a multi-institutional evaluation of factors predictive of toxicity and efficacy of bevacizumab for...

ORIGINAL ARTICLES

A multi-institutional evaluation of factorspredictive of toxicity and efficacy of bevacizumabfor recurrent ovarian cancerJ.D. WRIGHT*, A.A. SECORDy, T.M. NUMNUMz, R.P. ROCCONIz, M.A. POWELL§,A. BERCHUCKy, R.D. ALVAREZz, R.K. GIBB§, K. TRINKAUSk, J.S. RADER§ & D.G. MUTCH§*Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, Columbia University College ofPhysicians and Surgeons, New York, New York; yDivision of Gynecologic Oncology, Duke University MedicalCenter, Durham, North Carolina; zDivision of Gynecologic Oncology, University of Alabama at Birmingham,Birmingham, Alabama; §Division of Gynecologic Oncology, Washington University School of Medicine,St. Louis, Missouri; and kDivision of Biostatistics, Washington University School of Medicine, St. Louis, Missouri

Abstract. Wright JD, Secord AA, Numnum TM, Rocconi RP, Powell MA, Berchuck A, Alvarez RD, GibbRK, Trinkaus K, Rader JS, Mutch DG. A multi-institutional evaluation of factors predictive of toxicity and effi-cacy of bevacizumab for recurrent ovarian cancer. Int J Gynecol Cancer 2008;18:400–406.

While bevacizumab has shown activity in recurrent ovarian cancer, a higher than expected incidence of

bowel perforations has been reported in recent trials. We sought to determine factors associated with toxicityand tumor response in patients with relapsed ovarian cancer treated with bevacizumab. A retrospective

review of patients with recurrent ovarian cancer treated with bevacizumab was undertaken. Response was

determined radiographically and through CA125 measurements. Statistical analysis to determine factors asso-ciated with toxicity and response was performed. Sixty-two eligible patients were identified. The cohort had

received a median of 5 prior chemotherapy regimens. Single-agent bevacizumab was administered to 12

(19%), while 50 (81%) received the drug in combination with a cytotoxic agent. Grade 3–5 toxicities occurredin 15 (24%) patients, including grade 3–4 hypertension in 4 (7%), gastrointestinal perforations in 7%, and chy-

lous ascites in 5%. Development of chylous ascites and gastrointestinal perforations appeared to correlate

with tumor response. The overall response rate was 36% (4 complete response, 17 partial response), with sta-ble disease in 40%. A higher objective response rate was seen in the bevacizumab combination group com-

pared to single-agent treatment (43% vs 10%) (P ¼ 0.07). However, 29 grade 3–5 toxic episodes were seen in

the combination group vs only 1 in the single-agent bevacizumab cohort (P ¼ 0.071). We conclude that bev-acizumab demonstrates promising activity in recurrent ovarian cancer. The addition of a cytotoxic agent to

bevacizumab improved response rates at the cost of increased toxicity. Gastrointestinal perforations occurred

in 7%. The perforations occurred in heavily pretreated patients who were responding to therapy.

KEYWORDS: angiogenesis, bevacizumab, chylous ascites, gastrointestinal perforation, ovarian cancer.

In 2006, 20,180 new cases of ovarian cancer will bediagnosed and 15,310 women will die from the diseasein the United States(1). Standard therapy consists ofsurgical staging and cytoreduction followed by adju-vant platinum and taxane-based therapy. Despiteimprovements in survival, the majority of patientsultimately relapse(2–4). While a number of chemothera-peutic agents are active in the setting of recurrentovarian cancer, new strategies to prolong survival andimprove quality of life are needed.

Address correspondence and reprint requests to: Jason D. Wright,MD, Division of Gynecologic Oncology, Department of Obstetricsand Gynecology, Columbia University College of Physicians andSurgeons, 161 Fort Washington Avenue, 8th Floor, New York,NY 10032. Email: [email protected]

Presented at the 2006 Annual Meeting of the American Society ofClinical Oncology, Atlanta, Georgia, June, 2006.

doi:10.1111/j.1525-1438.2007.01027.x

# 2007, Copyright the AuthorsJournal compilation # 2007, IGCS and ESGO

Int J Gynecol Cancer 2008, 18, 400–406

The importance of angiogenesis in tumor develop-ment and progression has been well established(5–7).For ovarian cancer, increased vascular density andvascular endothelial growth factor (VEGF) levels havebeen correlated with aggressive clinical features andpoor outcome(8–12). In an analysis of patients withovarian cancer, the median survival was nearly three-fold higher in patients with low microvessel densitycounts compared to subjects with high microvesseldensity(8). Given these findings, therapeutic targetingof vascular growth and development has emerged asa rational target for a number of solid tumors(13,14).

The humanized, monoclonal antivascular endothe-lial growth factor antibody bevacizumab has shownpromise in the treatment of patients with relapsedovarian cancer(15–19). Despite impressive antitumoractivity, several studies of bevacizumab have noted anincreased risk of life-threatening gastrointestinal per-forations. Knowledge of clinical variables associatedwith response and toxicity would aid clinicians con-sidering bevacizumab-based therapy for women withrecurrent ovarian cancer. The goal of the present studywas to determine factors associated with tumor re-sponse and toxicity in patients with relapsed ovariancancer treated with bevacizumab.

Materials and methods

Patient population

A multi-institutional, retrospective analysis of factorsassociated with toxicity and response for women withrecurrent ovarian cancer treated with bevacizumabwas performed. Study approval was obtained fromthe institutional review boards of Washington Univer-sity School of Medicine, Duke University MedicalCenter, and the University of Alabama at Birmingham.Women with relapsed ovarian or primary peritonealcancer treated between November 1, 2004 and Decem-ber 30, 2005 were included in the analysis. All patientshad pathologically verified ovarian or primary perito-neal cancer. Patients were treated with the humanized,monoclonal antivascular endothelial growth factorantibody bevacizumab. Patients treated with single-agent bevacizumab or with bevacizumab in combina-tion with cytotoxic chemotherapy were included inthe analysis.

Toxicity assessment

Toxicity was assessed and documented according tothe National Cancer Institute’s Common Toxicity Cri-

teria version 3.0 guidelines. Subjects who receivedgreater than one dose of bevacizumab were includedin the toxicity analysis. Toxicity was determinedthrough review of medical records. Hematologic toxic-ity was assessed by serial complete blood counts. Pro-teinuria was determined from urinalysis and whenappropriate 24-hour urine protein collection. All gas-trointestinal perforations were verified with appropri-ate imaging modalities.

Assessment of response

Therapeutic efficacy was assessed in patients whoreceived greater than two doses of bevacizumab.Response was evaluated based on radiographic evalu-ation as well as through serial assessment of serumCA125 levels. Response Criteria in Solid Tumorswere utilized to determine antineoplastic response inwomen with measurable disease(20). As patients weretreated outside of a clinical trial, confirmatory imagingstudies were usually not available and not required todocument response(21). Additionally, responses wereclassified based on serial CA125 assessments based ona modification of the Rustin Criteria(22,23). Only patientswhose pretreatment CA125 was twice the upper limitof normal were assessed. A complete response (CR)was defined as normalization of CA125 (,35 U/mL)maintained for a minimum of 4 weeks. A partialresponse (PR) was defined as a 50% reduction inCA125 maintained for greater than 4 weeks. Pro-gressive disease was defined as at least a 25% increasein the serum CA125 level for at least 4 weeks. Patientswho did not meet any of the above criteria were con-sidered to have stable disease (SD). For each patient,a best response based upon the above schemas wasrecorded.

Statistics

Logistic regression was used to model the probabilityof response (CR or PR), clinical benefit (CR, PR, orSD), toxicity (presence/absence of grade 4–5), andbowel perforation (present/absent). Primary covari-ates of interest were patient’s age at treatment (,60 vs.60 years), extent of cytoreduction (optimal vs subop-timal), tumor stage (I/II vs III/IV), tumor histology(serous vs other), number of prior regimens (,5 vs.5), number of prior platinum-containing regimens(,2 vs .2), bevacizumab regimen (single-agent beva-cizumab vs bevacizumab combination), cumulativedose of bevacizumab (,3485 vs .3485 mg), numberof doses of bevacizumab (,6 vs .6), and pretreatmentCA125 levels (,426 vs .426).

Bevacizumab for recurrent ovarian cancer 401

# 2007 IGCS and ESGO, International Journal of Gynecological Cancer 18, 400–406

Results

Study population

Sixty-two patients who met the study inclusion criteriawere identified. The demographic characteristics ofthe study cohort are displayed in Table 1. Sixty pa-tients (97%) were initially diagnosed with ovarian can-cer, while two patients (3%) had primary peritonealtumors. Optimal surgical cytoreduction was achievedin 53% of the cohort. At presentation, the most com-mon stage was III, documented in 81%. The cohortwas heavily pretreated. At the time of initiation ofbevacizumab therapy, the patients had receiveda median of 5 prior regimens (range 1–15) includinga median of 2 prior platinum-containing regimens(range 1–5). Bevacizumab was administered as a sin-gle agent to 12 (19%) women, while 50 (81%) receivedthe drug in combination with one or more cytotoxicagents. A wide variety of combination regimens thatincluded cyclophosphamide, 5-fluorouracil, carbopla-tin, gemcitabine, topotecan, liposomal doxorubicin,docetaxel, and vinorelbine were utilized. The mostcommon combination regimen employed was bev-acizumab with oral cyclophosphamide (n ¼ 20). Themedian cumulative dose of bevacizumab received was3348 mg.

Predictors of response

The overall best response rate was 36.2%, while clini-cal benefit (CR, PR, or SD) was observed in 75.8%.This included 4 (6.9%) CRs and 17 (29.3%) PRs. Dis-ease stabilization was noted in 23 (39.6%) women,while progressive disease was seen in 14 (24.1%) pa-tients. When the analysis was limited to patients withradiographic follow-up, a response rate of 23.7% wasobserved (Table 2). An analysis of demographic, onco-logic, and treatment-related variables was performedto identify factors associated with tumor response andclinical benefit (Table 3, available online). A high pre-treatment CA125 level (P ¼ 0.02) as well as highcumulative bevacizumab dose (P ¼ 0.02) were associ-ated with tumor response (CR/PR). Similarly, seroushistology (P ¼ 0.006) and large cumulative bev-acizumab dose (P ¼ 0.003) were associated with clini-cal benefit (CR/PR/SD), while a high pretreatmentCA125 (P ¼ 0.05) was of marginal significance. Whilethere was a trend toward increased response rate inwomen with fewer prior treatment regimens (P ¼0.12), this did not reach statistical significance. Re-sponses were documented in 43.6% of women withserous histology compared to 21.1% of those withother histologic subtypes (P ¼ 0.14).

Predictors of toxicity

The grade 2–5 toxicities encountered are detailed inTable 4. The most common toxicity was myelosup-pression, including anemia in 30 (48%) and neu-tropenia in 18 (29%) patients. The most commonnonhematologic toxicities were proteinuria in 13 (21%)and hypertension in 10 (16%). While the majority ofthese side effects were mild, grade 3 and 4 hyperten-sion were each seen in two patients. Thromboembolicevents were noted in 3 (4.8%) women. Two (3.2%)hemorrhagic complications including an episode ofgrade 2 pulmonary bleeding and a grade 3 gastroin-testinal bleed were encountered.

Large volume chylous ascites developed in three(4.8%) patients and necessitated the discontinuation oftreatment in all three. The ascites developed between

Table 1. Patient demographics

Patients, n ¼ 62 (%)

Age (range) 52.5 years (28–79)Primary tumorOvary 60 (97)Primary peritoneal 2 (3)

CytoreductionOptimal 33 (53)Suboptimal 23 (37)Unknown 6 (10)

StageI 3 (5)II 3 (5)III 50 (81)IV 5 (8)Unknown 1 (2)

Bevacizumab therapySingle agent 12 (19)Cytotoxic combination 50 (81)

Prior therapyPrior regimens (range) 5 (1–15)Prior platinum-containingregimens (range)

2 (1–5)

Bevacizumab therapyDoses received (range) 6 doses (1–29)Cumulative dose (range) 3485 mg (300–15,560)

Table 2. Response (%)

CR PRDiseasestabilization Progression

Best response (n ¼ 58) 4 (6.9) 17 (29.3) 23 (39.6) 14 (24.1)Radiographic response

(n ¼ 38)2 (5.3) 7 (18.4) 10 (26.3) 19 (50.0)

Biochemical response(n ¼ 47)

1 (2.1) 15 (31.9) 21 (44.7) 10 (21.2)

402 J.D. Wright et al.


76 and 97 days after the start of bevacizumab. Allthree women were experiencing tumor responses asindicated by rapidly declining CA125 levels at thetime the ascites developed. Figure 1 displays a plot ofserum CA125 levels vs time in the women who devel-oped chylous ascites. Gastrointestinal perforationsoccurred in four (6.5%) women. The perforationsoccurred throughout the gastrointestinal tract includ-ing the small bowel in one patient, the descendingcolon in one, and the stomach at a gastrostomy tubesite in a third patient. The fourth patient developeda rectovaginal fistula as well as a perforation betweenthe stomach and an intra-abdominal tumor mass. Aclinically apparent bowel obstruction was present inonly the first patient. The perforations occurred from18 to 150 days after the initiation of bevacizumab.All of the perforations were noted in women whose

tumors were responding to bevacizumab as docu-mented by declining CA125 levels (Fig. 1). Only thepatient with the rectovaginal fistula had received priorradiotherapy.

An analysis was performed to identify factors asso-ciated with toxicity in the cohort (Table 5). We wereunable to document an association between any clinicalvariable and grade 3–5 toxicity. A similar analysis wasperformed for bowel perforations. Given the smallnumber of bowel perforations, we were unable to iden-tify a statistically significant association between anypatient or treatment-related factors and bowel perfora-tion. However, all four patients with bowel perfo-rations were very heavily pretreated and had largeintra-abdominal tumor burdens. The median numberof prior regimens in those who experienced a perfora-tion was 8.5 compared to 5 for the remainder of thecohort.

Combination therapy

To determine the effect of adding a cytotoxic agent tobevacizumab-based therapy, we compared the toxicityand efficacy of women who received single-agent bev-acizumab (n ¼ 12) to those who received bevacizu-mab in combination with a cytotoxic agent (n ¼ 50)(Table 6). There was a trend toward increased toxicityin the combination cohort. Grade 1–2 toxicities wererecorded in 58% of the single-agent group comparedto 72% of those who received combination therapy(P ¼ 0.49). Twenty-nine grade 3–5 adverse eventswere seen in the combination group vs only one in thesingle-agent bevacizumab cohort (P ¼ 0.071). Overall,34% of the women who received combination treat-ment experienced at least one grade 3–5 toxicity vs 8%of those treated with bevacizumab alone (P ¼ 0.15).Clinical benefit (CR/PR/SD) was seen in 80% of pa-tients treated with single-agent bevacizumab com-pared to 74% treated with combination therapy (P ¼1.00). However, there was a trend toward a higherobjective response rate (CR/PR) in the bevacizumabcombination group (43% vs 10%) (P ¼ 0.07).

Discussion

An accumulating body of data suggests that bevacizu-mab is associated with substantial activity in thesetting of recurrent ovarian cancer. However, themonoclonal antibody has also been correlated withpotentially significant toxicity in this setting. Responserates to single agent and combination bevacizumab-based therapy have ranged from 8–28%, with diseasestabilization in a further 55–67% of patients(15–18).

Table 3. Univariate analysis of factors associated withresponse (CR or PR) and clinical benefit (CR, PR, or SD) (Pub-lished online only)

Response(CR/PR)P value

Clinical benefit(CR, PR, SD)P value

Age 0.95 0.56Cytoreduction 0.53 0.65Stage 0.75 0.14Histology 0.16 0.006Number prior regimens 0.12 0.41Number prior platinum

regimens0.17 0.55

Doses of bevacizumab 0.078 0.12Cumulative bevacizumab

dose0.019 0.003

Pretreatment CA125 0.019 0.054Grade 4–5 toxicity 0.26 0.82Bowel perforation 0.49 0.97

Table 4. Grade 2–5 toxicity

Patients, n (%)

Toxicity Grade 2 Grade 3 Grade 4 Grade 5

Neutropenia 2 (3) 4 (6) 2 (3) —Thrombocytopenia — 1 (2) — —Anemia 14 (23) 2 (3) — —Nausea 1 (2) — — —Diarrhea 1 (2) 1 (2) — —Hypertension — 2 (3) 2 (3) —Proteinuria 5 (8) — — —Thromboembolic 1 (2) 2 (3) — —Hemorrhagic 1 (2) 1 (2) — —Ascites — 3 (5) — —Bowel obstruction — 4 (7) 1 (2) 1 (2)Gastrointestinal

perforation— — 3 (5) 1 (2)



We sought to determine parameters associated withclinical benefit as well as toxicity for women withrecurrent ovarian cancer treated with bevacizumab.While a number of bevacizumab-based regimens

and dosing schedules have been investigated, the clin-ical utility of combining a cytotoxic agent with bevaci-zumab for ovarian cancer remains uncertain. Initialstudies of single-agent bevacizumab in other solidtumors were associated with only modest responserates(24–26). Later trials of bevacizumab in combinationwith cytotoxic therapy were associated with moreimpressive response rates(13,27,28). The GynecologicOncology Group reported a response rate of 18% withdisease stabilization in 55% of patients with relapsedovarian cancer treated with bevacizumab at a dose of15 mg/kg every 3 weeks(15). In contrast, bevacizumabin combination with metronomic cyclophosphamide

yielded a 28% response rate with stable disease ina further 62% of participants(16). In our cohort, theresponse rate was higher in patients who receivedcombination therapy (10% vs 43%). However, clinicalutility (response or disease stabilization) was similarwhether or not bevacizumab was combined witha chemotherapeutic drug (80% vs 74%). As expected,multiagent therapy increased toxicity. Subjects whoreceived combination therapy were four times morelikely to experience grade 3–5 toxicity. This was man-ifested by 29 grade 3–5 toxic events in patients whoreceived combination therapy compared to only 1event in those who received bevacizumab alone.Thus, while combination therapy appears to improveresponse rates, multiagent therapy is also associatedwith added toxicity. The optimal combination regimenwill require further prospective study.

Figure 1. Development of chylous ascites andgastrointestinal perforations. Graph of CA125levels vs time since initiation of bevacizumab.‘‘X’’ denotes time of bowel perforation. They-axis indicates change in CA125 level, 1 0.4indicates a 40% increase in CA125, while 2 1.0indicates a 100% decrease in CA125. Chylousascites and bowel perforations correspond totumor response as evidenced by decliningCA125 levels.



To identify patients likely to experience toxicityfrom bevacizumab-based therapy, we examined theassociation between a number of clinical characteris-tics and toxicity. As in previous trials, manageable pro-teinuria (21%) and hypertension (16%) were frequentside effects. We were unable to identify any variablesthat were statistically associated with grade 3–5 toxic-ity. There was a strong trend toward added toxicitywhen bevacizumab was combined with a cytotoxicagent. We again noted the unusual toxicity of chylousascites. Chylous ascites occurred in three patients (5%)and developed between 76 and 97 days of initiatingtherapy. The mechanism of chylous is uncertain,however, given the association between VEGF-C andlymphangiogenesis it is possible that disruption ofVEGF-C by bevacizumab leads to lymphatic disrup-tion(29,30). All the three subjects who developed chy-lous ascites had tumors that were responding tobevacizumab-based therapy (Fig. 1). Data from a num-ber of studies of cetuximab and erlotinib have demon-strated an association between cutaneous adversereactions and tumor response and survival(31). Ina similar fashion, chylous ascites was a harbinger ofresponse in our cohort. Unfortunately, the ascitesproved debilitating to all three women who experi-enced the toxicity and necessitated discontinuation ofbevacizumab.

Gastrointestinal perforations are a worrisome com-plication of bevacizumab therapy for ovarian cancer.In a multi-institutional phase II evaluation of single-agent bevacizumab, a perforation rate of 11% (5 of 44patients) led to early study closure(17). Similarly, gas-trointestinal perforations were noted in 15% (2 of 13)

of ovarian cancer patients treated with bevacizumabin combination with erlotinib. The authors reportedthat both perforations occurred in patients who hadsmall bowel obstructions within the preceding 28 daysand both had peritoneal tumor implants greater than1 cm in diameter(18). Four perforations (7%) were iden-tified in our series. All four events occurred in subjectswho were very heavily pretreated (median of 8.5 priorregimens vs 5 for the remainder of the cohort) and hadlarge intra-abdominal tumor burdens. A striking find-ing in these patients was that the perforationsall occurred in women responding to treatment.

Table 6. Comparison of patients treated with single-agentbevacizumab vs patients treated with combination bev-acizumab/cytotoxic therapy

Single agent(n ¼ 12)

Combinationtherapy(n ¼ 50) P

Grade 1–2 toxicities 11 67 0.22Patients experiencinggrade 1–2 toxicity

7 (58.3) 36 (72.0) 0.49

Grade 3–5 toxicities 1 29 0.071Patients experiencinggrade 3–5 toxicitya

1 (8.3) 17 (34.0) 0.15

Toxicities (grade 3–5)Neutropenia — 6 (12.0)Thrombocytopenia — 1 (2.0)Anemia — 2 (2.0)Diarrhea — 1 (2.0)Hypertension — 4 (8.0)Thromboembolic — 2 (4.0)Hemorrhagic — 1 (2.0)Ascites — 3 (6.0)Bowel obstruction — 6 (12.0)Gastrointestinal

perforation1 (8.3) 3 (6.0)

Radiographic responseCR — 2 (5.9)PR — 7 (20.6)Stable disease 1 (25.0) 9 (26.5)Progression 3 (75.0) 16 (47.1)Not evaluable 8 16

Best responseCR — 4 (8.7)PR 1 (10.0) 16 (34.8)Stable disease 7 (70.0) 14 (30.4)Progression 2 (20.0) 12 (26.1)Not evaluable 2 4

Clinical benefit (CR/PR/SD)Radiographic 25.0% 52.9% 0.60Best 80.0% 73.9% 1.00

Response (CR/PR)Radiographic 0% 26.4% 0.55Best 10% 43.4% 0.07

aNumber of patients with grade 3–5 toxicities, several patientshad multiple toxicities.

Table 5. Univariate analysis of factors associated with toxic-ity (grade 3–5) and bowel perforation

Grade 3–5toxicity P value

BowelperforationP value

Age 0.89 0.96Cytoreduction 0.74 0.96Stage 0.25 0.39Histology 0.37 0.80Number prior regimens 0.26 0.95Number prior platinum

regimens0.72 0.13

Bevacizumab regimen 0.11 0.77Doses of bevacizumab 0.70 0.61Cumulative bevacizumab

dose0.58 1.00

Pretreatment CA125 0.99 1.00Grade 4–5 toxicity — 0.96Response (CR/PR) 0.26 0.49Clinical benefit (CR/PR/SD) 0.45 0.85



We hypothesize that angiogenic inhibition may resultin bowel wall weakening in segments of the gastroin-testinal tract already compromised by tumor orobstruction. Tumor response and tumor cell deathmay facilitate bowel perforation. This constellation ofobservations suggests that patients with heavy intra-peritoneal tumor burdens, particularly if associatedwith bowel obstruction or multiple prior chemothera-peutic regimens, should be treated with caution.Finally, consideration should be given to incorporatingbevacizumab earlier into the treatment of patientswith recurrent ovarian cancer.In conclusion, bevacizumab demonstrates promis-

ing activity in patients with recurrent ovarian and pri-mary peritoneal carcinoma. The addition of a cytotoxicagent to bevacizumab is associated with an improvedresponse rate at the cost of increased toxicity. Treatmentlimiting gastrointestinal perforations occurred in fourheavily pretreated patients with large intraperitonealtumor burdens. Further prospective studies of bevaci-zumab are warranted to determine the utility ofcombination therapy and the optimal timing of admin-istration for patients with relapsed ovarian cancer.

References

1 Jemal A, Siegel R, Ward E et al. Cancer statistics 2006. CA Cancer JClin 2006;56:106–30.

2 Spriggs D. Optimal sequencing in the treatment of recurrent ovar-ian cancer. Gynecol Oncol 2003;90(Pt 2):S39–44.

3 Armstrong DK. Relapsed ovarian cancer: challenges and manage-ment strategies for a chronic disease. Oncologist 2002;7(Suppl. 5):20–8.

4 Herzog TJ. Recurrent ovarian cancer: how important is it to treat todisease progression? Clin Cancer Res 2004;10:7439–49.

5 Jain RK. Normalization of tumor vasculature: an emerging conceptin antiangiogenic therapy. Science 2005;307:58–62.

6 Gasparini G, Longo R, Toi M, Ferrara N. Angiogenic inhibitors:a new therapeutic strategy in oncology. Nat Clin Pract Oncol2005;2:562–77.

7 Ferrara N, Kerbel RS. Angiogenesis as a therapeutic target. Nature2005;438:967–74.

8 Alvarez AA, Krigman HR, Whitaker RS, Dodge RK, Rodriguez GC.The prognostic significance of angiogenesis in epithelial ovariancarcinoma. Clin Cancer Res 1999;5:587–91.

9 Hollingsworth HC, Kohn EC, Steinberg SM, Rothenberg ML,Merino MJ. Tumor angiogenesis in advanced stage ovarian carci-noma. Am J Pathol 1995;147:33–41.

10 Stone PJ, Goodheart MJ, Rose SL, Smith BJ, DeYoung BR, Buller RE.The influence of microvessel density on ovarian carcinogenesis.Gynecol Oncol 2003;90:566–71.

11 Shen GH, Ghazizadeh M, Kawanami O et al. Prognostic significanceof vascular endothelial growth factor expression in human ovariancarcinoma. Br J Cancer 2000;83:196–203.

12 Paley PJ, Staskus KA, Gebhard K et al. Vascular endothelial growthfactor expression in early stage ovarian carcinoma. Cancer 1997;80:98–106.

13 Hurwitz H, Fehrenbacher L, Novotny W et al. Bevacizumab plusirinotecan, fluorouracil, and leucovorin for metastatic colorectal can-cer. N Engl J Med 2004;350:2335–42.

14 Yang JC, Haworth L, Sherry RM et al. A randomized trial of bev-acizumab, an anti-vascular endothelial growth factor antibody, formetastatic renal cancer. N Engl J Med 2003;349:427–34.

15 Burger RA, Sill M, Monk BJ, Greer B, Sorosky J. Phase II trial ofbevacizumab in persistent or recurrent epithelial ovarian cancer(EOC) or primary peritoneal cancer (PPC): a Gynecologic OncologyGroup (GOG) study. Am Soc Clin Oncol 2005;509.

16 Garcia AA, Oza AM, Hirte H et al. Interim report of a phase II clin-ical trial of bevacizumab (bev) and low dose metronomic oral cyclo-phosphamide (mCTX) in recurrent ovarian (OC) and primaryperitoneal carcinoma: a California Cancer Consortium Trial. Am SocClin Oncol 2005;5000.

17 Cannistra SA, Matulonis U, Penson R et al. Bevacizumab in patientswith advanced platinum-resistant ovarian cancer. Am Soc Clin Oncol2006.

18 Friberg G, Oza AM, Morgan RJ, Vokes EE, Gandara DR, FlemingGF. Bevacizumab plus erlotinib for patients with recurrent ovarianand fallopian tube cancer: preliminary results of a multi-centerphase II trial. Am Soc Clin Oncol 2006.

19 Cohn DE, Valmadre S, Resnick KE, Eaton LA, Copeland LJ, FowlerJM. Bevacizumab and weekly taxane chemotherapy demonstratesactivity in refractory ovarian cancer. Gynecol Oncol 2006;102:131–3.

20 Therasse P, Arbuck SG, Eisenhauer EA et al. New guidelines toevaluate the response to treatment in solid tumors. European Orga-nization for Research and Treatment of Cancer, National CancerInstitute of the United States, National Cancer Institute of Canada.J Natl Cancer Inst 2000;92:205–16.

21 Perez-Gracia JL, Munoz M, Williams G et al. Assessment of thevalue of confirming responses in clinical trials in oncology. EurJ Cancer 2005;41:1528–32.

22 Guppy AE, Nelstrop AE, Foster T, Agarwal R, Seckl MJ, Rustin GJ.A phase II study of sequential carboplatin, paclitaxel and topotecanin patients with previously untreated advanced ovarian cancer. Br JCancer 2004;90:810–4.

23 Rustin GJ. Use of CA-125 to assess response to new agents in ovar-ian cancer trials. J Clin Oncol 2003;21(Suppl. 10):187–93.

24 Cobleigh MA, Langmuir VK, Sledge GW et al. A phase I/II dose-escalation trial of bevacizumab in previously treated metastaticbreast cancer. Semin Oncol 2003;30(Suppl. 16):117–24.

25 Gordon MS, Margolin K, Talpaz M et al. Phase I safety and phar-macokinetic study of recombinant human anti-vascular endothelialgrowth factor in patients with advanced cancer. J Clin Oncol 2001;19:843–50.

26 Ryan CJ, Lin AM, Small EJ. Angiogenesis inhibition plus chemo-therapy for metastatic hormone refractory prostate cancer: historyand rationale. Urol Oncol 2006;24:250–3.

27 Johnson DH, Fehrenbacher L, Novotny WF et al. Randomized phaseII trial comparing bevacizumab plus carboplatin and paclitaxel withcarboplatin and paclitaxel alone in previously untreated locallyadvanced or metastatic non-small-cell lung cancer. J Clin Oncol2004;22:2184–91.

28 Herbst RS, Johnson DH, Mininberg E et al. Phase I/II trial evaluat-ing the anti-vascular endothelial growth factor monoclonal antibodybevacizumab in combination with the HER-1/epidermal growthfactor receptor tyrosine kinase inhibitor erlotinib for patients withrecurrent non-small-cell lung cancer. J Clin Oncol 2005;23:2544–55.

29 Tammela T, Enholm B, Alitalo K, Paavonen K. The biology of vas-cular endothelial growth factors. Cardiovasc Res 2005;65:550–63.

30 Karkkainen MJ, Haiko P, Sainio K et al. Vascular endothelial growthfactor C is required for sprouting of the first lymphatic vessels fromembryonic veins. Nat Immunol 2004;5:74–80.

31 Perez-Soler R, Saltz L. Cutaneous adverse effects with HER1/EGFR-targeted agents: is there a silver lining? J Clin Oncol 2005;23:5235–46.

Accepted for publication May 29, 2007



a multi-institutional evaluation of factors predictive of toxicity and efficacy of bevacizumab for...

Documents