Beruflich Dokumente
Kultur Dokumente
1608
Copyright 2015 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
varian cancer is the sixth most common cause of cancerrelated death in women worldwide.1 In Scandinavia, the
incidence has been relatively stable the last 20 years and is
among the highest in the world.2 The disease is characterized
by vague and not easily recognizable physical symptoms
(bloated abdomen, pelvic and abdominal pain, and urinary
urgency). Ovarian cancer is typically diagnosed at advanced
stage, and the survival is relatively poor. In Europe and the
United States, patients diagnosed in stage I have a 5-year
survival rate of 90%. However, almost 90% of the patients
are diagnosed in advanced stages (International Federation of
Gynecology and Obstetrics IIIYIV), with an overall 5-year
survival rate of less than 30%.3 Inheritance of mutations in
the BRCA1 and BRCA2 gene is associated with increased
risk of ovarian cancer.4 Other factors that increase the risk
of ovarian cancer are age, obesity, and hormone replacement
therapy. Reproductive history with full-term pregnancy,
breastfeeding, and use of contraceptives reduces the risk
of developing ovarian cancer.5 The treatment of epithelial
ovarian cancer (EOC) has improved over the last decades by
more effective surgery in combination with optimized chemotherapy. Despite improvements in the treatment, the overall
cure rate is poor.6 Prognostic and predictive markers are
needed to optimize and personalize therapy.
The high survival rate in patients diagnosed at an early
stage warrants the need for improved diagnostic tools for early
detection. Measurement of the widely used serum tumor
marker cancer antigen 125 (CA125), followed by ultrasound
imaging of the ovaries, has been suggested as a screening
tool.7 CA125 is a high-molecular-weight glycoprotein,8
which is highly expressed in normal epithelium in the female
genital tract, mammary ducts, and in peritoneal and pleural
cavities.9Y11 Serum levels of this tumor marker vary with the
menstruation cycle12 and are often increased in pregnancy13
and common benign conditions such as endometriosis.14
Serum levels of CA125 are raised in approximately 90% of
patients with advanced EOC10 but is also elevated in patients
with other malignancies such as lung, endometrioid, breast,
and colon cancer.9 In addition, less than 50% of patients
with ovarian cancer stage I have increased serum levels of
CA125.3,9,15 The low sensitivity and specificity limit the
value of CA125 as a single screening tool, and combining it
with additional biomarkers to create a multiple biomarker
panel might be more effective.
Human epididymis protein 4 (HE4) is proposed as a
promising serum tumor marker in screening for early stage
of EOC.16 Moore et al17 investigated 8 different biomarkers
isolated and in combination with CA125. Of those, HE4
showed the highest sensitivity of 72.9% and a specificity of
95% as a single marker. In combination with CA125, the
sensitivity increased to 76.4% at 95% specificity.
HE4 is a glycoprotein that belongs to a whey acidic
protein gene family of protease inhibitors18; however, the
specific biological function remains unknown. Although HE4
is highly expressed in EOC compared with normal ovarian
epithelium, increased expression has also been demonstrated
in other malignancies and in normal tissue from the respiratory tract.16,17,19Y21 In a study among Nordic women, the
serum levels of HE4 were significantly increased by age and
Copyright 2015 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
1609
Gislefoss et al
levels. A case was defined as an individual that had been diagnosed as having an ovarian cancer in the cohort, and a control
was an individual in the cohort that was alive and free of cancer,
except for nonmelanoma skin cancer, at the time of diagnosis.
Epithelial ovarian cancer cases were identified by
linking the Janus cohort to CRN. Cases with serum samples
donated more than 10 years before date of diagnosis or less
than 3 months before diagnosis were excluded from the study.
We identified 120 cases, of whom 77 women had donated
1 sample and 43 had donated 2 or more blood samples. The
study participants were enrolled mainly during the 1970s
(15%) and 1980s (72%; Fig. 1).
FIGURE 1. Times of blood collections and ovarian cancer diagnosis. Open circles represent times of blood collections;
solid circles, time of ovarian cancer diagnosis.
1610
Copyright 2015 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
No
61
51.7
18
12
15
4
7
3
15.5
9.5
11.2
3.5
6.0
2.6
Sample Processing
Samples were thawed at room temperature, mixed well,
distributed in aliquots, and refrozen on dry ice. To reduce
uncertainty from interassay variation, the samples were randomly placed in batches locating the matched case-control
sets into the same batch. Samples were shipped on dry ice
to the analyzing laboratories. All analyses were performed
on coded samples, blinding the laboratory staff to the casecontrol status.
Serum Analyses
Serum HE4 and CA125 were analyzed at Central
Laboratory, Oslo University Hospital. HE4 was measured in
duplicates using a manual HE4 EIA (Fujirebio Diagnostics
AB, Gothenburg, Sweden). The assay is a solid-phase, noncompetitive immunoassay based on a sandwich technique
using 2 mouse monoclonal antibodies. Two kit controls with
coefficients of variation (CVs) of 15.1% and 7.0% at concentrations of 45.4 and 376.5 pmol/L, respectively, were
analyzed both in the first and last sample positions for each
batch, in line with the manufacturers instructions. In addition,
1 serum control with a CV of 9.6% at a concentration of
51 pmol/L was analyzed once for each batch.
CA125 concentration was measured in duplicate using
an in-house routine assay run on the AutoDelfia platform
(PerkinElmer). This is a 3-step immunofluorometric assay
using 2 characterized monoclonal antibodies to CA125: the
OC125-like mAb K93 (biotinylated F(ab)2-fragment) and the
M11-like mAb K101 (Eu-labeled IgG). Three serum controls
with CVs of 4.7%, 3.6%, and 4.1% at mean concentrations of
Sample Quality
Biospecimens stored for up to 10 years were used in the
present study; thus, investigation of the sample quality was
necessary. The stability of HE4 and CA125 was examined by
Spearman correlation test for dependence between serum level
of the biomarkers and storage time in the control group. Neither
of the components showed a statistically significant correlation
to storage time (results not shown). This indicates that only
negligible degradation of the biomarkers may have occurred.
Statistical Analyses
The statistical analyses were performed by using the
program SPSS (Software Package for Social Sciences) version 19.36 Spearman correlation was performed to examine
the dependence between HE4, CA125, cotinine, storage time,
age at sampling, and lag time. All serum components had to
be log transformed to obtain normal distribution. Pairwise
difference in lg (log-transformed) HE4 and lg CA125 between
case and control samples was analyzed by a linear regression
TABLE 2. Baseline characteristics of the study
participants
Cases
Control
Characteristics
No. individuals
120
174
No. individuals with
43
V
serial samples
No. samples
174
174
Age at enrollment, 50.2 (45.0Y53.6) 50.2 (45.1Y53.2)
median (IQR), y
Age at diagnosis,
55.9 (51.3Y60.7)
V
median (IQR), y
Follow-up time,
6.53 (3.9Y8.4)
V
median (IQR), y
Smoking status, cotinine level
57 (47.6%)
48 (40.0%)
Nonsmokers
G5 nmol/L
Passive smokers
25 (20.8%)
36 (30.0%)
5Y85 nmol/L
Moderate smokers
26 (21.6%)
26 (21.6%)
85Y1700 nmol/L
Heavy smokers
12 (10.0%)
10 (8.4%)
91700 nmol/L
Copyright 2015 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
1611
Gislefoss et al
Component
HE4, pmol/L
Median (IQR)
Mean (CI)
CA125, kU/L
Median (IQR)
Mean (CI)
HE4, pmol/L
Median (IQR)
Mean (CI)
CA125, kU/L
Median (IQR)
Mean (CI)
Cases at
Enrollment
(n = 120)
Controls at
Enrollment
(n = 120)
48.3 (41.1Y59.0)
50.7 (25.9Y99.2)
44.9 (36.8Y57.8)
46.7 (25.9Y84.7)
19.4 (11.3Y28.2)
19.3 (3.8Y97.0)
13.6 (8.5Y23.4)
14.2 (3.6Y55.9)
Controls Matched
to Cases With
Lag Time G2 y
(n = 21)
54.2 (47.6Y85.2)
64.8 (25.1Y167.2)
42.7 (35.2Y52.9)
43.7 (26.9Y71.2)
30.5 (16.3Y67.1)
38.0 (4.5Y323.8)
12.3 (8.1Y21.3)
11.8 (3.3Y42.8)
model using lag time and difference in lg cotinine as independent variables. Lag time up to 4 years was particularly
investigated by including only women with samples collected
less than 4 years before diagnosis in the model. A significance
level of 5% was used.
RESULTS
Two unexplained high values of CA125, 355.1 and
719.5 pmol/L sampled 7.5 and 9.4 years before diagnosis,
respectively, were excluded before statistical analyses. Median with interquartile range (IQR) and retransformed mean
with confidence intervals (CIs) for serum levels of HE4 and
CA125 are presented in Table 3.
The Spearman correlation analysis showed a high
significant positive correlation between HE4 and cotinine
for cases (0.411; P G 0.001) and controls (0.484; P G 0.001).
Age at sampling was correlated with HE4 for cases (0.204;
DISCUSSION
This nested case-control study aimed to evaluate HE4
as a biomarker for early detection of EOC. The results indicated that serum levels of HE4 may rise up to 2 years before
diagnosis, whereas CA125 seemed to increase earlier.
Age and smoking have been reported as the 2 most
important determinants that influence the HE4 levels in
healthy women. Bolstad et al22 showed that serum levels of
HE4 in healthy Norwegian woman increased by age, an increase
of 9% and 20% at 40 and 50 years compared with 20-year level.
Smoking was related to an increase in HE4 serum levels. This
corresponds well to results from another study.24 In the present
study, we assessed the impact of age by matching controls by
age and smoking by including results from cotinine measurements in the Spearman correlation and in the linear regression
analyses. Both analyses demonstrated a statistically high
TABLE 4. Multiple linear regression for pairwise difference between cases and matched control samples
Multiple Linear Regression
Difference between invasive tumor and control
Samples (n = 174)
Lg HE4
Lg CA125
1612
Coefficient
P
Standardized A
Coefficient
P
Standardized A
Constant
Difference in Lg Cotinine
0.030
0.023
0.069
G0.001
0.555
j0.054
0.013
j0.189
0.163
G0.001
Copyright 2015 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
TABLE 5. Multiple linear regression for pairwise difference between cases with lag time 3 monthsY2 years and
matched control samples
Multiple Linear Regression
Difference between invasive tumor and control
Samples (n = 22)
Lg HE4
Lg CA125
Coefficient
P
Standardized A
Coefficient
P
Standardized A
Constant
Difference in Lg Cotinine
0.147
0.002
0.080
0.010
0.537
j0.079
0.269
j0.246
0.530
G0.001
a considerably lower age-related reference limit based on measurements from 801 healthy nonsmoking women and pointed out
that including smokers would have reduced the difference.
Smoking is highly correlated with the serum level of HE4. An
average increase of more than 20% in HE4 serum concentration
in moderate smokers compared with passive smoker confirms
that smoking has a significant impact on serum level of HE4.
The serum level of HE4 does not seem to be affected by menstrual cycle45 in contrast to CA12546; however, Anastasi et al52
reported nonsignificant menstrual fluctuation in serum HE4 and
CA125 levels in woman older than 35 years.
A potential weakness in studies based on archival
samples is the stability and general validity of the samples.
Selecting an appropriate study design is therefore important.
A nested case-control design was used to investigate the
difference in biomarker level between cases and controls. We
have matched for age at sampling as the levels of HE4 and
CA125 in serum are increased by age and time for blood draw
to secure equal storage time. Measurement of biomarkers in
archival samples may be influenced by analytic batch, storage,
and freeze-thaw cycles.47,48 The interassay variation was
minimized by placing the case and the matched control in the
same batch. Matching of cases and controls was important to
decrease bias of storage time and possibly differences in
preanalytical sample handling, assuming that both cases and
controls would be equally affected. The stability of biomarker
level was investigated by Spearman correlation test and
showed that neither of the components was statistically and
significantly correlated with storage time. The impact of
freeze-thaw cycles on HE4 and CA125 has not been investigated; however, unpublished data from the Janus Serumbank
show that many proteins are robust to several freeze-thaw
cycles. Although matching may introduce selection bias in
a biomarker study, a case-control design is the preferred approach to deal with batch effects, storage effects, and freezethaw cycles.48
The search for new and better markers for early detection of EOC is important, and although there are both
affirmative and negative results with regard to the diagnostic
performance of HE4, a recent meta-study of HE4 vs CA125
concludes that HE4 is superior to CA125 for identification
of EOC in women with suspected gynecologic disease.49 A
number of studies have observed that HE4 seems to be released from tumor earlier than CA125, and that the HE4
levels are significantly higher than CA125 in the initial stages
Copyright 2015 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
1613
Gislefoss et al
ACKNOWLEDGMENT
We express our gratitude to all persons who have
donated blood to the Janus Serumbank of Norway and to all
enthusiasts that made the sample collection possible. The
Norwegian Cancer Society provided funding for the cotinine
analyses.
REFERENCES
1. Bray F, Loos AH, Tognazzo S, et al. Ovarian cancer in Europe:
cross-sectional trends in incidence and mortality in 28 countries,
1953Y2000. Int J Cancer. 2005;113:977Y990.
2. Cancer in Norway 2012 - Cancer incidence, mortality,
survival and prevalence in Norway. Oslo: Cancer Registry of
Norway; 2014.
3. Jacobs IJ, Menon U. Progress and challenges in screening for
early detection of ovarian cancer. Mol Cell Proteomics.
2004;3:355Y366.
4. King MC, Marks JH, Mandell JB, et al. Breast and ovarian
cancer risks due to inherited mutations in BRCA1 and BRCA2.
Science. 2003;302:643Y646.
5. Ovarian cancer; causes, risk factors, and prevention topics
Available at: http://www.cancer.org/cancer/ovariancancer/
detailedguide/ovarian-cancer-risk-factors. Accessed July 9, 2015.
6. Luvero D, Milani A, Ledermann JA. Treatment options in
recurrent ovarian cancer: latest evidence and clinical potential.
Ther Adv Med Oncol. 2014;6:229Y239.
7. Urban N. Screening for ovarian cancer. We now need a definitive
randomised trial. BMJ. 1999;319:1317Y1318.
8. Yin BW, Dnistrian A, Lloyd KO. Ovarian cancer antigen CA125
is encoded by the MUC16 mucin gene. Int J Cancer.
2002;98:737Y740.
9. Bast RC Jr. Status of tumor markers in ovarian cancer screening.
J Clin Oncol. 2003;21:200sY205s.
10. Bast RC Jr, Klug TL, St John E, et al. A radioimmunoassay
using a monoclonal antibody to monitor the course of epithelial
ovarian cancer. N Engl J Med. 1983;309:883Y887.
11. Makar AP, Kristensen GB, Kaern J, et al. Prognostic value of
pre- and postoperative serum CA 125 levels in ovarian cancer:
new aspects and multivariate analysis. Obstet Gynecol.
1992;79:1002Y1010.
12. Brumsted JR, McBean JH, Deaton JL, et al. CA-125 secretion
by luteal phase endometrium in vitro. Hum Reprod.
1990;5:682Y684.
13. Sarandakou A, Protonotariou E, Rizos D. Tumor markers in
biological fluids associated with pregnancy. Crit Rev Clin Lab
Sci. 2007;44:151Y178.
1614
Copyright 2015 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
Copyright 2015 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
1615