Xu Y, Shen Z, Wiper DW, Wu M, Morton RE, Elson P, Kennedy AW, Belinson J, Markman M, Casey G. Lysophosphatidic Acid as a Potential Biomarker for Ovarian and Other Gynecologic Cancers. JAMA. 1998;280(8):719-723. doi:10.1001/jama.280.8.719
From the Departments of Gynecology and Obstetrics (Drs Xu, Wiper, Kennedy, Belinson, and Casey), Cancer Biology (Drs Xu, Shen, and Casey and Ms Wu), Cell Biology (Dr Morton), and Biostatistics (Dr Elson) and the Cancer Center (Drs Xu, Markman, Casey, and Elson), Cleveland Clinic Foundation, and the Department of Chemistry, Cleveland State University (Dr Shen), Cleveland, Ohio.
Context.— Lysophosphatidic acid (LPA) has been shown to stimulate proliferation
of ovarian cancer cells and is present in the ascitic fluid of patients with
Objectives.— To determine whether elevated levels of LPA are present in plasma from
patients with ovarian cancer and other gynecologic malignancies compared with
healthy controls and to evaluate whether an elevated LPA plasma level may
be a biomarker for these diseases.
Design.— A research assay was used to measure total LPA levels in plasma from
healthy women and women with different diseases. All LPA assays and comparison
of LPA levels and CA125 (an ovarian cancer biomarker) levels were performed
by observers blinded to patient status or group.
Setting.— The Cleveland Clinic Foundation.
Participants.— A convenience sample of 48 healthy control women, 48 women with ovarian
cancer, 36 women with other gynecologic cancers, 17 women with benign gynecologic
diseases, 11 women with breast cancer, and 5 women with leukemias.
Main Outcome Measures.— Total LPA levels in plasma samples from patients and controls.
Results.— Patients in the ovarian cancer group had significantly higher plasma
LPA levels (mean, 8.6 µmol/L; range, 1.0-43.1 µmol/L) compared
with the healthy control group (mean, 0.6 µmol/L; range, <0.1-6.3
µmol/L) (P<.001). Elevated plasma LPA levels
were detected in 9 of 10 patients with stage I ovarian cancer, 24 of 24 patients
with stage II, III, and IV ovarian cancer, and 14 of 14 patients with recurrent
ovarian cancer. Of 36 patients with other gynecologic cancers, 33 also showed
higher LPA levels (mean, 14.9 µmol/L; range, <0.1-63.2 µmol/L),
compared with healthy controls (P<.001). Elevated
plasma LPA levels were detected in 5 of 48 controls and 4 of 17 patients with
benign gynecologic diseases and in no women with breast cancer or leukemia.
In comparison, among a subset of patients with ovarian cancer, 28 of 47 had
elevated CA125 levels, including 2 of 9 patients with stage I disease.
Conclusions.— Plasma LPA levels may represent a potential biomarker for ovarian cancer
and other gynecologic cancers. However, these findings are preliminary and
require confirmation in larger studies.
PATIENTS WITH ovarian cancer have the highest mortality rate among women
with gynecologic cancers, with an estimated 14500 deaths from ovarian cancer
in 1998 in the United States.1 More than two
thirds of patients with ovarian cancer have widespread metastatic disease
at initial diagnosis.1 The outlook for women
with advanced disease remains poor, with a 5-year survival rate of no more
than 15%.2 This dismal outcome is due, at least
in part, to the failure to detect the disease at stage I, when the long-term
survival rate may approach 90%.1,2
Methods for earlier detection are essential to improve prognosis and overall
survival of patients with ovarian cancer.
The CA125 remains the most widely used biomarker for the detection and
management of epithelial ovarian cancer, even though this marker is not highly
sensitive and lacks specificity. For example, CA125 is not consistently elevated
in serum from patients with early-stage ovarian cancer and may be elevated
in patients with benign gynecologic diseases.2,3
Measurement of serum CA125 in conjunction with ultrasound screening as a second-line
test confers higher specificity but detects only about half of stage I ovarian
cancers.4 Although other markers have been
developed,5- 7 none
has proved to be sufficiently sensitive for widespread use.
Previous reports have shown that ascitic fluid from patients with ovarian
cancer can stimulate the proliferation of ovarian cancer cells both in vitro
and in vivo.8,9 We recently purified
and characterized a factor from the ascites of patients with ovarian cancer.
This factor is comprised of various species of lysophosphatidic acid (LPA),
termed ovarian cancer activating factor.10- 12 The LPA stimulates
the proliferation of cancer cells, intracellular calcium release, and tyrosine
phosphorylation, including mitogen-activated protein kinase activation.10- 12 This suggests that
ovarian cancer activating factor or LPA may play a biological role in ovarian
cancer cell growth. The LPA has been shown to be a multifunctional signaling
molecule in fibroblasts and other cells.13- 15
In this study, we analyzed plasma LPA levels of women with ovarian cancer
and other malignant and benign diseases to determine whether elevated plasma
LPA levels represent a biomarker for gynecologic malignancies.
We enrolled a convenience sample of patients who were seen in the Department
of Gynecology and Obstetrics at the Cleveland Clinic Foundation, Cleveland,
Ohio, during 2 periods, June 1995 to January 1996 and July 1996 to April 1997.
Patients with breast cancer or leukemia were seen at the Cleveland Clinic
Cancer Center and were enrolled in the study in December 1996. All female
patients with cancer who visited the Department of Gynecology and Obstetrics
at the Cleveland Clinic Foundation during the defined periods were regarded
as eligible for entry into the study. No patients who were asked refused to
participate. Whole blood specimens were obtained from patients with ovarian
cancer, including 10 patients with stage I disease, 24 patients with stages
II, II, and IV, and 14 patients with recurrent ovarian cancer. Blood specimens
were obtained from patients with other gynecologic cancers, including 15 patients
with primary peritoneal papillary serous adenocarcinoma, 15 patients with
endometrial cancer, and 6 patients with cervical cancer. Seventeen patients
with benign gynecologic conditions, 11 patients with breast cancer, and 5
patients with leukemia also were enrolled.
Cancer diagnosis was confirmed for all patients following a pathologic
review of tumors. One patient with an ovarian tumor of low malignant potential
was not included in the study. Clinical stage was determined according to
International Federation of Gynecologists and Obstetricians criteria, and
the histologic subtype was evaluated according to the World Health Organization
classification.16 Whole blood specimens also
were obtained from 48 healthy female controls. Controls were identified during
the same study periods as patients with gynecologic cancer. Subjects were
pooled from 2 sources at the Cleveland Clinic Foundation: healthy volunteers
of any age and women older than 50 years without cancer who were attending
outpatient clinics for routine physical examination.
The study was approved by the Institutional Review Board of the Cleveland
Clinic Foundation. A signed informed consent was obtained from all participants.
The LPA is produced and released by activated platelets during coagulation
and therefore is a normal constituent of serum,17- 22
but it is not detectable in whole blood or fresh platelet-poor plasma from
healthy individuals.19- 22
To prevent platelet activation and phospholipase activity, blood samples were
collected in EDTA-containing tubes. Whole blood was centrifuged at 580g for 5 minutes. The supernatant was transferred to a microcentrifuge
tube and centrifuged at 8000g for 5 minutes to remove
remaining platelets. Plasma was either processed immediately or stored at−70°C
before lipid extraction.
Lipid extraction was performed at 0°C to 4°C to minimize damage
to ester bonds, using a slight modification of published methods.22 The LPA was separated from other lipids on thin-layer
chromatographic plates, which were developed using a solvent system of chloroform-methanol-ammonium
hydroxide (65:35:5.5). Sample spots were scraped from the silica gel plates
into glass centrifuge tubes. The LPA was hydrolyzed in 1-mol/L ethanolic potassium
hydroxide and transmethylated in the presence of behenic acid (internal standard)
with the boric chloride-methanol reagent (Supelco, Bellefonte, Pa). The fatty
acid methyl esters were extracted with petroleum ether, dried under nitrogen,
and dissolved in chloroform. A gas chromatography unit (model 5710A; Hewlett-Packard,
Wilmington, Del), equipped with a column (1.83×2 mm) coated with 3%
SP-2310, 2% SP-2300 on 100/120 Chromosorb WAW (Supelco), was used to analyze
LPA levels. Two standard curves were obtained using 2 fatty methyl ester standard
mixtures (Nu Check Prep Inc, Elysian, Minn).
Levels of CA125 were determined as a routine analysis from patients
with gynecologic cancers by radioimmunoassay (Abbott Laboratories, Atlanta,
Ga). The LPA levels were compared with CA125 data only if they had been performed
within the same week. Except for 1 patient with stage I disease, CA125 data
were available for all patients with ovarian cancer.
All samples were coded and analyzed by individuals who had no knowledge
of the subject's disease status. Codes were revealed on completion of the
plasma LPA analyses. Nonparametric methods were used to analyze the data because
of the limited number of patients with some cancers and the large number of
patients with undetectable LPA levels. Comparisons of patient groups were
performed using the Wilcoxon rank sum test,23
stratified by age (<50 years, 50-64 years, or >64 years; the age groups
were chosen to ensure approximately equal number of subjects in each) and
the Kruskal-Wallis test.23 A 0.1-µmol/L
LPA value was used in calculations for which LPA levels were below the level
of detection. To evaluate the diagnostic accuracy of LPA as a marker for ovarian
cancer and other gynecologic cancers, nonparametric receiver operating characteristic
curves24 were examined and a cutoff value of
1.3 µmol/L was identified as optimizing both the sensitivity and specificity
of the assay. A cutoff of 35 U/mL was used to define elevated CA125 levels.25 The McNemar test26
was used to compare the proportions of patients with ovarian cancer who had
elevated LPA levels, elevated CA125 levels, or both. All statistical significance
testing was 2-sided, and P values less than .05 were
considered to be statistically significant. Data analyses were carried out
using SAS (Statistical Analysis Software, version 6.12, SAS Institute Inc,
Cary, NC) and StatXact (version 2.04, CYTEL Software Corp, Cambridge, Mass).
The ages, stages, grades, histological types, sizes of the tumors, and
treatment status of the patients with ovarian cancer are shown in Table 1. Plasma LPA levels of patients
with ovarian cancer were significantly higher than those of healthy controls
1, Table 2). There were
no statistically significant differences in total LPA levels among patients
with primary ovarian cancer who had blood samples obtained preoperatively,
postoperatively, or postchemotherapy (Figure
1), and, therefore data from these subgroups have been combined.
The median (range) values of total LPA levels in the plasma from patients
with stage I (10 patients), stages II, III, and IV (24 patients), and recurrent
(14 patients) ovarian cancer were 2.4 (1.0-32.3) µmol/L, 5.2 (1.8-43.1)
µmol/L, and 4.1 (1.4-33.8) µmol/L, respectively, compared with
0.1 (<0.1-6.3) µmol/L for 48 controls (Table 2). Plasma LPA levels were higher than the 1.3-µmol/L
cutoff value in 47 (98%) of 48 patients with ovarian cancer. The LPA levels
were elevated in 9 (90%) of 10 patients with stage I ovarian cancer, and all
patients with stages II, III, and IV ovarian cancer (24 of 24) (P<.001 compared with controls) or recurrent ovarian cancer (14 of
14) (P<.001 compared with controls). The 1 false-negative
result occurred in a patient with a focal stage I clear cell adenocarcinoma
present in a 12-cm endometriotic cyst. The patient also had multiple other
sites of histologically documented pelvic endometriosis. There were no statistically
significant differences in LPA levels between patients with different stages
of ovarian cancer or recurrent ovarian cancer (stage I vs stages II-IV [P=.39]; stage I vs recurrent [P=.18];
stages II-IV vs recurrent [P=.58]).
Among healthy controls (median age, 49.5 years; range, 22-76 years),
elevated plasma LPA levels were detected in 5 (10%) of 48 cases, all 5 of
whom were older than 45 years (ages 46, 48, 57, 62, and 76 years) (Figure 1).
Patients with primary peritoneal (median age, 63.0 years; range, 30-73
years), endometrial (8 with stage I and 7 with stages II, III, and IV; median
age, 62.0 years; range, 38-73 years), and cervical cancers (2 with stage I
and 4 with advanced stages; median age, 52.5 years; range, 43-76 years) also
had statistically significant higher plasma LPA levels than controls (P<.001) (Figure 2, Table 2). Elevated LPA levels were detected
in 13 (87%) of 15 patients with peritoneal cancer, 14 (93%) of 15 patients
with endometrial cancer, and 6 (100%) of 6 patients with cervical cancer.
Patients with benign gynecologic diseases had statistically significant
higher levels of plasma LPA (9 leiomyoma, 7 benign adnexal masses, and 1 endometriosis;
median age, 45.0 years; range, 40-91 years) than controls (P=.004). However, plasma LPA levels above the 1.3-µmol/L cutoff
were detected in only 4 (24%) of 17 patients (Figure 3). There was a statistically significant difference between
patients with benign gynecologic disease vs patients with gynecologic cancers
No patients with breast cancer (0 of 11) (2 with stage I, 1 with stage
II, and 8 with stage III or IV; median age, 56.0 years; range, 44-74 years)
or leukemias (0 of 5) (4 with acute myelocytic leukemia and 1 with acute lymphocytic
leukemia; median age, 50.0 years; range, 39-56 years) had elevated plasma
LPA levels (Figure 3, Table 2) (P=.20 and .29, respectively,
when these 2 groups compared with controls).
Overall, patients with gynecologic cancers (n=84) had significantly
higher LPA levels compared with patients with other cancers (n=16) or no cancer
(n=48) (P<.001) (Table 2). The patients with gynecologic cancer generally were older
than other groups. However, adjusting for age, the difference between LPA
levels in patients with gynecologic cancers compared with patients with other
cancers or no cancer remained statistically significant (P<.001).
A cutoff value of 1.3 µmol/L for LPA levels was determined to
maximize the sensitivity and specificity of the test results within this study
population using the nonparametric receiver operating characteristics curve
(Figure 4). Using the cutoff of
1.3 µmol/L for LPA, the sensitivity and the specificity among all patients
with gynecologic cancers and all other subgroups (controls, patients with
benign gynecologic diseases, breast cancer, and leukemia) were 95% and 89%,
respectively. However, this cutoff value may be applicable only to this study
population and will have to be reevaluated following larger studies.
The CA125 and LPA values were compared in patients with ovarian cancers
who had blood samples drawn and assayed within the same week, using the cutoff
value of 35 U/mL for CA125 and 1.3 µmol/L for LPA. Of the 9 patients
with stage I ovarian cancer for whom data was obtained using both assays,
8 (89%) had elevated LPA levels, and 2 (22%) also had an elevated CA125 level
(P=.03). The one false-negative LPA level was in
a patient whose CA125 level (26 U/mL) was also below the specified cutoff
value (35 U/mL). Among 24 patients with stages II, III, and IV ovarian cancer,
24 (100%) had elevated LPA levels and 13 (54%) had elevated CA125 levels (P=.001). All 14 patients with recurrent ovarian cancer
had elevated LPA levels, and 12 (86%) of the 14 had elevated CA125 levels
(P=.50). Overall, 47 (98%) of 48 patients with ovarian
cancer had LPA levels higher than 1.3 µmol/L, and 28 (57%) of 47 had
CA125 levels higher than 35 U/mL.
We believe that the most important finding of this study is that elevated
plasma LPA levels were detected in patients with early-stage ovarian cancer
compared with controls. In addition, a comparison of available CA125 values
with LPA levels suggests that plasma LPA may represent a more sensitive marker
for ovarian cancer, particularly stage I disease. The plasma LPA assay offers
the possibility of earlier diagnosis of ovarian cancer, a disease that is
associated with a poor outcome mainly because it is rarely detected at early
However, a number of issues need to be addressed. Our results are preliminary
and are based on a limited study population. Further studies will be required
to determine the general usefulness of LPA as a biomarker for gynecologic
cancers and whether a combination of the LPA and CA125 assays will prove even
more useful for cancer detection. In this study, we used a 1.3-µmol/L
cutoff value to optimize for specificity and sensitivity using receiver operating
characteristic curve analysis. We recognize that this value needs to be reevaluated
following large-scale studies.
In this study, 47 (98%) of 48 patients with ovarian cancer and 80 (95%)
of 84 women with any gynecologic malignancy had elevated levels of LPA, whereas
no patients with breast cancer or leukemia showed elevated plasma LPA levels.
Five of 48 healthy female controls and 4 of 17 patients with benign gynecologic
diseases (1 with leiomyoma, 2 with benign adnexal masses, and 1 with endometriosis)
had elevated plasma LPA levels. To represent a useful test for the detection
of gynecologic cancers, any assay needs to have a low or minimal false-positive
rate. The reasons for the false-positive results in the present study are
not clear. From limited longitudinal studies, we have determined that patients
with active ovarian cancer consistently demonstrate elevated LPA levels from
serial blood LPA tests (Xu et al, unpublished data, 1998). In contrast, LPA
levels less than 1.3 µmol/L were detected on repeat tests from several
controls in this study who initially showed higher levels of LPA (only the
initial test results are reported herein).
The false-positive results we observed may suggest that other factors
influence the release of activated LPA into plasma. For example, certain diseases
may influence activated LPA release, including inflammatory processes, hypercholesterolemia,
or diabetes mellitus. Future studies therefore need to include not only prospective
or longitudinal CA125 and LPA comparative analyses but also investigations
that will determine whether plasma LPA levels are influenced by other medical
conditions that may affect the production, secretion, and circulation of LPA.
Longitudinal studies are also required to assess how well LPA levels correlate
with disease status and thus may be used as a marker for monitoring treatment,
progression, and recurrence.
The source of the elevated plasma LPA in patients with gynecologic cancers
remains to be determined. Our data suggest that ovarian cancer cells may be
a source of LPA as we have shown that ovarian cancer cells, but not breast
cancer or leukemia cells, secrete LPA following stimulation by phorbol 12-myristate-13-acetate
(Shen et al, unpublished data, 1998).27
In summary, our findings suggest that elevated plasma LPA levels represent
a potential biomarker for gynecologic cancers and early-stage ovarian cancer
in particular. However, this study is preliminary in nature and requires validation
in larger multicenter studies. General application will require the development
of a simple assay for LPA, such as an enzyme-linked immunoassay or a radioimmunoassay.