Novel compositions and methods for identification, assessment, prevention, and therapy of ovarian cancer

The present application is a divisional of U.S. patent application Ser. No. 10/361,112, filed Feb. 7, 2003 (pending), which claims the benefit of U.S. Provisional Application Ser. No. 60/384,042, filed May 29, 2002 (abandoned), and US. Provisional Application Ser. No. 60/355,388, filed Feb. 8, 2002 (abandoned). The entire contents of each of the above-referenced patent applications are incorporated herein by this reference.

The increased number of cancer cases reported in the United States, and, indeed, around the world, is a major concern. Currently there are only a handful of detection and treatment methods available for specific types of cancer, and these provide no absolute guarantee of success. In order to be most effective, these treatments require not only an early detection of the malignancy, but also a reliable assessment of the severity of the malignancy.

Ovarian cancer is responsible for significant morbidity and mortality in populations around the world. Ovarian cancer is classified, on the basis of clinical and pathological features, in three groups, namely epithelial ovarian cancer (EOC; >90% of ovarian cancer in Western countries), germ cell tumors (circa 2-3% of ovarian cancer), and stromal ovarian cancer (circa 5% of ovarian cancer; Ozols et al., 1997, Cancer Principles and Practice of Oncology, 5th ed., DeVita et al., Eds. pp. 1502). Relative to EOC, germ cell tumors and stromal ovarian cancers are more easily detected and treated at an early stage, translating into higher/better survival rates for patients afflicted with these two types of ovarian cancer.

There are numerous types of ovarian tumors, some of which are benign, and others of which are malignant. Treatment (including non-treatment) options and predictions of patient outcome depend on accurate classification of the ovarian cancer. Ovarian cancers are named according to the type of cells from which the cancer is derived and whether the ovarian cancer is benign or malignant. Recognized histological tumor types include, for example, serous, mucinous, endometrioid, and clear cell tumors. In addition, ovarian cancers are classified according to recognized grade and stage scales.

In grade I, the tumor tissue is well differentiated from normal ovarian tissue. In grade II, tumor tissue is moderately well differentiated. In grade III, the tumor tissue is poorly differentiated from normal tissue, and this grade correlates with a less favorable prognosis than grades I and II. Stage I is generally confined within the capsule surrounding one (stage IA) or both (stage IB) ovaries, although in some stage I (i.e. stage IC) cancers, malignant cells may be detected in ascites, in peritoneal rinse fluid, or on the surface of the ovaries. Stage II involves extension or metastasis of the tumor from one or both ovaries to other pelvic structures. In stage IIA, the tumor extends or has metastasized to the uterus, the fallopian tubes, or both. Stage IIB involves extension of the tumor to the pelvis. Stage IIC is stage IIA or IIB in which malignant cells may be detected in ascites, in peritoneal rinse fluid, or on the surface of the ovaries. In stage III, the tumor comprises at least one malignant extension to the small bowel or the omentum, has formed extrapelvic peritoneal implants of microscopic (stage IIIA) or macroscopic (<2 centimeter diameter, stage IIIB; >2 centimeter diameter, stage IIIC) size, or has metastasized to a retroperitoneal or inguinal lymph node (an alternate indicator of stage IIIC). In stage IV, distant (i.e. non-peritoneal) metastases of the tumor can be detected.

The durations of the various stages of ovarian cancer are not presently known, but are believed to be at least about a year each (Richart et al., 1969, Am. J. Obstet. Gynecol. 105:386). Prognosis declines with increasing stage designation. For example, 5-year survival rates for patients diagnosed with stage I, II, III, and IV ovarian cancer are 80%, 57%, 25%, and 8%, respectively.

Despite being the third most prevalent gynecological cancer, ovarian cancer is the leading cause of death among those afflicted with gynecological cancers. The disproportionate mortality of ovarian cancer is attributable to a substantial absence of symptoms among those afflicted with early-stage ovarian cancer and to difficulty diagnosing ovarian cancer at an early stage. Patients afflicted with ovarian cancer most often present with non-specific complaints, such as abnormal vaginal bleeding, gastrointestinal symptoms, urinary tract symptoms, lower abdominal pain, and generalized abdominal distension. These patients rarely present with paraneoplastic symptoms or with symptoms which clearly indicate their affliction. Presently, less than about 40% of patients afflicted with ovarian cancer present with stage I or stage II. Management of ovarian cancer would be significantly enhanced if the disease could be detected at an earlier stage, when treatments are much more generally efficacious.

Ovarian cancer may be diagnosed, in part, by collecting a routine medical history from a patient and by performing physical examination, x-ray examination, and chemical and hematological studies on the patient. Hematological tests which may be indicative of ovarian cancer in a patient include analyses of serum levels of proteins designated CA125 and DF3 and plasma levels of lysophosphatidic acid (LPA). Palpation of the ovaries and ultrasound techniques (particularly including endovaginal ultrasound and color Doppler flow ultrasound techniques) can aid detection of ovarian tumors and differentiation of ovarian cancer from benign ovarian cysts. However, a definitive diagnosis of ovarian cancer typically requires performing exploratory laparotomy of the patient.

Potential tests for the detection of ovarian cancer (e.g., screening, reflex or monitoring) may be characterized by a number of factors. The “sensitivity” of an assay refers to the probability that the test will yield a positive result in an individual afflicted with ovarian cancer. The “specificity” of an assay refers to the probability that the test will yield a negative result in an individual not afflicted with ovarian cancer. The “positive predictive value” (PPV) of an assay is the ratio of true positive results (i.e. positive assay results for patients afflicted with ovarian cancer) to all positive results (i.e. positive assay results for patients afflicted with ovarian cancer+positive assay results for patients not afflicted with ovarian cancer). It has been estimated that in order for an assay to be an appropriate population-wide screening tool for ovarian cancer the assay must have a PPV of at least about 10% (Rosenthal et al., 1998, Sem. Oncol. 25:315-325). It would thus be desirable for a screening assay for detecting ovarian cancer in patients to have a high sensitivity and a high PPV. Monitoring and reflex tests would also require appropriate specifications.

Owing to the cost, limited sensitivity, and limited specificity of known methods of detecting ovarian cancer, screening is not presently performed for the general population. In addition, the need to perform laparotomy in order to diagnose ovarian cancer in patients who screen positive for indications of ovarian cancer limits the desirability of population-wide screening, such that a PPV even greater than 10% would be desirable.

Prior use of serum CA 125 level as a diagnostic marker for ovarian cancer indicated that this method exhibited insufficient specificity for use as a general screening method. Use of a refined algorithm for interpreting CA125 levels in serial retrospective samples obtained from patients improved the specificity of the method without shifting detection of ovarian cancer to an earlier stage (Skakes, 1995, Cancer 76:2004). Screening for LPA to detect gynecological cancers including ovarian cancer exhibited a sensitivity of about 96% and a specificity of about 89%. However, CA125-based screening methods and LPA-based screening methods are hampered by the presence of CA125 and LPA, respectively, in the serum of patients afflicted with conditions other than ovarian cancer. For example, serum CA 125 levels are known to be associated with menstruation, pregnancy, gastrointestinal and hepatic conditions such as colitis and cirrhosis, pericarditis, renal disease, and various non-ovarian malignancies. Serum LPA is known, for example, to be affected by the presence of non-ovarian gynecological malignancies. A screening method having a greater specificity for ovarian cancer than the current screening methods for CA125 and LPA could provide a population-wide screening for early stage ovarian cancer.

Presently greater than about 60% of ovarian cancers diagnosed in patients are stage III or stage IV cancers. Treatment at these stages is largely limited to cytoreductive surgery (when feasible) and chemotherapy, both of which aim to slow the spread and development of metastasized tumor. Substantially all late stage ovarian cancer patients currently undergo combination chemotherapy as primary treatment, usually a combination of a platinum compound and a taxane. Median survival for responding patients is about one year. Combination chemotherapy involving agents such as doxorubicin, cyclophosphamide, cisplatin, hexamethylmelamine, paclitaxel, and methotrexate may improve survival rates in these groups, relative to single-agent therapies. Various recently-developed chemotherapeutic agents and treatment regimens have also demonstrated usefulness for treatment of advanced ovarian cancer. For example, use of the topoisomerase I inhibitor topectan, use of amifostine to minimize chemotherapeutic side effects, and use of intraperitoneal chemotherapy for patients having peritoneally implanted tumors have demonstrated at least limited utility. Presently, however, the 5-year survival rate for patients afflicted with stage III ovarian cancer is 25%, and the survival rate for patients afflicted with stage IV ovarian cancer is 8%.

It would therefore be beneficial to provide specific methods and reagents for the diagnosis, staging, prognosis, monitoring, and treatment of diseases associated with ovarian cancer, or to indicate a predisposition to such for preventative measures. The present invention is directed towards these needs.

The present invention is directed to the identification of markers that can be used to determine the sensitivity or resistance of ovarian cancer to a therapeutic agent. By examining the expression of one or more of the identified markers, whose expression correlates with sensitivity to a therapeutic agent or resistance to a therapeutic agent, in a sample of ovarian cancer cells, it is possible to determine whether a therapeutic agent or combination of agents will be most likely to reduce the growth rate of the ovarian cancer cells and can further be used in selecting appropriate treatment agents. In one embodiment, the invention is further directed to the identification of markers that can be used to determine the sensitivity or resistance of ovarian tumors to a therapeutic agent. By examining the expression of one or more of the identified markers, whose expression correlates with sensitivity to a therapeutic agent or resistance to a therapeutic agent, in a sample of ovarian tumor cells, it is possible to determine whether a therapeutic agent or combination of agents will be most likely to reduce the growth rate of the ovarian tumor cells and can further be used in selecting appropriate treatment agents. The markers of the present invention, whose expression correlates with sensitivity or with resistance to an agent, are identified in Table 1 as n1-n78 and further characterized in Table 16.

The invention further comprises the use of certain combinations of markers, wherein the expression of one or more markers, correlates with sensitivity or resistance to a therapeutic agent. Preferred combinations of markers referred to herein as “marker sets” are set forth in Tables 2-15. In addition, by examining the expression of the individual markers of the newly-identified marker sets (also referred to as the “expression profile” of the marker set), it is possible to determine whether a therapeutic agent or combination of agents will be most likely to reduce the growth rate of the ovarian cancer cells and can further be used in selecting appropriate treatment agents. Moreover, by examining the expression of individual markers and marker sets, it is also possible to determine whether a patient will most likely experience early or late recurrence of ovarian cancer growth. This information can be used in selecting an appropriate treatment.

Table 1 lists all of the markers of the invention (and comprises markers listed in Tables 2-16) which are designated with a marker identification number (“No.”), the image Clone ID (“Image Clone ID”), the gene corresponding to the marker (“Gene Name”), the Accession Number (“Accession”), the GeneBank number (“GI number”), and the Reference Sequence Nucleic Accession Number (“RefSeq”).

Tables 2-15 list marker sets, comprised of multiple individual markers, which are designated with a marker identification number within each marker set (“Number”), the marker identification numbers set forth in Table 1 (“Marker Number”), the Image clone identification number (“Image Clone Id”), the feature selection (“Feature Selection”), and the classification error rate of the model (“Classification Error Rate of the Model”). Table 2 is a preferred marker set.

Table 16 identifies the 78 individual markers of the present invention (n1-n78). The marker identification numbers are set forth in Table 16 (“No”), the Image clone identification number (“Image Clone Id”), and the Signal-to-noise (“SNR score”). Table 16 lists markers using SNR statistics applied to 18,539 genes of the 51 ovarian samples. In particular, the markers in Table 16 with negative SNR values are correlated with resistance to an agent (referred to herein as “resistance markers”), and the markers with positive SNR scores are correlated with sensitivity to an agent (referred to herein as “sensitivity markers”).