Ten-year follow-up of a phase 2 study of dose-intense paclitaxel with cisplatin and cyclophosphamide as initial therapy for poor-prognosis, advanced-stage epithelial ovarian cancer




The objective of this study was to assess activity and toxicity in patients with newly diagnosed, advanced-stage epithelial ovarian cancer (EOC) who were receiving dose-intense paclitaxel, cyclophosphamide, cisplatin, and filgrastim delivered with a flexible dosing schedule.


Patients with stage III/IV EOC received cyclophosphamide 750 mg/m2, followed by a 24-hour infusion of paclitaxel 250 mg/m2 and cisplatin 75 mg/m2 on Day 2. Filgrastim began on Day 3 at 10 μg/kg daily for 9 days. Patients received 6 cycles of all drugs. Those who achieved a pathologic complete response or had microscopic residual disease at the conclusion of 6 cycles of therapy received an additional 2 to 4 cycles of paclitaxel with cyclophosphamide. Patients who had an objective response continued on cyclophosphamide and paclitaxel.


Sixty-two patients were enrolled. Thirty-two of 62 patients had stage IIIC disease, and 26 of 62 patients had stage IV disease. According to an intent-to-treat analysis, 55 patients (89%) experienced a clinical complete remission. At a median potential follow-up of 11.4 years, the median progression-free survival was 18.9 months, and the median survival was 5.4 years. The most serious toxicity was grade 3/4 neutropenic fever (35%). Although all participants developed peripheral neuropathy, improvement in neuropathic symptoms began with the decrease or cessation of paclitaxel.


The studied regimen yielded a high response rate and encouraging overall survival. The current data and those reported by the Japanese Gynecologic Oncology Group suggest that further study is warranted of dose-dense or dose-intense paclitaxel regimens in women with newly diagnosed, advanced-stage EOC. Cancer 2010. © 2010 American Cancer Society.

Epithelial ovarian cancer (EOC) is the leading cause of death from gynecologic malignancies in the United States, and 15,520 deaths are expected in 2008. 1 It continues to be associated with a poor prognosis and advanced-stage disease at presentation. 2 Most women with advanced-stage disease can anticipate optimal surgical cytoreduction and a subsequent response to platinum-based chemotherapy, yet most will experience tumor recurrence and eventual death from EOC. Sustained remissions and cures remain rare in women with advanced-stage disease.

Many approaches have been explored to optimize and improve the initial treatment. Interval surgical cytoreduction has been tested without notable advantage in the paclitaxel era. Other approaches in initial systemic therapy include triple-drug or sequential doublet regimens, 3, 4 maintenance therapy, 5 introduction of novel treatment regimens (including molecularly targeted agents for initial treatment), 6 and dose intensification. 7-11 Limited advances, albeit statistically significant, have been observed. Recent reports indicate that dose-dense taxane therapy benefits patients with breast cancer and EOC. After standard adjuvant therapy, patients with breast cancer who received weekly paclitaxel had improved disease-free survival (progression-free survival [PFS]) and overall survival (OS) compared with patients who received paclitaxel every 3 weeks. 12 More relevant are recent results from the Japanese Gynecologic Oncology Group (JGOG), who demonstrated improved PFS in women with stage II through IV disease who treated with carboplatin coupled with weekly paclitaxel at a dose of 80 mg/m2 compared with standard schedule paclitaxel (180 mg/m2) delivered every 3 weeks. 11

In this article, we report the activity, toxicity, and 10-year survival outcome of women with EOC in a phase 2 trial of a dose-intense paclitaxel regimen. Prospective goals were to define the surgical response rate, a flexible filgrastim dosing schedule to maintain paclitaxel dose density, individualization of treatment duration, and portability through multi-institutional collaboration. Secondary objectives were the characterization of OS and PFS and the kinetics of neurotoxicity.

Cisplatin and cyclophosphamide, with differing and potentially synergistic mechanisms of action, were combined with dose-intensified paclitaxel. 13-16 Paclitaxel stabilizes microtubule assembly 17 and slows the repair of DNA lesions, including cisplatin adducts and cyclophosphamide cross-links. 16, 18 We defined the maximum tolerated dose of paclitaxel as 250 mg/m2 given as a 24-hour infusion with flexible filgrastim support 19 and determined its safety when combined with cyclophosphamide 750 mg/m2 and cisplatin 75 mg/m2. 20 Administration of the triplet to patients with advanced stage EOC led to an 89% clinical response rate and a 61% surgical response rate (microscopic and pathologic) in a single-institution study of patients with EOC. 20 That phase 2 study defined the activity and tolerability of this dose-intense paclitaxel combination.


Patient Eligibility and Pretreatment Evaluation

The study was approved by the institutional review boards at the National Cancer Institute (NCI) and Massachusetts General Hospital (MGH). Eligibility criteria included women with untreated, newly diagnosed, International Federation of Gynecology and Obstetrics stage III/IV EOC; initiation of therapy within 6 weeks of staging laparotomy; creatinine clearance ≥60 mL/minute; good end-organ function; an Eastern Cooperative Oncology Group performance status between 0 and 2; and informed consent. Patients were excluded for a prior history of invasive cancer, myocardial infarction, or unstable dysrhythmia within 6 months of study entry or for gastrointestinal bleeding within 1 month of entry.

Drug Administration

Agents were provided by the Cancer Therapy Evaluation Program, NCI. NCI patients received their first cycle in the hospital and received subsequent cycles as outpatients, whereas MGH patients received all cycles in the hospital. The premedication, treatment, and dose modification schema are shown in Tables 1 and 2. Cisplatin was limited to 6 cycles to minimize the risk of allergic reactions and cumulative neurotoxicity. Cyclophosphamide and paclitaxel were continued in patients who required >6 cycles to achieve the best clinical response and after surgical reassessment. Subsequent cycles were initiated if nonhematologic toxicity was grade <3 and if their absolute granulocyte count was ≥1500 cells/mm3.

Table 1. Drug Administration and Dose Modification Schema
Administration Day21-Day Cycle
  • CI indicates continuous infusion.

  • a

    Premedication: Day 1: 20 mg of oral dexamethasone 14 hours and 7 hours before paclitaxel; cimetidine at a dose of 300 mg and diphenhydramine at a dose of 25 mg 30 minutes before paclitaxel. Day 2: dexamethasone at a dose of 20 mg and mannitol diuresis.

  • b

    Filgrastim was continued until the total granulocyte count was >30,000/dL or for 9 days, whichever was longer.

  • c

    Ondansetron was administered at a dose of 8 mg intravenously before cyclophosphamide and then orally every 4 to 8 hours for 4 to 6 doses.

  • d

    Prochlorperazine SR was administered every 12 hours on Days 1 and 2 beginning 1 hour before cyclophosphamide followed by 30 mg every 12 hours for 4 doses and 15 mg every 12 hours for 4 to 6 doses on Days 5 and 6.

Cyclophosphamide, 750 mg/m2X      
Paclitaxel, 250 mg/m224-h CI      
Cisplatin, 75 mg/m2 X     
Filgrastim, 10 μg/kg/db  Daily on D3-D13XXXX
Table 2. Dose Modification
ToxicityFirst OccurrenceSecond OccurrenceThird Occurrence
  1. F&N indicates fever and neutropenia; G, filgrastim; T, paclitaxel; C, cyclophosphamide; P, cisplatin.

Neutropenia >5 d with or without F&NG, 20/μg/kg/dT, 200 mg/m2C, 500 mg/m2
Grade 4 thrombocytopeniaT, 200 mg/m2P, 50 mg/m2T, 150 mg/m2
Grade 3 neurotoxicityT, 200 mg/m2T, 175 mg/m2T, 150 mg/m2
Grade 4 neurotoxicityT, 200 mg/m2  
 P, 50 mg/m2  

Toxicity Assessment and Dose Modification

Drug compliance and toxicity were assessed at each visit and were graded using NCI Common Toxicity Criteria version 1.0 (CTC v1.0). Proprioception, vibration, and pinprick sensory modalities; deep tendon reflexes; Romberg maneuver; and tandem gait were evaluated in each cycle. The paclitaxel dose was modified for objective clinical findings with functional impairment consistent with CTC v1.0 grade 3 neurologic toxicity.

Outcome Assessment

Patients underwent clinical assessment every cycle and radiographic assessment every 2 cycles. Patients who had normal physical examination, CA 125 levels, and radiographic studies after 6 cycles (clinical complete response [CCR]) underwent surgical reassessment. Therapy after surgical assessment was guided based on intraoperative findings. Two cycles of paclitaxel and cyclophosphamide were administered to patients who had pathologically negative findings, whereas women who had microscopic residual disease received 2 to 4 additional cycles. Patients who had unresectable gross residual tumor (pathologic partial response [PPR]) were referred to other treatments. Patients who did not attain a CCR after 6 cycles received from 2 up to 4 additional cycles of paclitaxel and cyclophosphamide if they still were responding. Continued persistent disease led to a change in therapy. All patients were followed for survival.

Disease outcomes were measured using radiographic imaging studies and/or physical examinations. Response was documented as a CCR, or a partial response (PR) that lasted ≥4 weeks, or progressive disease (PD). A PR required a reduction >50% in the sum of the products of bidirectional measurements of tumor masses. Standard PD definitions were used. Disease response was classified further at surgery. Patients who had CCR and underwent surgical assessment were categorized with a pathologic CR (pCR), microscopic residual disease (micro), or a PPR. Off-study criteria included failure to achieve a CCR, PPR, or PD as the best response; grade ≥3 nonhematologic toxicity; and voluntary withdrawal.

Statistical Analysis

The primary objectives were to identify the pCR rate and to determine regimen portability to another academic site; the study was defined to accrue up to 80 participants. Assessments of OS and PFS were secondary objectives and were calculated from the on-study date until the date of death, disease progression, or last follow-up. OS and PFS probabilities were computed using the Kaplan-Meier method. Survival was censored at the time of the most recent follow-up.


Patient Characteristics

This cohort had multiple indicators of a poor prognosis (Table 3). All patients underwent attempts at optimal debulking during surgical staging, and 42% had pathologically documented stage IV disease. Residual disease was present at enrollment in 35 patients (44%), most of whom had hepatic parenchymal involvement; 93% had radiologic or clinically measurable disease, and 7% had residual effusions only. Forty-three percent of patients underwent optimal debulking (<1 cm; n = 27).

Table 3. Clinical Characteristics
CharacteristicNo. of Patients (%)
  • NCI indicates National Cancer Institute; MGH, Massachusetts General Hospital; LN, lymph nodes.

  • a

    Multiple sites within 1 organ were counted once.

Total patients accrued46 (75)16 (25)62 (100)
No. evaluable for response451661
No. evaluable for toxicity461662
Median age [range], y50 [43-69]55 [40-69]52 [40-69]
Stage on study   
 IIIA011 (1)
 IIIB213 (5)
 IIIC23932 (52)
 IVa21526 (42)
 Papillary serous241337 (59)
 Endometrioid/transitional10010 (16)
 Undifferentiated404 (7)
 Clear cell213 (5)
 Primary peritoneal628 (13)
Tumor grade   
 1011 (1)
 29211 (18)
 3371249 (79)
Sites of disease   
 Liver13619 (31)
 Spleen314 (6)
 Lung/chest parenchymal235 (8)
 Mesenteric/omental22628 (45)
 Retroperitoneal LN17219 (31)
 Pelvis27835 (56)
 Pleural effusion18422 (35)
 Ascites21324 (39)

Response and Duration

The pCR rate, which was the primary study objective, is presented in Tables 4 and 5. CCRs occurred in 55 patients (89%), and 2 patients had a PR (overall response rate, 92%) (Table 4). All but 1 patient underwent an invasive reassessment. pCRs occurred in 26 patients (47%), and microscopic residual was identified in 20 patients (36%), yielding an 84% overall pathologic response rate (Table 5). A patient who had microscopic disease at reassessment had a pCR identified at surgery for an adhesion-induced small bowel obstruction. The median potential follow-up for the cohort was 11.4 years, the median PFS was 18.9 months (Fig. 1), and the median OS was 5.4 years (5-year OS rate, 53%). When the results were restricted to patients with stage IV disease who had parenchymal liver or splenic involvement, the median PFS was 15.8 months, and the median OS was 2.9 years, (5-year OS rate, 42%).

Table 4. Clinical Response: Intent-to-Treat (N=62)
Type of ResponseNo. of Patients (%)
  • NCI indicates National Cancer Institute; MGH, Massachusetts General Hospital; CCR, clinical complete remission; PR, partial remission; SD, stable disease; NE, not evaluable, RR, response rate.

  • a

    Patient withdrew after 2 cycles with unconfirmed PR and tinnitus.

CCR401555 (89)
PR202 (3)
SD303 (5)
Overall clinical RR42 (91)15 (94)57 (92)
Table 5. Pathologic Response After Surgical Assessment of Patients in Clinical Complete Remission (N=55)
Type of ResponseNo. of Patients (%)
  • NCI indicates National Cancer Institute; MGH, Massachusetts General Hospital; CR, complete response.

  • a

    Includes a patient who did not undergo second-look surgery but had microscopic residual identified during a procedure that was performed to relieve a small bowel obstruction.

Pathologic CR161026 (47)
Microscopic residuala18220 (36)
Macroscopic residual639 (16)
Pathologic CR and microscopic residual: Eligible for surgery34/40 (85)12/15 (80)46/55 (84)
Figure 1.

Kaplan-Meier plots are shown for (A) progression-free survival (n = 62), (B) overall survival (n = 62), (C) progression-free survival for National Cancer Institute (NCI) patients who had liver/splenic involvement (n = 19), and (D) overall survival for NCI patients who had liver/splenic involvement (n = 19).

Drug Administration and Toxicity

A secondary objective was maintenance of dose density in the first 6 cycles. Patients received a median of 10 courses and 7 courses at NCI and MGH, respectively. Details of dose modifications are shown in Table 6. The administered dose was maintained in 89% of patients during the first 6 cycles, and 10 patients (16%) required increased filgrastim doses. Toxicity is presented in Table 7. Seventeen patients (27%) had grade 3 neutropenic fever without a documented source, and 5 patients had grade 4 neutropenic fever that required aggressive management. Filgrastim-induced bone pain was grade 1/2 in 19% of patients and was reduced by premedication with acetaminophen and diphenhydramine. The most common adverse events were grade 2 fatigue (37%), peripheral neuropathy (37%), nausea (27%), and diarrhea (24%). One patient withdrew from the study with grade 2 tinnitus after 2 cycles. No deaths and no severe anaphylactoid reactions occurred on study. There have been no secondary malignancies.

Table 6. Treatment and Dose Modifications in All Cycles for All Patients (N=62)
Dose ChangesNo. of Patients (%)Median No. of Cycles [Range]Reason for Dose Modificationa
  • ANC indicates absolute neutrophil count; PN, peripheral neuropathy; F&N, fatigue and neutropenia; GCSF, granulocyte–colony-stimulating factor filgrastim; T, paclitaxel; C, cyclophosphamide.

  • a

    There was more than 1 reason for dose modification in some patients.

  • b

    For delayed thrombocytopenia.

Full dose55 (89)      
GCSF, 20 μg/kg/d       
 Cycles 1-610 (16)2 [2-6]100000
 Cycle >63 (5)830000
T, 200 mg/m2       
 Cycles 1-64 (6)3 [3-4]12100
 Cycle >612 (19)8 [7-10]29400
T, 175 mg/m2       
 Cycles 1-62 (5)5 [4-6]00101
 Cycle >66 (15)8 [8-10]15101
T, 150 mg/m2       
 Cycles 1-61601000
 Cycle >61901000
C, 500 mg/m2       
 Cycles 1-61310000
 Cycle >63 (5)9 [7-9]20001
P, 50 mg/m2       
 Cycles 1-63b600011
Table 7. Toxicity
ToxicityNo. of Patients (%)
Grade 1Grade 2Grade 3Grade 4
  • a

    Does not include patients with neutropenic fever.

  • b

    National Cancer Institute cohort (n=46).

  • c

    Gilbert syndrome.

Alopecia4 (6)48 (77)0 (0)0 (0)
Allergic reaction2 (3)1 (1)0 (0)0 (0)
Anorexia27 (44)11 (17)0 (0)0v
Arthralgia/myalgia19 (31)12 (19)0 (0)0 (0)
Bone pain29 (47)12 (19)0 (0)0 (0)
Constipation14 (23)5 (8)1 (1)0 (0)
Dehydration/intravenous fluids5 (8)1 (1)0 (0)0 (0)
Depression10 (16)8 (13)0 (0)0 (0)
Diarrhea18 (29)15 (24)8 (13)0 (0)
Dizziness30 (48)2 (3)0 (0)0 (0)
Fatigue19 (31)23 (37)8 (13)0 (0)
Fever11 (17)12 (19)2 (3)0 (0)
Edema15 (24)3 (5)0 (0)0 (0)
Headache15 (24)2 (3)0 (0)0 (0)
Hypotension1 (1)3 (5)0 (0)0 (0)
 Neutropenia1 (1)2 (3)17 (27)a5 (8)a
 Neutropenic fever0 (0)0 (0)17 (27)5 (8)
 Thrombocytopenia12 (26)18 (39)16 (35)0 (0)
Laboratory abnormalitiesb    
 Anemia0 (0)26 (46)20 (43)0 (0)
 Hypomagnesemia33 (72)3 (6)6 (13)0 (0)
 Elevated creatinine0 (0)1 (2)0 (0)0 (0)
 Transaminitis21 (46)5 (11)2 (4)2 (4)
 Hyperbilirubinemia0 (0)2 (4)0 (0)1 (2)c
Mucositis17 (27)5 (8)0 (0)0 (0)
Nausea26 (42)17 (27)3 (5)1 (1)
Neuropathy23 (37)19 (31)12 (19)0 (0)
Rash7 (11)1 (1)0 (0)0 (0)
Tinnitus2 (3)3 (5)0 (0)0 (0)
Vomiting24 (39)11 (17)5 (8)1 (1)
Weight loss3 (5)2 (3)1 (1)0 (0)

The natural history of treatment-induced neuropathy was a secondary objective analyzed at NCI (Table 8). The cisplatin dose was not modified for early peripheral neuropathy but was reduced for 1 patient who developed impaired hearing during Cycle 2. Gabapentin, oxycodone, sertraline, and amitriptyline provided symptomatic relief. Neurologic improvement began with paclitaxel dose reduction or treatment discontinuation. Nineteen patients (41%) had complete resolution of neuropathy with mean time to resolution of 14 months (median, 11 months; range 3-36 months).

Table 8. Follow-Up of Patients With Peripheral Neuropathy (National Cancer Institute Patients; N=46)a
GradeNo. of PatientsMedian Cycle of Onset [Range]Clinical ManifestationPhysical ExaminationDuration [Range], wkbMean Duration, wkNo. of Patients Treated with Pain Relief Medication (%)
  • NCI indicates National Cancer Institute; DTRs, deep tendon reflexes.

  • a

    Each grade manifested per patient is presented.

  • b

    Duration implies either the time to resolution or improvement to a lower grade and excludes patients whose condition was not resolved.

1423 [1-10]Subjective numbness and tingling; gait dysfunctionLoss of DTRs with time4 [0.1-3400]19.54 (9)
2267 [2-10]Subjective functional limitations; pain in extremities; increased tingling and numbnessAbnormal examination: decreased light touch, proprioception, or vibration; wide-based gait16 [0.1.-144]336 (23)
3127 [3-10]Functional limitations; ambulatory difficulty; painful neuropathy; loss of balance; loss of fine motor skillsDecreased/loss of proprioception, light touch, pin prick, vibration; abnormal cerebellar examination; fine motor skill dysfunction15 [0.1-28]91 (8)


We hypothesized that dose-intense paclitaxel, in combination with cyclophosphamide and cisplatin and flexible filgrastim dosing, would yield improved pCR or microscopic response rates compared with historic controls and that this might translate into improved long-term outcomes. We individualized the number of administered treatment cycles, gave consolidation therapy after reassessment surgery, and maintained dose intensity with flexible filgrastim dosing. The portability of this regimen was demonstrated through a multi-institutional collaboration between the NCI and MGH. Administration of this 3-drug regimen resulted in 92% CCR and 84% pCR or microscopic residual disease. The median OS was 5.4 years, and the median PFS was 18.9 months (the median PFS was 15.8 months for patients with stage IV parenchymal organ involvement). Although a direct comparison is neither feasible nor appropriate, at the least, outcomes are comparable and may be superior to outcomes in institutional historic controls and to the outcomes reported in phase 3 carboplatin/paclitaxel control arms for optimally debulked patients. 4, 20, 21

Since the initiation of this trial in 1995, several reports have suggested a benefit of dose-dense paclitaxel in breast and ovarian cancers. 7, 8, 11, 12 Patients with breast cancer who received weekly paclitaxel 80 mg/m2 over 1 hour for 12 doses after standard adjuvant therapy experienced improved PFS and OS compared with patients who were randomized to receive paclitaxel 175 mg/m2 over 3 hours every 3 weeks for 4 doses. 12 A phase 2 trial in patients with stage IIB through IV EOC who received weekly paclitaxel 100 mg/m2 followed by carboplatin AUC, 2 for a total of 18 cycles (which were interrupted by treatment breaks every 6 cycles), had a median PFS of 21 months, and an OS of 43 months. 22 The JGOG recently reported their randomized trial of patients who had stage II through IV EOC and received carboplatin (AUC, 6) with either dose-dense paclitaxel (80 mg/m2 weekly) or paclitaxel 180 mg/m2 every 3 weeks. 11 Twenty percent of those patients had stage II disease, and 45% underwent optimal debulking. PFS was significantly improved in the dose-dense paclitaxel arm (median, 27.9 months vs 17.1 months; P = .0014). OS at 2 years favored the dose-dense cohort (83.6% vs 77.7%; P = .05). Our median PFS of 18.9 months in a cohort of patients with stage IIIC/IV disease suggests the potential of our regimen.

Randomized clinical trials to date have failed to demonstrate the superiority of a triplet regimen using pegylated liposomal doxorubicin or gemcitabine against standard carboplatin and paclitaxel (PFS, 15.4-16.4 months). 4 To our knowledge, no triplets have incorporated dose-dense or dose-intense paclitaxel. In a randomized trial, the addition of epirubicin to paclitaxel and carboplatin (AUC, 5) increased toxicity and did not improve activity. 23 In the absence of a randomized trial, it is impossible to determine the contribution of each component in the triplet to the observed activity and toxicity.

Biasing against overall PFS may be the nonstandard and aggressive follow-up approach used at the NCI of computed tomography scans, physical examinations, and CA 125 measurements every other month for the first year. This may have resulted in the earlier detection of disease recurrence, often before notable increases in CA 125. This detection of early recurrence was followed by prompt therapeutic intervention and, thus, may account for the relatively short median PFS. However, in view of the JGOG data, it is also possible that the dose-intense paclitaxel played a role. The JGOG regimen administered paclitaxel at a dose of 240mg/m2 over 3 weeks, comparable to our dose of 250 mg/m2 every 3 weeks. A previous Gynecologic Oncology Group study failed to demonstrate improved survival in patients who were randomized to receive paclitaxel 200 mg/m2 or cisplatin 100 mg/m2 every 3 weeks compared with patients who received paclitaxel 135 mg/m2 and cisplatin 75 mg/m2; however, the interpretation of this trial was complicated because of an imbalance in the treatment arms. 24 Our dose-intense regimen and the JGOG dose-dense regimen used higher net doses of paclitaxel in combination with a platinum compound. Unlike regimens that contain carboplatin, there was no evidence of long-term bone marrow injury.

All patients experienced some peripheral neuropathy, which is a major reason for discontinuing cisplatin use in EOC, but it was grade ≤2 in the majority of patients. Despite the assumption that cisplatin was the cause of neuropathy in this regimen, neuropathy began to resolve after paclitaxel was decreased or discontinued. We previously described neuropathy as the dose-limiting toxicity of high-dose, single-agent paclitaxel and demonstrated that toxicity was not augmented by the addition of cisplatin. 19, 20 Early dose reduction of paclitaxel for neuropathy in this study occurred in only 3 patients, whereas 9 patients required late dose modification after 6 cycles, consistent with a late and cumulative process. Resolution of neuropathy was observed in 41% of patients, and the median time to resolution was 11 months (range, 3-36 months).

In summary, this triplet dose-intense paclitaxel-based regimen resulted in a high pathologic response rate and encouraging OS. Outcome analyses of this generally poor-prognosis group of patients with advanced-stage disease usually are inferior, which makes the current findings potentially noteworthy. The JGOG results and ours reported herein suggest that the combination of dose-dense/intense paclitaxel and a platinum may warrant further study for the treatment of patients with advanced-stage EOC.


We thank P. Davis, RN; A. Jones-Wells, RN; M. Raggio, RN; and C. Annunziata, MD, PhD for their assistance and A. Bicher, MD for participation.


Supported by the Intramural Research Program of the National Institutes of Health, National Cancer Institute, Center for Cancer Research. Patient care support was provided to the Massachusetts General Hospital investigators from Amgen Corporation.