Ovarian Cancer Essay, Research Paper
Ovarian Cancer
Of all gynecologic malignancies, ovarian cancer continues to have the
highest mortality and is the most difficult to diagnose. In the United States
female population, ovarian cancer ranks fifth in absolute mortality among
cancer related deaths (13,000/yr). In most reported cases, ovarian cancer,
when first diagnosed is in stages III or IV in about 60 to 70% of patients
which further complicates treatment of the disease (Barber, 3).
Early detection in ovarian cancer is hampered by the lack of appropriate
tumor markers and clinically, most patients fail to develop significant symptoms
until they reach advanced stage disease. The characteristics of ovarian cancer
have been studied in primary tumors and in established ovarian tumor cell lines
which provide a reproducible source of tumor material. Among the major
clinical problems of ovarian cancer, malignant progression, rapid emergence of
drug resistance, and associated cross-resistance remain unresolved. Ovarian
cancer has a high frequency of metastasis yet generally remains localized
within the peritoneal cavity. Tumor development has been associated with
aberrant, dysfunctional expression and/or mutation of various genes. This can
include oncogene overexpression, amplification or mutation, aberrant tumor
suppressor expression or mutation. Also, subversion of host antitumor immune
responses may play a role in the pathogenesis of cancer (Sharp, 77).
Ovarian clear cell adenocarcinoma was first described by Peham in 1899
as “hypernephroma of the ovary” because of its resemblance to renal cell
carcinoma. By 1939, Schiller noted a histologic similarity to mesonephric
tubules and classified these tumors as “mesonephromas.” In 1944, Saphir and
Lackner described two cases of “hypernephroid carcinoma of the ovary” and
proposed “clear cell” adenocarcinoma as an alternative term. Clear cell tumors
of the ovary are now generally considered to be of mullerian and in the genital
tract of mullerian origin. A number of examples of clear cell adenocarcinoma
have been reported to arise from the epithelium of an endometriotic cyst
(Yoonessi, 289). Occasionally, a renal cell carcinoma metastasizes to the
ovary and may be confused with a primary clear cell adenocarcinoma.
Ovarian clear cell adenocarcinoma (OCCA) has been recognized as a
distinct histologic entity in the World Health Organization (WHO) classification
of ovarian tumors since 1973 and is the most lethal ovarian neoplasm with an
overall five year survival of only 34% (Kennedy, 342). Clear cell
adenocarcinoma, like most ovarian cancers, originates from the ovarian
epithelium which is a single layer of cells found on the surface of the ovary.
Patients with ovarian clear cell adenocarcinoma are typically above the age of
30 with a median of 54 which is similar to that of ovarian epithelial cancer in
general. OCCA represents approximately 6% of ovarian cancers and bilateral
ovarian involvement occurs in less that 50% of patients even in advanced cases.
The association of OCCA and endometriosis is well documented (De La
Cuesta, 243). This was confirmed by Kennedy et al who encountered histologic or
intraoperative evidence of endometriosis in 45% of their study patients.
Transformation from endometriosis to clear cell adenocarcinoma has been
previously demonstrated in sporadic cases but was not observed by Kennedy et al.
Hypercalcemia occurs in a significant percentage of patients with OCCA.
Patients with advanced disease are more typically affected than patients with
nonmetastatic disease. Patients with OCCA are also more likely to have Stage I
disease than are patients with ovarian epithelial cancer in general (Kennedy,
348).
Histologic grade has been useful as an initial prognostic determinant in
some studies of epithelial cancers of the ovary. The grading of ovarian clear
cell adenocarcinoma has been problematic and is complicated by the multiplicity
of histologic patterns found in the same tumor. Similar problems have been
found in attempted grading of clear cell adenocarcinoma of the endometrium
(Disaia, 176). Despite these problems, tumor grading has been attempted but has
failed to demonstrate prognostic significance. However, collected data suggest
that low mitotic activity and a predominance of clear cells may be favorable
histologic features (Piver, 136).
Risk factors for OCCA and ovarian cancer in general are much less clear
than for other genital tumors with general agreement on two risk factors:
nulliparity and family history. There is a higher frequency of carcinoma in
unmarried women and in married women with low parity. Gonadal dysgenesis in
children is associated with a higher risk of developing ovarian cancer while
oral contraceptives are associated with a decreased risk. Genetic and candidate
host genes may be altered in susceptible families. Among those currently under
investigation is BRCA1 which has been associated with an increased
susceptibility to breast cancer. Approximately 30% of ovarian adenocarcinomas
express high levels of HER-2/neu oncogene which correlates with a poor prognosis
(Altcheck, 375-376). Mutations in host tumor suppresser gene p53 are found in
50% of ovarian carcinomas. There also appears to be a racial predilection, as
the vast majority of cases are seen in Caucasians (Yoonessi, 295).
Considerable variation exists in the gross appearance of ovarian clear cell
adenocarcinomas and they are generally indistinguishable from other epithelial
ovarian carcinomas. They could be cystic, solid, soft, or rubbery, and may also
contain hemorrhagic and mucinous areas (O’Donnell, 250). Microscopically, clear
cell carcinomas are characterized by the presence of variable proportions of
clear and hobnail cells. The former contain abundant clear cytoplasm with often
centrally located nuclei, while the latter show clear or pink cytoplasm and
bizarre basal nuclei with atypical cytoplasmic intraluminal projections. The
cellular arrangement may be tubulo acinar, papillary, or solid, with the great
majority displaying a mixture of these patterns. The hobnail and clear cells
predominate with tubular and solid forms, respectively (Barber, 214).
Clear cell adenocarcinoma tissue fixed with alcohol shows a high
cytoplasmic glycogen content which can be shown by means of special staining
techniques. Abundant extracellular and rare intracellular neutral mucin mixed
with sulfate and carboxyl group is usually present. The clear cells are
recognized histochemically and ultrastructurally (short and blunt microvilli,
intercellular tight junctions and desmosomes, free ribosomes, and lamellar
endoplasmic reticulum). The ultrastructure of hobnail and clear cells resemble
those of the similar cells seen in clear cell carcinomas of the remainder of the
female genital tract (O’Brien, 254). A variation in patterns of histology is
seen among these tumors and frequently within the same one.
Whether both tubular components with hobnail cells and the solid part
with clear cells are required to establish a diagnosis or the presence of just
one of the patterns is sufficient has not been clearly established. Fortunately,
most tumors exhibit a mixture of these components. Benign and borderline
counterparts of clear cell ovarian adenocarcinomas are theoretical possibilities.
Yoonessi et al reported that nodal metastases could be found even when the
disease appears to be grossly limited to the pelvis (Yoonessi, 296).
Examination of retroperitoneal nodes is essential to allow for more factual
staging and carefully planned adjuvant therapy.
Surgery remains the backbone of treatment and generally consists of
removal of the uterus, tubes and ovaries, possible partial omentectomy, and
nodal biopsies. The effectiveness and value of adjuvant radiotherapy and
chemotherapy has not been clearly demonstrated. Therefore, in patients with
unilateral encapsulated lesions and histologically proven uninvolvement of the
contralateral ovary, omentum, and biopsied nodes, a case can be made for (a)no
adjuvant therapy after complete surgical removal and (b) removal of only the
diseased ovary in an occasional patient who may be young and desirous of
preserving her reproductive capacity (Altchek, 97). In the more adv-anced
stages, removal of the uterus, ovaries, omentum, and as much tumor as possible
followed by pelvic radiotherapy (if residual disease is limited to the pelvis)
or chemotherapy must be considered. The chemotherapeutic regimens generally
involve adriamycin, alkylating agents, and cisPlatinum containing combinations
(Barber, 442).
OCCA is of epithelial origin and often contains mixtures of other
epithelial tumors such as serous, mucinous, and endometrioid. Clear cell
adenocarcinoma is characterized by large epithelial cells with abundant
cytoplasm. Because these tumors sometimes occur in association with
endometriosis or endometrioid carcinoma of the ovary and resemble clear cell
carcinoma of the endometrium, they are now thought to be of mullerian duct
origin and variants of endometrioid adenocarcinoma. Clear cell tumors of the
ovary can be predominantly solid or cystic. In the solid neoplasm, the clear
cells are arranged in sheets or tubules. In the cystic form, the neoplastic
cells line the spaces. Five-year survival is approximately 50% when these
tumors are confined to the ovaries, but these tumors tend to be aggressive and
spread beyond the ovary which tends to make 5-year survival highly unlikely
(Altchek, 416).
Some debate continues as to whether clear cell or mesonephroid
carcinoma is a separate clinicopathological entity with its own distinctive
biologic behavior and natural history or a histologic variant of endometrioid
carcinoma. In an effort to characterize clear cell adenocarcinoma, Jenison et
al compared these tumors to the most common of the epithelial malignancies, the
serous adenocarcinoma (SA). Histologically determined endometriosis was
strikingly more common among patients with OCCA than with SA. Other
observations by Jenison et al suggest that the biologic behavior of clear cell
adenocarcinoma differs from that of SA. They found Stage I tumors in 50% of the
observed patient population as well as a lower incidence of bilaterality in OCCA
(Jenison, 67-69). Additionally, it appears that OCCA is characteristically
larger than SA, possibly explaining the greater frequency of symptoms and signs
at presentation.
Risk Factors
There is controversy regarding talc use causing ovarian cancer. Until
recently, most talc powders were contaminated with asbestos. Conceptually,
talcum powder on the perineum could reach the ovaries by absorption through the
cervix or vagina. Since talcum powders are no longer contaminated with asbestos,
the risk is probably no longer important (Barber, 200). The high fat content of
whole milk, butter, and meat products has been implicated with an increased risk
for ovarian cancer in general.
The Centers for Disease Control compared 546 women with ovarian cancer
to 4,228 controls and reported that for women 20 to 54 years of age, the use of
oral contraceptives reduced the risk of ovarian cancer by 40% and the risk of
ovarian cancer decreased as the duration of oral contraceptive use increased.
Even the use of oral contraceptives for three months decreased the risk. The
protective effect of oral contraceptives is to reduce the relative risk to 0.6
or to decrease the incidence of disease by 40%. There is a decreased risk as
high as 40% for women who have had four or more children as compared to
nulliparous women. There is an increase in the incidence of ovarian cancer
among nulliparous women and a decrease with increasing parity. The “incessant
ovulation theory” proposes that continuous ovulation causes repeated trauma to
the ovary leading to the development of ovarian cancer. Incidentally, having
two or more abortions compared to never having had an abortion decreases one’s
risk of developing ovarian cancer by 30% (Coppleson, 25-28).
Etiology
It is commonly accepted that cancer results from a series of genetic
alterations that disrupt normal cellular growth and differentiation. It has
been proposed that genetic changes causing cancer occur in two categories of
normal cellular genes, proto-oncogenes and tumor suppressor genes. Genetic
changes in proto-oncogenes facilitate the transformation of a normal cell to a
malignant cell by production of an altered or overexpressed gene product. Such
genetic changes include mutation, translocation, or amplification of proto-
oncogenes Tumor suppressor genes are proposed to prevent cancer. Inactivation
or loss of these genes contributes to development of cancer by the lack of a
functional gene product. This may require mutations in both alleles of a tumor
suppressor gene. These genes function as regulatory inhibitors of cell
proliferation, such as a DNA transcription factor, or a cell adhesion molecule.
Loss of these functions could result in abnormal cell division or gene
expression, or increased ability of cells in tissues to detach. Cancer such as
OCCA most likely results from the dynamic interaction of several genetically
altered proto-oncogenes and tumor suppressor genes (Piver, 64-67).
Until recently, there was little evidence that the origin of ovarian was
genetic. Before 1970, familial ovarian cancer had been reported in only five
families. A familial cancer registry was established at Roswell Park Cancer
Institute in 1981 to document the number of cases occurring in the United States
and to study the mode of inheritance. If a genetic autosomal dominant
transmission of the disease can be established, counseling for prophylactic
oophorectomy at an appropriate age may lead to a decrease in the death rate from
ovarian cancer in such families.
The registry at Roswell Park reported 201 cases of ovarian cancer in 94
families in 1984. From 1981 through 1991, 820 families and 2946 cases had been
observed. Familial ovarian cancer is not a rare occurrence and may account for
2 to 5% of all cases of ovarian cancer. Three conditions that are associated
with familial ovarian cancer are (1) site specific, the most common form, which
is restricted to ovarian cancer, and (2) breast/ovarian cancer with clustering
of ovarian and breast cases in extended pedigrees (Altchek, 229-230). One
characteristic of inherited ovarian cancer is that it occurs at a significantly
younger age than the non-inherited form.
Cyto
have revealed frequent alterations of chromosomes 1,3,6, and 11. Many proto-
oncogenes have been mapped to these chromosomes, and deletions of segments of
chromosomes (particularly 3p and 6q) in some tumors is consistent with a role
for loss of tumor suppressor genes. Recently, a genetic linkage study of
familial breast/ovary cancer suggested linkage of disease susceptibility with
the RH blood group locus on chromosome 1p.
Allele loss involving chromosomes 3p and 6q as well as chromosomes 11p,
13q, and 17 have been frequently observed in ovarian cancers. Besides allele
loss, point mutations have been identified in the tumor suppressor gene p53
located on chromosome17p13. Deletions of chromosome 17q have been reported in
sporadic ovarian tumors suggesting a general involvement of this region in
ovarian tumor biology. Allelic loss of MYB and ESR genes map on chromosome 6q
near the provisional locus for FUCA2, the locus for a-L-fucosidase in serum.
Low activity of a-L-fucosidase in serum is more prevalent in ovarian cancer
patients. This suggests that deficiency of a-L-fucosidase activity in serum may
be a hereditary condition associated with increased risk for developing ovarian
cancer. This together with cytogenetic data of losses of 6q and the allelic
losses at 6q point to the potential importance of chromosome 6q in hereditary
ovarian cancer (Altchek, 208-212).
Activation of normal proto-oncogenes by either mutation, translocation,
or gene amplification to produce altered or overexpressed products is believed
to play an important role in the development of ovarian tumors. Activation of
several proto-oncogenes (particularly K-RAS, H-RAS, c-MYC, and HER-2/neu)
occurs in ovarian tumors. However, the significance remains to be determined.
It is controversial as to whether overexpression of the HER-2/neu gene in
ovarian cancer is associated with poor prognosis. In addition to studying
proto-oncogenes in tumors, it may be beneficial to investigate proto-oncogenes
in germ-line DNA from members of families with histories of ovarian cancer
(Barber, 323-324). It is questionable whether inheritance or rare alleles of
the H-RAS proto-oncogene may be linked to susceptibility to ovarian cancers.
Diagnosis and Treatment
The early diagnosis of ovarian cancer is a matter of chance and not a
triumph of scientific approach. In most cases, the finding of a pelvic mass is
the only available method of diagnosis, with the exception of functioning tumors
which may manifest endocrine even with minimal ovarian enlargement.
Symptomatology includes vague abdominal discomfort, dyspepsia, increased
flatulence, sense of bloating, particularly after ingesting food, mild digestive
disturbances, and pelvic unrest which may be present for several months before
diagnosis (Sharp, 161-163).
There are a great number of imaging techniques that are available.
Ultrasounds, particularly vaginal ultrasound, has increased the rate of pick-up
of early lesions, particularly when the color Doppler method is used.
Unfortunately, vaginal sonography and CA 125 have had an increasing number of
false positive examinations. Pelvic findings are often minimal and not helpful
in making a diagnosis. However, combined with a high index of suspicion, this
may alert the physician to the diagnosis.
These pelvic signs include:
Mass in the ovarian area
Relative immobility due to fixation of adhesions
Irregularity of the tumor
Shotty consistency with increased firmness
Tumors in the cul-de-sac described as a handful of knuckles
Relative insensitivity of the mass
Increasing size under observation
Bilaterality (70% for ovarian carcinoma versus 5% for benign cases)
(Barber, 136)
Tumor markers have been particularly useful in monitoring treatment,
however, the markers have and will probably always have a disadvantage in
identifying an early tumor. To date, only two, human gonadotropin (HCG) and
alpha fetoprotein, are known to be sensitive and specific. The problem with
tumor markers as a means of making a diagnosis is that a tumor marker is
developed from a certain volume of tumor. By that time it is no longer an early
but rather a biologically late tumor (Altchek, 292).
Many reports have described murine monoclonal antibodies (MAbs) as
potential tools for diagnosing malignant ovarian tumors. Yamada et al attempted
to develop a MAb that can differentiate cells with early malignant change from
adjacent benign tumor cells in cases of borderline malignancy. They developed
MAb 12C3 by immunizing mice with a cell line derived from a human ovarian tumor.
The antibody reacted with human ovarian carcinomas rather than with germ cell
tumors. MAb 12C3 stained 67.7% of ovarian epithelial malignancies, but
exhibited an extremely low reactivity with other malignancies. MAb 12C3
detected a novel antigen whose distribution in normal tissue is restricted.
According to Yamada et al, MAb 12C3 will serve as a powerful new tool for the
histologic detection of early malignant changes in borderline epithelial
neoplasms. MAb 12C3 may also be useful as a targeting agent for cancer
chemotherapy (Yamada, 293-294).
Currently there are several serum markers that are available to help
make a diagnosis. These include CA 125, CEA, DNB/70K, LASA-P, and serum inhibin.
Recently the urinary gonadotropin peptide (UCP) and the collagen-stimulating
factor have been added. Although the tumor markers have a low specificity and
sensitivity, they are often used in screening for ovarian cancer. A new tumor
marker CA125-2 has greater specificity than CA125. In general, tumor markers
have a very limited role in screening for ovarian cancer.
The common epithelial cancer of the ovary is unique in killing the
patient while being, in the vast majority of the cases, enclosed in the
anatomical area where it initially developed: the peritoneal cavity. Even with
early localized cancer, lymph node metastases are not rare in the pelvic or
aortic areas. In most of the cases, death is due to intraperitoneal
proliferation, ascites, protein loss and cachexia. The concept of debulking or
cytoreductive surgery is currently the dominant concept in treatment.
The first goal in debulking surgery is inhibition of debulking surgery
is inhibition of the vicious cycle of malnutrition, nausea, vomiting, and
dyspepsia commonly found in patients with mid to advanced stage disease.
Cytoreductive surgery enhances the efficiency of chemotherapy as the survival
curve of the patients whose largest residual mass size was, after surgery, below
the 1.5 cm limit is the same as the curve of the patients whose largest
metastatic lesions were below the 1.5 cm limit at the outset (Altchek, 422-424).
The aggressiveness of the debulking surgery is a key question surgeons
must face when treating ovarian cancers. The debulking of very large metastatic
masses makes no sense from the oncologic perspective. As for extrapelvic masses
the debulking, even if more acceptable, remains full of danger and exposes the
patient to a heavy handicap. For these reasons the extra-genital resections
have to be limited to lymphadenectomy, omentectomy, pelvic abdominal peritoneal
resections and rectosigmoid junction resection. That means that stages IIB and
IIC and stages IIIA and IIB are the only true indications for extrapelvic
cytoreductive surgery. Colectomy, ileectomy, splenectomy, segmental hepatectomy
are only exceptionally indicated if they allow one to perform a real optimal
resection. The standard cytoreductive surgery is the total hysterectomy with
bilateral salpingoophorectomy. This surgery may be done with aortic and pelvic
lymph node sampling, omentectomy, and, if necessary, resection of the
rectosigmoidal junction (Barber. 182-183).
The concept of administering drugs directly into the peritoneal cavity
as therapy of ovarian cancer was attempted more than three decades ago. However,
it has only been within the last ten years that a firm basis for this method of
drug delivery has become established. The essential goal is to expose the tumor
to higher concentrations of drug for longer periods of time than is possible
with systemic drug delivery. Several agents have been examined for their
efficacy, safety and pharmacokinetic advantage when administered via the
peritoneal route.
Cisplatin has undergone the most extensive evaluation for regional
delivery. Cisplatin reaches the systemic compartment in significant
concentrations when it is administered intraperitoneally. The dose limiting
toxicity of intraperitoneally administered cisplatin is nephrotoxicity,
neurotoxicity and emesis. The depth of penetration of cisplatin into the
peritoneal lining and tumor following regional delivery is only 1 to 2 mm from
the surface which limits its efficacy. Thus, the only patients with ovarian
cancer who would likely benefit would be those with very small residual tumor
volumes. Overall, approximately 30 to 40% of patients with small volume
residual ovarian cancer have been shown to demonstrate an objective clinical
response to cisplatin-based locally administered therapy with 20 to 30% of
patients achieving a surgically documented complete response. As a general rule,
patients whose tumors have demonstrated an inherent resistance to cisplatin
following systemic therapy are not considered for treatment with platinum-based
intraperitoneal therapy (Altchek, 444-446).
In patients with small volume residual disease at the time of second
look laparotomy, who have demonstrated inherent resistance to platinum-based
regimens, alternative intraperitoneal treatment programs can be considered.
Other agents include mitoxantrone, and recombinant alpha-interpheron.
Intraperitoneal mitoxanthone has been shown to have definite activity in small
volume residual platinum-refractory ovarian cancer. Unfortunately, the dose
limiting toxicity of the agent is abdominal pain and adhesion formation,
possibly leading to bowel obstruction. Recent data suggests the local toxicity
of mitoxanthone can be decreased considerably by delivering the agent in
microdoses.
Ovarian tumors may have either intrinsic or acquired drug resistance.
Many mechanisms of drug resistance have been described. Expression of the MDR1
gene that encodes the drug efflux protein known as p-glycoprotein, has been
shown to confer the characteristic multi-drug resistance to clones of some
cancers. The most widely considered definition of platinum response is response
to first-line platinum treatment and disease free interval. Primary platinum
resistance may be defined as any progression on treatment. Secondary platinum
resistance is the absence of progression on primary platinum-based therapy but
progression at the time of platinum retreatment for relapse (Sharp, 205-207).
Second-line chemotherapy for recurrent ovarian cancer is dependent on
preferences of both the patient and physician. Retreatment with platinum
therapy appears to offer significant opportunity for clinical response and
palliation but relatively little hope for long-term cure. Paclitaxel (trade
name: Taxol), a prototype of the taxanes, is cytotoxic to ovarian cancer.
Approximately 20% of platinum failures respond to standard doses of paclitaxel.
Studies are in progress of dose intensification and intraperitoneal
administration (Barber, 227-228). This class of drugs is now thought to
represent an active addition to the platinum analogs, either as primary therapy,
in combination with platinum, or as salvage therapy after failure of platinum.
In advanced stages, there is suggestive evidence of partial
responsiveness of OCCA to radiation as well as cchemotherapy, adriamycin,
cytoxan, and cisPlatinum-containing combinations (Yoonessi, 295). Radiation
techniques include intraperitoneal radioactive gold or chromium phosphate and
external beam therapy to the abdomen and pelvis. The role of radiation therapy
in treatment of ovarian canver has diminished in prominence as the spread
pattern of ovarian cancer and the normal tissue bed involved in the treatment of
this neoplasm make effective radiation therapy difficult. When the residual
disease after laparotomy is bulky, radiation therapy is particularly ineffective.
If postoperative radiation is prescribed for a patient, it is important that
theentire abdomen and pelvis are optimally treated to elicit a response from the
tumor (Sharp, 278-280).
In the last few decades, the aggressive attempt to optimize the
treatment of ovarian clear cell adenocarcinoma and ovarian cancer in general has
seen remarkable improvements in the response rates of patients with advanced
stage cancer without dramatically improving long-term survival. The promises of
new drugs with activity when platinum agents fail is encouraging and fosters
hope that, in the decades to come, the endeavors of surgical and pharmacoogical
research will make ovarian cancer an easily treatable disease.
Bibliography
Altchek, A., & Deligdisch, L. (1996). Diagnosis and Management of Ovarian
Disorders. New York: Igaku Shoin.
Barber, H. (1993). Ovarian Carcinoma: Etiology, Diagnosis, and Treatment. New
York: Springer Verlag.
Coppleson, M. (Ed.). (1981). Gynecologic Oncology (vol. 2). New York:
Churchill Livingstone.
Current Clinical Trials Oncology. (1996). Green Brook, NJ: Pyros Education.
De La Cuesta, R., & Eichorn, J. (1996). Histologic transformation of benign
endometriosis to early epithelial ovarian cancer. Gynecologic Oncology,
60, 238- 244.
Disaia, P, & Creasman, W. (1989). Clinical Gynecologic Oncology (3rd ed.). St.
Louis: Mosby.
Jenison, E., Montag, A., & Griffiths, T. (1989). clear cell adenocarcinoma of
the ovary: a clinical analysis and comparison with serous carcinoma. Gynecologic
Oncology, 32, 65-71.
Kennedy, A., & Biscotti, C. (1993). Histologic correlates of progression-free
interval and survival in ovarian clear cell adenocarcinoma. Gynecologic Oncology, 50,
334-338.
Kennedy, A., & Biscotti, C. (1989). Ovarian clear cell adenocarcinoma.
Gynecologic Oncology, 32, 342-349.
O’Brien, M., Schofield, J., & Tan, S. (1993). Clear cell epithelial ovarian
cancer: Bad prognosis only in early stages. Gynecologic Oncology, 49, 250-254.
O’Donnell, M, & Al-Nafussi, A. (1995). Intracytoplasmic lumina and mucinous
inclusions in ovarian carcinoma. Histopathology, 26, 181-184.