C33Awere stably transfected with a SPARC expression vector.Increased SPARC expression was confirmed by Western blotanalysis. The effect of forced SPARC expression on cellproliferation was determined using timed cell counts and colonyformation assays.

Results. Normal cervical tissue specimens expressed highlevels of SPARC. SPARC expression was absent in stage Icancer specimens but was noted to trend toward normal levels instage II and III samples. SPARC expression in the cervicalcancer cell lines was low when compared to normal cervixspecimens. Forced SPARC expression significantly inhibitedgrowth of C33A. This effect was not seen in CaSki or HeLacells.

Conclusions. SPARC expression is interrupted as the cervixadvances from normal tissue to cancer. Induced expression ofSPARC inhibits growth of C33A cells in vitro. The function ofendogenous SPARC in cervical cancer remains unclear.However, the differential effect of SPARC expression inC33A as compared to HeLa and CaSki cells might provide aunique opportunity to elucidate the mechanism by whichSPARC functions in the progression of cervical cancer.

doi:10.1016/j.ygyno.2007.08.014

5Paclitaxel Maintenance Chemotherapy FollowingIntraperitoneal Chemotherapy for Ovarian CancerNicole Davis, April Rogers, William RobinsonTexas Tech/Harrington Cancer Center, Amarillo, TX

Objectives. To determine the feasibility and morbidity of twotreatment regimens for optimally debulked FIGO stage IIIovarian, fallopian tube, or primary peritoneal cancers. Thetreatment regimens consist of: (1) combined intraperitoneal/intravenous(IP/IV) cisplatin/paclitaxel and 12 cycles of main-tenance paclitaxel IV; (2) IV only carboplatin/paclitaxel and 12cycles of maintenance paciltaxel IV.

Methods. One hundred two subjects identified by casereview were treated between January 2003 and December 2006for FIGO stage III ovarian, fallopian tube, or primary peritonealcancer. All subjects underwent optimal debulking surgery,received either IP/IV or IV primary chemotherapy for 6 cycles,and had a complete clinical response. The subjects were thenoffered and agreed to receive maintenance paclitaxel IV for anadditional 12 months. Demographic and clinical data werecollected from all subjects in an effort to identify specifictoxicities and estimate tolerability of the regimens.

Results. Forty-five subjects received combined IP/IVchemotherapy, versus 57 who received IV therapy alone.Stratification by IP/IV vs. IV administration was not associatedwith significant differences in age, ethnicity, primary tumor site,tumor histology or incidence of bowel resection at the time ofsurgery. Cell type, primary tumor site, and incidence of bowelsurgery were similar in both groups. Toxicities noted includefatigue, neuropathy, myelosuppression, and nausea/vomiting

and were also similar in both groups. 29/47 subjects (61.7%) inthe IP/IV group completed 12 cycles of maintenance paclitaxelchemotherapy versus 18/55 (32.7%) in the IV group (p=0.006).The mean number of cycles completed by the IP/IV group was8.6, while the IV group completed a mean of 5.8 cycles(p=0.002). In those subjects who received less than 12 cycles,the mean number of cycles completed by the IP/IV group was3.1, versus 2.8 in the IV group. The reasons for stopping weresimilar in each group and included neuropathy (33), fatigue (8),myelosuppression (7), and disease progression (6).

Conclusions. Patients who received combined IP/IV che-motherapy were more likely to complete maintenance therapythan those who only received IV chemotherapy. Patients whocomplete IP/IV chemotherapy may be more willing to toleratethe toxicity and inconvenience of maintenance chemotherapythan those who receive IV. Also, patients who stop maintenancetherapy usually do so early in the course. Additional resourcesdirected at physical and emotional support of patients during thefirst 1–3 cycles of maintenance chemotherapy may allow moreto complete the 12 month regimen.

doi:10.1016/j.ygyno.2007.08.015

6Predictors for Success in a Phase III Trial—an Analysis of315 Phase II Chemotherapy Trials for Advanced CancersS.M. Ueda, D.S. Kapp, V.E. Sugiyama, C. Stave, J.Y. Shin,B.J. Monk, B.I. Sikic, K. Osann, J.K. ChanStanford University School of Medicine, Stanford, CAUniversity of California, San Francisco School of Medicine,San Francisco, CA

Objectives. To determine the characteristics of phase IIstudies that predict for subsequent “positive” phase III chemo-therapy trials.

Methods. All phase III clinical trials with their prior phase IIstudies on chemotherapy in cancer treatment published from1985 to 2006 were extracted. Predictive factors for positivephase III trials were analyzed using the Chi-square test andlogistic regression.

Results. Of the 315 phase III clinical trials with precedingphase II studies, lung, breast, gastrointestinal, gynecologic,genitourinary, and dermatological cancers comprised of 40.0%,8.4%, 16.8%, 9.8%, 6.3%, and 4.1% of the studies. 199 (63.2%)phase III trials were “positive” and 116 (36.8%) were“negative”. Over the 10 year study periods 1985–1995 and1996–2006, the percentage of phase II studies that led topositive phase III trials increased from 53.8% to 69.7%(p=0.004). The reported results of phase II trials (negative vs.positive vs. equivocal) were predictive for phase III success(31.2% vs. 65.0% vs. 60.0%; p=0.024), respectively. Theinterval between the publication of phase II and III studies (0–2,3–5, N5 years) was associated with the success of phase III trial(74.2%, 61.8%, and 54.7%, respectively; p=0.022). Thepublication of the phase II studies in journals with an impact

362 ABSTRACTS / Gynecologic Oncology 107 (2007) 360–381