 20.6–58.5 kg/m2). Detailed clinical characteristics of patients with endometrial cancer are presented in Table 1 and Table 2. The mean number of lymph nodes obtained during lymphadenectomy was 19.5 (range, 1–41). The mean number of tumour-positive lymph nodes was 3.6 (range, 1–13; n = 7), 53 patients did not have tumourpositive lymph nodes, and for one patient this information was missing. Lymphadenectomy was not carried out for three patients, and lymphovascular involvement was seen for nine patients (14.1%). According to the International Federation of Gynaecology and Obstetrics classification (34), 43 patients (67.2%) had stage IA, 11 (17.2%) had stage IB, two (3.1%) had stage IIIA, one (1.6%) had stage IIIB, four (6.3%) had stage IIIC, and two (3.1%) had stage IVB. The mean follow-up time for these patients was 39.9 months (36.0 ± 12.1 months; n = 61), and six patients (9.8%) died with a mean time of 10.4 months. Four (6.6%) of these died due to malignant neoplasms of the corpus uteri, one (1.6%) due to pancreatic malignant neoplasm, and for one patient (1.6%) this information was not available. For the remaining 55 patients (90.2%) there were no records of disease recurrence, and the median diseasefree survival was 40.2 months (mean, 38.1 ± 9.5 months).

Table 1 Detailed clinical characteristics of the study participants.

Age b49.9 years 50–59.9 years 60–69.9 years 70-79.9 years N80 years BMI Normal 18.6–24.9 Overweight 25–29.9 Class I obesity 30–34.9 Class II obesity 35–39.9 Class III obesity 40–49.9 Class IV obesity 50–59.9 NA Menopausal status Premenopausal Postmenopausal NA Diabetes Yes No NA Arterial hypertension Yes No NA Hormonal therapy Yes No NA Medication intake Yes No NA Smoking status Regular smoker Occasional smoker Non-smoker Former smoker NA

pa values

Control, n = 69 (100%)

Case, n = 64 (100%)

5 (7.25%) 20 (28.99%) 23 (33.33%) 20 (28.99%) 1 (1.45%)

2 (3.13%) 16 (25.00%) 27 (42.19%) 17 (26.56%) 2 (3.13%)

ns

19 (27.54%) 25 (36.23%) 15 (21.74%) 5 (7.25%) 1 (1.45%) 0 4 (5.80%)

12 (18.75%) 18 (28.13%) 16 (25.00%) 9 (14.06%) 7 (10.94%) 1 (1.56%) 1 (1.56%)

0.002

6 (8.70%) 55 (79.71%) 8 (11.59%)

3 (4.69%) 60 (93.75%) 1 (1.56%)

ns

10 (14.49%) 44 (63.77%) 15 (21.74%)

12 (18.75%) 45 (70.31%) 7 (10.94%)

ns

25 (36.23%) 29 (42.03%) 15 (21.74%)

36 (56.25%) 21 (32.81%) 7 (10.94%)

ns

0 50 (72.46%) 19 (27.54%)

2 (3.13%) 44 (68.75%) 18 (28.13%)

ns

48 69.57%) 13 (18.84%) 8 (11.59%)

46 (71.88%) 13 (20.31%) 5 (7.81%)

ns

6 (8.70%) 4 (5.80%) 46 (66.67%) 10 (14.49%) 3 (4.35%)

5 (7.81%) 2 (3.13%) 51 (79.69%) 6 (9.38%) 0

ns

pa values were calculated using non-parametric Wilcoxon-Mann-Whitney test or Kruskal-Wallis H test for continuous variables and Fisher's exact test for categorical variables, ns = not significant.

Please cite this article as: T. Knific, et al., Novel algorithm including CA-125, HE4 and body mass index in the diagnosis of endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.07.130

4

T. Knific et al. / Gynecologic Oncology xxx (2017) xxx–xxx

Table 2 Serum CA-125 and HE4 levels in patients with endometrial cancer and control patients. Patient group

n (%)a

pb

CA-125 (kU/L) Median

Mean ± SD

Range

64 (48.1) 69 (51.9)

19.58 13.07

38.13 ± 108.86 15.03 ± 6.89

5.46–880.10 6.70–35.82

57 (89.1) 5 (7.8) 1 (1.6)

18.41 47.41

22.38 ± 16.07 191.24 ± 339.66

Endometrioid carcinoma differentiation Well differentiated G1 38 (66.7) Moderately differentiated G2 13 (22.8) Poorly differentiated G3 6 (10.5)

16.34 22.36 30.39

Myometrial invasion No invasion b50% into the myometrium N50% into the myometrium

14 (21.9) 34 (53.1) 15 (23.4)

Metastasis Without metastasis With metastasis

Disease status Cancer Benign Histology Type I: Endometrioid Type II: Serous Clear cell

Lymphovascular invasion No Yes Total a b

pb

HE4 (pmol/L) Median

Mean ± SD

Range

b0.001

82.81 57.80

127.94 ± 191.81 63.46 ± 30.43

40.95–1500.00 32.83–259.50

b0.001

5.46–108.60 12.64–880.10

0.01

81.52 130.91

99.87 ± 69.58 390.16 ± 563.66

40.95–455.30 57.08–1500.00

ns

18.27 ± 9.62 25.86 ± 13.79 40.89 ± 34.35

5.46–46.39 10.68–49.75 13.67–108.60

0.03

76.53 85.73 122.55

90.70 ± 55.68 118.97 ± 107.45 116.57 ± 40.56

40.95–337.20 45.65–455.30 64.30–175.70

ns

19.83 18.15 22.36

22.62 ± 14.02 46.33 ± 147.83 35.43 ± 34.42

6.29–46.76 5.46–880.10 12.27–126.10

ns

64.42 80.33 119.20

69.34 ± 25.09 130.10 ± 244.64 175.98 ± 133.93

40.95–127.10 43.62–1500.00 64.25–455.30

0.001

53 (82.8) 10 (15.6)

18.18 35.39

21.36 ± 12.01 129.12 ± 266.85

5.46–49.75 8.14–880.10

0.03

79.79 155.45

91.20 ± 48.10 320.01 ± 441.76

40.95–337.20 46.36–1500.00

ns

54 (84.4) 9 (14.1)

18.61 25.80 19.58

21.81 ± 12.28 138.38 ± 281.35 38.13 ± 108.86

5.46–49.75 12.27–880.10 5.46–880.10

ns

77.74 175.70 82.81

88.08 ± 39.31 364.13 ± 453.78 127.94 ± 191.81

40.95–227.00 64.25–1500.00 40.95–1500.00

0.003

, for one patient the histology information is not available. , p values were calculated using non-parametric Wilcoxon-Mann-Whitney tests or Kruskal-Wallis H tests with post-hoc tests and Holm correction; ns, not significant.

Fig. 1. Box plots comparing serum CA-125 and HE4 levels between the different study groups. Box-and-whisker plots with median, minimum and maximum and circles indicating outliers for serum CA-125 (A, C, E) and HE4 (B, D, F-H) levels. (A, B) Control patients and patients with endometrial cancer. (C, D) Control patients and patients with endometrioid endometrial cancer. (E, F) Control patients and patients with two histological subtypes of endometrial cancer. (G) Absence or presence of deep myometrial invasion. (H) Absence or presence of lymphovascular invasion. Serum levels of the biomarkers are represented on a logarithmic scale. * p b 0.05, ** p b 0.01, *** p b 0.001.

Please cite this article as: T. Knific, et al., Novel algorithm including CA-125, HE4 and body mass index in the diagnosis of endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.07.130

T. Knific et al. / Gynecologic Oncology xxx (2017) xxx–xxx

5

The initial basic model included HE4 levels (AUC = 0.765). Inclusion of serum CA-125 levels (AUC = 0.790) and addition of BMI further improved the model (AUC = 0.804). The model with HE4, CA-125 and BMI showed the best diagnostic characteristics, with adjusted odds ratios of 11.0, 2.4 and 23.1, respectively, and adjusted p-values of 0.00022, 0.03147 and 0.00733, respectively (Fig. 2, Table 3). Addition of other variables did not provide any significant improvement of the model, including for age, diabetes, arterial hypertension, hormonal therapy, smoking status, and recent medication intake. We also looked at how well the model performs after we excluded 3 patients that have been on hormonal therapy or oral contraceptives. The AUC value slightly increased (from 0.804 to 0.814) while there were no significant changes in terms of sensitivity, specificity, positive and negative predictive value of the proposed diagnostic model.

3.3. HE4 levels allow prediction of deep myometrial invasion and lymphovascular invasion

Fig. 2. Diagnostic ROC curve. ROC curve exhibits improving effects of adding variables to the model for differentiation between endometrial cancer patients and control patients.

The control group included 69 patients with mean age of 63.5 ± 9.3 years (range, 44–83 years) and mean BMI of 28.3 ± 4.8 kg/m2 (range, 20.3–42.2 kg/m2). Detailed description of control patients is presented in Table 1. There was a statistically significant difference in BMI between the case and control groups (p = 0.002), but there were no differences in age distribution, menopausal status, hormonal therapy, medication intake, smoking status, diabetes or arterial hypertension (Table 1).

3.2. A model with serum CA-125 and HE4 levels and body mass index has good diagnostic characteristics The results of Wilcoxon rank-sum tests showed that serum CA-125 and HE4 levels significantly differed between the case and control patients (p = 2.67 × 10−4, 1.36 × 10−7, respectively; Table 2, Fig. 1). Further subgroup analysis according to BMI (BMI b 30 kg/m2 versus BMI N 30 kg/m2) revealed significant differences between cases and controls for HE4 levels, while serum CA-125 levels were significantly different only in the group with BMI b 30 kg/m2. This was confirmed with logistic regression models, which showed crude odds ratios of 13.1 (p = 0.00001) for HE4, 3.5 (p = 0.00044) for CA-125, and 21.9 (p = 0.00226) for BMI. Differences in serum CA-125 and HE4 levels according to menopausal status, diabetes and arterial hypertension are presented in Supplementary Table 2.

To evaluate the prognostic value of serum CA-125 and HE4 levels, subgroup analysis was performed for the cases group and results are presented in Table 2. When G1 and G3 tumours were compared for patients with endometriod endometrial cancer, there was a significant difference in serum CA-125 levels (p = 0.02) (Fig. 1). The stratification of patients according to deep myometrial invasion (N50% of myometrium, yes/no) revealed that only serum HE4 levels differentiated between the patients according to this subclassification, as there was no significant difference for CA-125 (Fig. 1). Serum HE4 levels enabled differentiation among: (i) patients without myometrial invasion; (ii) patients with b50% myometrial invasion; and (iii) patients with N50% myometrial invasion (p b 0.05). Also stratification according to lymphovascular invasion (yes/no) showed significant differences only for serum HE4 levels (p = 0.003) (Fig. 1). The prognostic model was constructed based on information on deep myometrial invasion (yes/no) and lymphovascular invasion (yes/ no) as the binary outcome. The initial model included only serum HE4 levels. Addition of other variables that were obtained before surgery and that might improve a prognostic model did not show any significant impact on the model, including for serum CA-125 levels, age, BMI, diabetes, arterial hypertension and tumour differentiation. Serum HE4 levels thus provided the best prognostic potential in terms of differentiating patients according to the depth of their myometrial invasion (AUC = 0.776) and the presence of lymphovascular invasion (AUC = 0.810) (Fig. 3). The prognostic value was also evaluated for serum CA-125 and HE4 levels in low-risk patients with grades 1 and 2 endometrioid endometrial cancer, where lymphadenectomy can be avoided. Serum HE4 levels significantly differed when patients were stratified according to deep myometrial invasion (yes/no) (p = 3.39 × 10−4) with AUC of 0.839, sensitivity of 90.9% and specificity of 72.5%. In patients with lymphovascular invasion median serum HE4 levels were higher, but this difference did not reach significance (p = 0.06).

Table 3 Diagnostic and prognostic models of preoperative serum CA-125 and HE4 levels. Model

Prediction

HE4 CA125 BMI HE4 HE4

Disease status⁎

Crude OR (95%CI) Crude p-value Adjusted OR (95% CI) Adjusted p-value AUC

13.1 (4.2, 40.8) 3.5 (1.7, 7.0) 21.9 (3, 158.3) Myometrial invasion 3.9 (1.3, 11.4) Lymphovascular invasion 11.5 (2.4, 55.3)

0.00001 0.00044 0.00226 0.01341 0.00230

11.0 (3.1, 39.5) 2.4 (1.1, 5.4) 23.1 (2.3, 230.0) NA NA

0.00022 0.03147 0.00733 NA NA

Sensitivity (%) Specificity (%)

Predictive value (%) Positive

Negative

0.804

66.7

84.6

80.8

84.6

0.776 0.81

86.7 66.7

66.7 81.5

44.8 37.5

66.7 81.5

⁎ Algorithm for the diagnostic model: ln (odds) = 0.884 (ln[CA-125]) + 2.402 (ln[HE4]) + 3.142 (ln[BMI]) − 23.284).

Please cite this article as: T. Knific, et al., Novel algorithm including CA-125, HE4 and body mass index in the diagnosis of endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.07.130

6

T. Knific et al. / Gynecologic Oncology xxx (2017) xxx–xxx

4. Discussion

Fig. 3. Prognostic ROC curves. ROC curves based on logistic regression modelling for prognostic value according to the presence of myometrial invasion or lymphovascular invasion in patients with endometrial cancer.

3.4. Serum HE4 levels are significantly associated with patient survival For the Kaplan-Meier survival analysis, the patients were divided into two groups according to the calculated optimal cut-off points: serum CA125, 33.8 kU/L; and serum HE4 levels, 175.7 pmol/L. The results showed that only serum HE4 levels were significantly associated with patient survival (p = 0.01), while serum CA-125 levels did not have any significant impact on patient survival (Supplementary Fig. 1). For the Cox proportional hazard model, serum CA-125 and HE4 levels were included in the model as continuous variables, whereby the serum levels of both of these biomarkers had a statistically significant influence on patient survival (CA-125: hazard ratio, 1.003; 95% confidence interval, 1.001– 1.006; HE4: hazard ratio, 1.002; 95% confidence interval, 1.001–1.004; p b 0.05 for both). Age, BMI, tumour differentiation, lymphovascular invasion and myometrial invasion did not improve the overall fit of the model. As expected, lymphovascular invasion itself had a strong impact on overall survival of the cancer patients (hazard ratio, 28.3; 95% confidence interval, 3.2–253.9; p = 0.003). The proportional hazards assumption was satisfied in all three survival models.

3.5. Two patients with extreme serum CA-125 and HE4 levels There was an outlier patient for the serum levels of both biomarkers, with high serum CA-125 (880 kU/L) and HE4 (above the measuring range) levels. This patient was 62 years old, a non-smoker, obese and postmenopausal. Histology revealed serous endometrial carcinoma with b50% myometrial invasion, but with lymphovascular invasion and metastases on the ovary and in the lymph nodes, peritoneum, sigma and bowel. Additionally, the tumour had invaded the cervix and was massively necrotic. The patient died within 2 years of the operation. Another patient with similar clinico-pathological characteristics also had highly elevated serum CA125 and HE4 levels. This patient had serous endometrial carcinoma, but with N50% myometrial invasion and metastasis in the omentum, connective and adipose tissues, istmus and lymph nodes, with cancerous lymphovascular invasion in the ovary. This patient had the shortest endometrial cancer related survival, as she died 4 months after the surgery.

Pathohistological examinations remain the gold standard for diagnosis of endometrial cancer as there are still no valid non-invasive biomarkers or any panel of biomarkers that might accurately predict the presence and extent of endometrial cancer [13]. Diagnostic biomarkers would be helpful for screening asymptomatic high-risk women, such as primarily patients with breast cancer treated with tamoxifen, patients with metabolic syndrome, or patients with Lynch syndrome [25,26]. Prognostic biomarkers that would stratify patients with high risk for progression from those with good prognosis would be invaluable in clinical practice, to assist clinicians in the management and treatment of patients with endometrial cancer. Currently, there are no validated diagnostic or prognostic biomarkers for endometrial cancer although algorithms including CA-125 and HE4 show great potential [13,21,23,24,27]. Over the past 6 years, there have been several studies in patients with endometrial cancer that have examined the diagnostic [16] and prognostic [17,18,28,29] values of CA-125 as a single biomarker or in combinations with other tumour markers [30,31]. Although these studies have revealed the limited diagnostic accuracy of CA-125, they also indicated its prognostic value, as elevated serum CA-125 levels correlated with higher stage, higher grade, increased depth of myometrial invasion, lymph node metastases, presence of lymphovascular space involvement, and extra-uterine spread [17,18]. The combination of serum CA-125 levels with serum HE4 levels, compared to serum CA125 levels alone, showed better diagnostic [22,24,32] and similar prognostic [33–35] potential as a biomarker for endometrial cancer. In the present study, the best diagnostic characteristics were seen for a model/algorithm with serum CA-125 and HE4 levels and BMI. This algorithm stratified patients with endometrial cancer and control patients with the relatively high AUC of 0.80. In line with published data, the present results indicate that serum HE4 levels can differentiate between patients with lymphovascular invasion (p b 0.05) and deep myometrial invasion (p b 0.05), while serum CA-125 levels lack this kind of prognostic value. The present data support the study of Brennan et al. [33] that showed that serum HE4 levels with a cut-off of 70 pmol/L were a better predictor for deep myometrial invasion (AUC = 0.76) than serum CA-125 levels (AUC = 0.65). Importantly, our study indicate that HE4 levels differentiate patients with low-grade (G1 or G2) cancer with deep myometrial invasion from patients without invasion with an AUC of 0.839, sensitivity of 90.9% and specificity of 72.5%. The logistic regression model of Brennan et al. [36] that combined age with serum CA-125 and HE4 levels predicted disease recurrence with an AUC of 0.87, but did not improve the sensitivity and specificity of serum HE4 levels alone. In the present case– control study, there were no recurrences of cancer in 90.2% of the patients, and thus prediction of disease recurrence was not examined. Published studies have reported that high serum CA-125 levels correlate with lymph node metastasis [13], which was also shown in the present study. The associations of high serum CA-125 and/or HE4 levels with extra-uterine spread of endometrial cancer, and with shorter patient survival have also been reported previously [12,22,23,28,30], and the case of two patients with serous endometrial cancer described above support these findings. In summary, the present study shows that serum CA-125 and HE4 levels have good diagnostic potential, while serum HE4 levels show superior prognostic value compared to serum CA-125 levels. A model that combines preoperative serum CA-125 and HE4 levels and BMI has the optimal diagnostic characteristics for separation of patients with endometrial cancer from control patients, with AUC of 0.804, sensitivity of 66.7%, and specificity of 84.6%. As a prognostic biomarker, serum HE4 levels have the optimal characteristics for prediction of myometrial invasion, with AUC of 0.776, and the presence of lymphovascular invasion, with AUC of 0.810. Preoperative serum HE4 levels enable stratification of these patients into different risks groups, which might help clinicians to identify patients who would benefit from more aggressive treatment,

Please cite this article as: T. Knific, et al., Novel algorithm including CA-125, HE4 and body mass index in the diagnosis of endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.07.130

T. Knific et al. / Gynecologic Oncology xxx (2017) xxx–xxx

while limiting the overtreatment of those who are at low risk for extrauterine spread and disease recurrence. Validation on a larger sample size would allow evaluation of the models within a subgroup of patients with different histological types, grade, stage, depth of myometrial invasion, present lymphovascular invasion and tumour diameter. Acknowledgements The authors thank the study participants, the personnel of the Department of Obstetrics and Gynaecology, University Medical Centre, Ljubljana, Slovenia, and especially Milena Osredkar and Dr. Leon Meglič, MD, for their support in the enrolling of the study participants. The authors also thank Mrs. Vera Troha Poljančič at the University Medical Centre Ljubljana, Clinical Institute for Clinical Chemistry and Biochemistry for processing the samples, Mrs. Teja Fabjan for her help with the experiments, Dr. Matej Bračko MD for consulting in the histological evaluation, and Dr. Chris Berrie for critical reading of the manuscript. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ygyno.2017.07.130. References [1] Cancer in Slovenia, Epidemiology and Cancer Registry, 2016, Institute of Oncology Ljubljana, Ljubljana: Ljubljana, 2012 2012. [2] J. Ferlay, I. Soerjomataram, R. Dikshit, S. Eser, C. Mathers, M. Rebelo, et al., Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012, Int. J. Cancer 136 (2015) E359–86. [3] Y. Sonoda, R.R. Barakat, Screening and the prevention of gynecologic cancer: endometrial cancer, Best Pract. Res. Clin. Obstet. Gynaecol. 20 (2006) 363–377. [4] H. Hampel, A. de la Chapelle, The search for unaffected individuals with Lynch syndrome: do the ends justify the means? Cancer Prev. Res. (Phila.) 4 (2011) 1–5. [5] P. Morice, A. Leary, C. Creutzberg, N. Abu-Rustum, E. Darai, Endometrial cancer, Lancet 387 (2016) 1094–1108. [6] G. Bogani, S.C. Dowdy, W.A. Cliby, F. Ghezzi, D. Rossetti, A. Mariani, Role of pelvic and para-aortic lymphadenectomy in endometrial cancer: current evidence, J. Obstet. Gynaecol. Res. 40 (2014) 301–311. [7] W.T. Creasman, C.P. Morrow, B.N. Bundy, H.D. Homesley, J.E. Graham, P.B. Heller, Surgical pathologic spread patterns of endometrial cancer. A gynecologic oncology group study, Cancer 60 (1987) 2035–2041. [8] S. Capriglione, F. Plotti, A. Miranda, R. Ricciardi, G. Scaletta, A. Aloisi, et al., Utility of tumor marker HE4 as prognostic factor in endometrial cancer: a single-center controlled study, Tumour Biol. 36 (2015) 4151–4156. [9] H.M. Werner, H.B. Salvesen, Current status of molecular biomarkers in endometrial cancer, Curr. Oncol. Rep. 16 (2014) 403. [10] B. Dobrzycka, B. Mackowiak-Matejczyk, K.M. Terlikowska, M. Kinalski, S.J. Terlikowski, Utility of HE4 to identify patients with endometrioid endometrial cancer who may require lymphadenectomy, Adv. Med. Sci. 61 (2016) 23–27. [11] I.S. Haldorsen, H.B. Salvesen, What is the best preoperative imaging for endometrial cancer? Curr. Oncol. Rep. 18 (2016) 25. [12] I. Mutz-Dehbalaie, D. Egle, S. Fessler, M. Hubalek, H. Fiegl, C. Marth, et al., HE4 is an independent prognostic marker in endometrial cancer patients, Gynecol. Oncol. 126 (2012) 186–191. [13] T.L. Rizner, Discovery of biomarkers for endometrial cancer: current status and prospects, Expert. Rev. Mol. Diagn. 16 (2016) 1315–1336. [14] N. Zamani, M. Modares Gilani, F. Zamani, M.H. Zamani, Utility of pelvic MRI and tumor markers HE4 and CA125 to predict depth of myometrial invasion and cervical involvement in endometrial cancer, J Family Reprod Health 9 (2015) 177–183. [15] J.M. Niloff, T.L. Klug, E. Schaetzl, V.R. Zurawski Jr., R.C. Knapp, R.C. Bast Jr., Elevation of serum CA125 in carcinomas of the fallopian tube, endometrium, and endocervix, Am. J. Obstet. Gynecol. 148 (1984) 1057–1058.

7

[16] B.W. Kim, Y.E. Jeon, H. Cho, E.J. Nam, S.W. Kim, S. Kim, et al., Pre-treatment diagnosis of endometrial cancer through a combination of CA125 and multiplication of neutrophil and monocyte, J. Obstet. Gynaecol. Res. 38 (2012) 48–56. [17] J. Nicklin, M. Janda, V. Gebski, T. Jobling, R. Land, T. Manolitsas, et al., The utility of serum CA-125 in predicting extra-uterine disease in apparent early-stage endometrial cancer, Int. J. Cancer 131 (2012) 885–890. [18] Y.L. Chen, C.Y. Huang, T.Y. Chien, S.H. Huang, C.J. Wu, C.M. Ho, Value of pre-operative serum CA125 level for prediction of prognosis in patients with endometrial cancer, Aust. N. Z. J. Obstet. Gynaecol. 51 (2011) 397–402. [19] R. Drapkin, H.H. von Horsten, Y. Lin, S.C. Mok, C.P. Crum, W.R. Welch, et al., Human epididymis protein 4 (HE4) is a secreted glycoprotein that is overexpressed by serous and endometrioid ovarian carcinomas, Cancer Res. 65 (2005) 2162–2169. [20] R.G. Moore, A.K. Brown, M.C. Miller, D. Badgwell, Z. Lu, W.J. Allard, et al., Utility of a novel serum tumor biomarker HE4 in patients with endometrioid adenocarcinoma of the uterus, Gynecol. Oncol. 110 (2008) 196–201. [21] E. Bignotti, M. Ragnoli, L. Zanotti, S. Calza, M. Falchetti, S. Lonardi, et al., Diagnostic and prognostic impact of serum HE4 detection in endometrial carcinoma patients, Br. J. Cancer 104 (2011) 1418–1425. [22] L. Zanotti, E. Bignotti, S. Calza, E. Bandiera, G. Ruggeri, C. Galli, et al., Human epididymis protein 4 as a serum marker for diagnosis of endometrial carcinoma and prediction of clinical outcome, Clin. Chem. Lab. Med. 50 (2012) 2189–2198. [23] S.L. Antonsen, E. Hogdall, I.J. Christensen, M. Lydolph, A. Tabor, A. Loft Jakobsen, et al., HE4 and CA125 levels in the preoperative assessment of endometrial cancer patients: a prospective multicenter study (ENDOMET), Acta Obstet. Gynecol. Scand. 92 (2013) 1313–1322. [24] B. Omer, S. Genc, O. Takmaz, A. Dirican, Z. Kusku-Kiraz, S. Berkman, et al., The diagnostic role of human epididymis protein 4 and serum amyloid-A in early-stage endometrial cancer patients, Tumour Biol. 34 (2013) 2645–2650. [25] B. Trabert, N. Wentzensen, A.S. Felix, H.P. Yang, M.E. Sherman, L.A. Brinton, Metabolic syndrome and risk of endometrial cancer in the united states: a study in the SEERmedicare linked database, Cancer Epidemiol. Biomark. Prev. 24 (2015) 261–267. [26] K.H. Lu, M. Dinh, W. Kohlmann, P. Watson, J. Green, S. Syngal, et al., Gynecologic cancer as a “sentinel cancer” for women with hereditary nonpolyposis colorectal cancer syndrome, Obstet. Gynecol. 105 (2005) 569–574. [27] S.K. Saarelainen, N. Peltonen, T. Lehtimaki, A. Perheentupa, M.H. Vuento, J.U. Maenpaa, Predictive value of serum human epididymis protein 4 and cancer antigen 125 concentrations in endometrial carcinoma, Am. J. Obstet. Gynecol. 209 (142) (2013) e1–6. [28] A. Chao, Y.H. Tang, C.H. Lai, C.J. Chang, S.C. Chang, T.I. Wu, et al., Potential of an agestratified CA125 cut-off value to improve the prognostic classification of patients with endometrial cancer, Gynecol. Oncol. 129 (2013) 500–504. [29] B.P. Goksedef, H. Gorgen, S.Y. Baran, M. Api, A. Cetin, Preoperative serum CA 125 level as a predictor for metastasis and survival in endometrioid endometrial cancer, J. Obstet. Gynaecol. Can. 33 (2011) 844–850. [30] E. Myriokefalitaki, G. Vorgias, G. Vlahos, A. Rodolakis, Prognostic value of preoperative Ca125 and Tag72 serum levels and their correlation to disease relapse and survival in endometrial cancer, Arch. Gynecol. Obstet. 292 (2015) 647–654. [31] E. Baser, T. Gungor, C. Togrul, O. Turkoglu, S. Celen, Preoperative prediction of poor prognostic parameters and adjuvant treatment in women with pure endometrioid type endometrial cancer: what is the significance of tumor markers? Eur. J. Gynaecol. Oncol. 35 (2014) 513–518. [32] J. Presl, Z. Novotny, O. Topolcan, P. Vlasak, R. Kucera, R. Fuchsova, et al., CA125 and HE4 levels in a Czech female population diagnosed with endometrial cancer in preoperative management, Anticancer Res. 34 (2014) 327–331. [33] D.J. Brennan, A. Hackethal, A.M. Metcalf, J. Coward, K. Ferguson, M.K. Oehler, et al., Serum HE4 as a prognostic marker in endometrial cancer—a population based study, Gynecol. Oncol. 132 (2014) 159–165. [34] R.G. Moore, C.M. Miller, A.K. Brown, K. Robison, M. Steinhoff, G. Lambert-Messerlian, Utility of tumor marker HE4 to predict depth of myometrial invasion in endometrioid adenocarcinoma of the uterus, Int. J. Gynecol. Cancer 21 (2011) 1185–1190. [35] N. Prueksaritanond, P. Cheanpracha, M. Yanaranop, Association of serum HE4 with primary tumor diameter and depth of myometrial invasion in endometrial cancer patients at Rajavithi hospital, Asian Pac. J. Cancer Prev. 17 (2016) 1489–1492. [36] D.J. Brennan, A. Hackethal, K.P. Mann, I. Mutz-Dehbalaie, H. Fiegl, C. Marth, et al., Serum HE4 detects recurrent endometrial cancer in patients undergoing routine clinical surveillance, BMC Cancer 15 (2015) 33.

Please cite this article as: T. Knific, et al., Novel algorithm including CA-125, HE4 and body mass index in the diagnosis of endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.07.130