Professional Documents
Culture Documents
2.lukka2005 52.5 20ses
2.lukka2005 52.5 20ses
DOI: 10.1200/JCO.2005.06.153 J Clin Oncol 23:6132-6138. © 2005 by American Society of Clinical Oncology
6132
Downloaded from jco.ascopubs.org on March 25, 2015. For personal use only. No other uses without permission.
Copyright © 2005 American Society of Clinical Oncology. All rights reserved.
Hypofractionation in Prostate Cancer
the presumed low ␣/ ratio of prostate tumors compared with chest x-ray and bone scan were performed within 12 weeks of
the surrounding normal tissue. This randomized trial was de- random assignment. Patients were required to have a PSA mea-
signed to determine whether a hypofractionated treatment surement within 6 weeks of commencing radiotherapy.
Patients were randomly assigned to receive prostate irradia-
regimen of 52.5 Gy in 20 fractions administered over 28 days
tion consisting of either 66 Gy in 33 fractions over 45 days (long
was as effective as the conventional treatment regimen of 66 Gy arm) or 52.5 Gy in 20 fractions over 28 days (short arm). Radiation
in 33 fractions administered over 45 days for men with early- was delivered daily, from Monday through Friday. The intention
stage prostate cancer. was to treat the prostate gland alone with a 1.5-cm margin. CT
planning was mandatory for all patients. Shielding, where appro-
priate, was used to keep the planning target volume under 1,000
PATIENTS AND METHODS mL. Patients were treated in the supine position with a full bladder
before each treatment. The target volume was treated using a
Study Population four-field box technique or using three fields in patients with
Men with early-stage adenocarcinoma of the prostate (T1-2 prosthetic hips. The borders defining the treatment volume in-
according to International Union Against Cancer TNM classifica- cluded a 1.5-cm margin on all sides except posteriorly, where, at
tion) were eligible for the trial. Patient exclusion criteria were as the discretion of the radiation oncologist, the margin could be
follows: a prostate-specific antigen (PSA) level of more than 40 reduced to 1 cm. Shielding was done using either multileaf colli-
g/L; previous therapy for prostate carcinoma (other than biopsy mator or poured blocks. Differential weighting was used to achieve
or transurethral resection of the prostate); previous hormone optimal dose distribution. The dose was prescribed at the iso-
therapy; prior or active malignancy other than nonmelanoma skin center, and the dose within the target volume was required to be
cancer, colon cancer, or thyroid cancer treated a minimum of 5 within 5% of the isocenter dose. All patients were treated with a
years before the trial and presumed cured; a simulated volume ⱖ 10-MV linear accelerator. To confirm adequate coverage, port
exceeding 1,000 mL; previous pelvic radiotherapy; presence of films were taken of the anterior and lateral pelvic fields of each
inflammatory bowel disease; diagnosis of serious nonmalignant patient in the treatment position on the first day of treatment. For
disease that would preclude radiotherapy or surgical biopsy; geo- further quality assurance, the first 30 patients from each center
graphically inaccessible for follow-up; a psychiatric or addictive underwent real-time review. If this was satisfactory, random spot
disorder that would preclude obtaining informed consent or ad- checks were performed on 20% of the remaining plans.
herence to protocol; inability to commence radiotherapy within
26 weeks of the date of last prostatic biopsy; and failure to provide Follow-Up
informed consent to participate in the clinical trial. After completion of radiation therapy, patients were ob-
Participating centers included eight Ontario regional cancer served after 4 weeks, at 6 months after random assignment, and
centers (Hamilton, Princess Margaret Hospital, Toronto Sunny- then every 6 months thereafter. At each visit, patients underwent a
brook, Sudbury, Kingston, Windsor, London, and Thunder Bay) medical history and physical examination with a digital rectal
and eight additional Canadian centers (the Vancouver and Fraser examination, except for the first 4-week postradiotherapy visit.
Valley Cancer Centres in British Columbia, CancerCare Manitoba Urinary and rectal symptoms were assessed at baseline and then
in Winnipeg, the Nova Scotia Cancer Centre in Halifax, the Saint weekly during radiotherapy (four times for the 52.5-Gy group and
John Regional Hospital in New Brunswick, the Dr Leon Richard six times for the 66-Gy group) to document acute toxicity. Subse-
Oncology Centre in Moncton, the Newfoundland Cancer Treat- quently, toxicity was assessed at 2 weeks (by telephone) and at 4
ment and Research Foundation in St John’s, and the Saskatoon weeks after radiotherapy, at 6 weeks after radiotherapy for the
Cancer Centre in Saskatchewan). Recruitment for the centers 52.5-Gy group, and at visits every 6 months thereafter. PSA was
outside Ontario was coordinated through the Clinical Trials measured at each follow-up visit starting with the 6-month post-
Group of the National Cancer Institute of Canada. Written in- randomization visit. Transrectal ultrasound-guided biopsy of the
formed consent was obtained from eligible patients before assign- prostate was scheduled 2 years after completion of radiotherapy.
ment to treatment. The institutional review board at each Patient follow-up continued until clinical evidence of metastases
participating center approved the study protocol. was found.
www.jco.org 6133
Downloaded from jco.ascopubs.org on March 25, 2015. For personal use only. No other uses without permission.
Copyright © 2005 American Society of Clinical Oncology. All rights reserved.
Lukka et al
outcomes included local persistence of tumor on biopsy of the For the toxicity assessment, acute toxicity was defined as
prostate at 2 years (biopsy positivity), overall survival, and acute events occurring up to 5 months after randomization, and late
and late radiation-induced toxicity. toxicity included events occurring after 5 months. Risk differences
The criterion for local disease recurrence was based on the (long arm minus short arm) and 95% CIs were calculated using
clinical evaluation of the prostate at the time of digital rectal the modified Wilson score method.14 All statistical analyses, which
examination. Signs or symptoms of local recurrence were con- were based on the intent-to-treat principle, were undertaken by
firmed through prostate biopsy. Biopsies reported as suspicious the study statistician using data provided by the Coordinating and
were classified as positive, and biopsies reported as equivocal were Methods Centre of the Ontario Clinical Oncology Group located
classified as negative. Criteria for distant disease recurrence of in Hamilton, Ontario, Canada.
metastases outside of the prostate included recurrent tumor found
in regional pelvic lymph nodes, bone (abnormal bone x-rays or
RESULTS
bone scan), liver (abnormal liver scan, ultrasound, or CT scan),
and lung (abnormal chest x-ray consistent with metastases).
Asymptomatic patients who presented during follow-up with an Study Population
increasing PSA were investigated at the physician’s discretion with Nine hundred thirty-six patients were recruited for the
a further bone scan and an abdominal and pelvic CT scan (if the trial between March 1995 and December 1998; 470 patients
bone scan was negative). If both scans were negative, a prostate were randomly assigned to receive 66 Gy in 33 fractions
biopsy was undertaken to determine clinically the cause of the over 45 days, and 466 patients were randomly assigned to
increase. In patients with only PSA failure, effort was made to
receive 52.5 Gy in 20 fractions over 33 days. The median
prolong the interval between treatment and biopsy to avoid false-
positive biopsy. Treatment at the time of relapse for metastatic follow-up time for all patients was 5.7 years (minimum, 4.5
symptoms, clinical evidence of recurrence or residual disease, or years; maximum, 8.3 years). The treatment groups were
PSA failure was at the discretion of the treating physician. Physi- well balanced in terms of baseline characteristics and risk
cians were asked to refrain from intervening based solely on the group stratification (Table 1).
results of the 2-year postradiotherapy biopsy. During the study, seven patients did not receive radia-
The investigating physician or clinical trials nurse assessed tion (five patients in the long arm and two patients in the
radiation toxicity using the standardized National Cancer Insti- short arm), one patient in the short arm crossed over to the
tute of Canada toxicity scale and graded toxicity according to
long arm, and three patients in the long arm received alter-
specific criteria for each symptom on a 5-point scale ranging from
0 (nontoxic) to 4 (severe toxicity). The effects of radiation therapy nate radiation dose and fraction combinations (22 Gy in 11
on the GI system (anorexia, diarrhea, GI bleeding, nausea, pain/ fractions, 52 Gy in 26 fractions, and 52.8 Gy in 20 fractions).
cramping, proctitis, and vomiting), the genitourinary system All other patients completed the prescribed treatment.
(bladder changes, cystitis, fistula formation, frequency, hematu- BCF
ria, ureteral obstruction, and genitourinary pain), the skin, and
fatigue/general malaise were evaluated using the 5-point scale.
At the time of analysis, 236 patients in the long arm and
263 patients in the short arm experienced BCF (Fig 1). Table
Statistical Analysis 2 lists the events according to the outcome type and radia-
The study was designed as a noninferiority investigation re- tion therapy schedule. The 100 patients who died during the
quiring 450 patients per treatment arm. The sample size calcula- study of non–prostate cancer causes without experiencing
tion was based originally on the ability to demonstrate that the BCF (54 patients in the long arm and 46 patients in the short
percentage of patients with biopsy positivity at 2 years in the short
arm) were censored on the date of death. At 5 years, the
arm was no worse than the long arm (suggested in the literature to
be 25%) by more than an absolute difference of 7.5%, with a
Kaplan-Meier estimates of BCF in the long arm and short
one-sided ␣ ⫽ .05 and a power of 80%. With time from random arm were 52.95% and 59.95%, respectively. The difference
assignment to BCF as the primary outcome measure and a BCF was ⫺7.0% (90% CI, ⫺12.58% to ⫺1.42%). Because the
probability over 5 years estimated at approximately 40% in either lower bound is less than the predefined tolerance of ⫺7.5%,
group, we estimated that a sample size of 940 men provided we could not exclude the possibility of the short arm being
approximately 80% power to demonstrate noninferiority with the inferior. Overall, a hazard ratio of 1.18 (95% CI, 0.99 to
same 7.5% tolerance. 1.41) in favor of the long arm was estimated.
Overall survival was defined as the time from random assign-
Biochemical failure was also assessed using the
ment to death from any cause or to the date of the last visit for
patients still alive. For the time to BCF outcome, patients without Vancouver15 and Houston16 definitions of post-treatment
events were censored on the date of their last visit or prostate- failure. Using the Vancouver criteria, 50.0% of patients in the
unrelated death. Treatment groups were compared using a two- long arm and 55.8% of patients in the short arm experienced
sided 90% CI for the difference (long arm minus short arm) of the PSA failure. Using the Houston definition, 37.7% of patients in
cumulative BCF probability at 5 years. A lower confidence limit the long arm and 42.3% of patients in the short arm experi-
greater than ⫺7.5% indicated noninferiority. All time to event enced PSA failure.
outcomes were summarized using the Kaplan-Meier method. In
addition, hazard ratios (short arm relative to long arm) and their Overall Survival
respective 95% CIs were calculated for BCF and overall survival Overall, there were 166 deaths during the study period
using the Cox proportional hazards model. (89 deaths in the long arm and 77 deaths in the short arm).
Downloaded from jco.ascopubs.org on March 25, 2015. For personal use only. No other uses without permission.
Copyright © 2005 American Society of Clinical Oncology. All rights reserved.
Hypofractionation in Prostate Cancer
www.jco.org 6135
Downloaded from jco.ascopubs.org on March 25, 2015. For personal use only. No other uses without permission.
Copyright © 2005 American Society of Clinical Oncology. All rights reserved.
Lukka et al
Downloaded from jco.ascopubs.org on March 25, 2015. For personal use only. No other uses without permission.
Copyright © 2005 American Society of Clinical Oncology. All rights reserved.
Hypofractionation in Prostate Cancer
comparable late toxicity.24,25 The small but significant false-negative rates.34,35 Last, differentiating residual tumor
increase in acute toxicity could be related to a higher net rate from postradiotherapy effects can be difficult to judge even
of stem-cell depletion in the rectal and bladder mucosa. when assessed by an experienced tumor pathologist.32 The
This depletion results from an increased rate of dose accumu- postradiotherapy biopsy results presented here are based on
lation in the hypofractionated arm compared with the con- the report from the original pathologist. Overall, caution
ventional arm (13.1 v 10 Gy/wk, respectively). This same should be exercised when interpreting the 2-year biopsy
mechanism has been purported to cause the increase in acute results, particularly when drawing comparisons to the re-
toxicity observed in hyperfractionation protocols.26 sults of biochemical failure.
Compared with other tumors, several studies have con- In the early 1990s when this trial was designed, a four-
cluded that the ␣/ ratio for prostate cancer is low, in the field technique was the standard method of radiation deliv-
range of 1.0 to 3.0.11,27 However, the debate on this issue ery. In recent years, conformal radiation has evolved as the
continues. Some recent publications have concluded that technique more often used, and in some centers, intensity-
the low ␣/ ratio reported in previous studies may actually modulated radiation therapy is used. These radiotherapy
be an artifact resulting from tumor hypoxia.28-30 If the ␣/ tools allow for higher doses of radiation to be delivered to
value for prostate cancer is less than for normal tissue the prostate and for the margin around the prostate to be
(nominally 3 Gy), it could be argued that a high dose per decreased.36-38 Standard total doses of external-beam radi-
fraction regimen would result in a therapeutic gain by im- ation used to treat early-stage prostate cancer patients now
proving tumor control with acceptable late toxicity.25,27 range from 70 to 78 Gy.39,40 It could be argued that the total
Models based on a low ␣/ ratio predict that a therapeutic gain dose of 66 Gy delivered in 2-Gy fractions used in the long
would require a larger fraction size and a higher biologic effec- arm would be biologically comparable to 70.2 Gy delivered
tive dose than what was used in this study. In fact, for all ␣/ in 1.8-Gy fractions.10,11 However, this dose is still low com-
ratios between 0.85 and infinity, the ␣/ model predicts that pared with current clinical doses. Both the technique and
our chosen hypofractionated regimen would result in a lower dose of radiation used in this study limit the generalizability of
tumor control rate relative to the conventional fractionation the results to current practice. Recent information supports
schedule. Therefore, given our current understanding of the the use of adjuvant hormones and potentially pelvic radiother-
␣/ ratio, the results of this study do not contribute meaning- apy in the treatment of high-risk patients. However, when the
fully to the debate on the ␣/ ratio for prostate cancer. Another study was designed, this was not universally accepted as stan-
interpretation of this data arrived at by comparison of the dard practice and, hence, was not part of the study design.
study results with the dose-response curve from published This is the first reported randomized trial comparing a
retrospective studies has suggested that the results are consis- high-dose hypofractionated radiation schedule to a longer
tent with a low ␣/ ratio for prostate cancer.25,31 conventional dose per fraction schedule. Given the results
The dose chosen in the hypofractionated regimen was of this study, we cannot exclude the possibility that the
based on clinical experience from two Canadian centers and chosen hypofractionated radiation treatment regimen is infe-
published studies at the time of study design.5-8 Given the rior to the standard regimen for early-stage prostate cancer
acute toxicity results in this study, it would have been im- patients. However, this is only a preliminary analysis and dis-
possible to further escalate the dose using our treatment cussion of the results; 5 years is not considered a long period of
technique. Thus, evaluation of high dose per fraction regi- follow-up for prostate cancer. Further evaluation at 10 years of
mens using techniques that reduce toxicity, such as the outcomes between the two arms of the study is awaited.
conformal-beam radiation or intensity-modulated therapy, The results discussed here do have important implications for
is recommended.9 our understanding of the radiobiology of prostate cancer. Fu-
The percentage of positive biopsies reported in this trial ture studies evaluating high dose per fraction regimens are
was similar in the long and short treatment arms, which is a encouraged using modern methods of radiation delivery.
finding that somewhat contradicts the results of biochemi-
■ ■ ■
cal failure. Postradiotherapy biopsies are no longer a gold
standard for measurement of treatment effect because they Acknowledgment
are associated with inherent problems that limit interpreta- We thank S. Bouma for manuscript preparation; T.
tion. First, there is evidence suggesting that 2 years is too Finch, S. Chambers, and Q. Guo for data management;
early and that biopsies should not be taken before 30 to 36 and Drs Daya, Jones, Scrigley, Wright, Dayes, Wu, Poon,
months after radiation to minimize false-positive rates.32 Hodson, Fyles, Dixon, Milosevic, Rheaume, Sagar, Youssef,
Second, sampling error of biopsies as a result of number and and Whelan for their assistance with this study
length of obtained cores can affect biopsy evaluation.33
Several studies have documented that increased biopsy Authors’ Disclosures of Potential
sampling (standard of six biopsies v 10 or 12) has a better Conflicts of Interest
probability of detecting prostate cancer and may lower The authors indicated no potential conflicts of interest.
www.jco.org 6137
Downloaded from jco.ascopubs.org on March 25, 2015. For personal use only. No other uses without permission.
Copyright © 2005 American Society of Clinical Oncology. All rights reserved.
Lukka et al
15. Pickles T, Duncan GG, Kim-Sing C, et al: cancer: Implications for the alpha/beta ratio. Int J
REFERENCES PSA relapse definitions: The Vancouver rules Radiat Oncol Biol Phys 57:391-401, 2003
show superior predictive power. Int J Radiat 29. Orton C: In regard: Incorporating clinical
1. Hank GE: External beam radiation therapy Oncol Biol Phys 43:699-700, 1999 measurements of hypoxia into local tumor con-
for clinically localized prostate cancer: Patterns 16. American Society for Therapeutics Radiol- trol modeling of prostate cancer. Int J Radiat
of care studies in the US. NCI Monogr 7:75-84, ogy and Oncology Consensus Panel: PSA Re- Oncol Biol Phys 58:1637, 2004
1988 30. Nahum A, Chapman J: In response: Incor-
lapse Definitions. New Orleans, LA, American
2. Sagerman RH, Chun HC, King GA, et al: porating clinical measurements of hypoxia into
Society for Therapeutics Radiology and Oncol-
External beam radiotherapy for carcinoma of the local tumor control modeling of prostate cancer.
ogy, 2002
prostate. Cancer 63:2468-2474, 1989 Int J Radiat Oncol Biol Phys 58:1637-1639, 2004
17. Freiha FS, Bagshaw MA: Carcinoma of the
3. Zagars GK, Von Eschenback AC, Johnson 31. Zietman A: A randomized trial comparing
prostate: Results of post-irradiation biopsy. Pros-
DE, et al: The role of radiotherapy in stages A2 two fractionation schedules for patients with
tate 5:19-25, 1984
and B carcinoma of the prostate. Int J Radiat
18. Kiesling VJ, McAninch JW, Goekel JL, et localized prostate cancer. Presented at 45th An-
Oncol Biol Phys 14:701-709, 1988
al: External beam radiotherapy for adenocarci- nual Meeting of the American Society for Ther-
4. Perez CA, Pilepich MV, Garcia D, et al:
noma of the prostate: A clinical follow-up. J Urol apeutic Radiology and Oncology, Salt Lake City,
Definitive radiation therapy in carcinoma of the
124:851-854, 1980 UT, October 19-23, 2003
prostate localized to the pelvis: Experience at the
19. Crook J, Robertson S, Collin G, et al: 32. Crook J, Malone S, Perry G, et al: Postra-
Mallinckrodt Institute of Radiology. NCI Monogr
Clinical relevance of TRUS, biopsy, and serum diotherapy prostate biopsies: What do they really
7:87-94, 1988
PSA following radiation therapy for carcinoma of mean? Results for 498 patients. Int J Radiat
5. Duncan W, Warde P, Catton CN, et al:
the prostate. Int J Radiat Oncol Biol Phys 27:31- Oncol Biol Phys 48:355-367, 2000
Carcinoma of the prostate: Results of radical
37, 1993 33. Ruijter E, van Leenders G, Miller G, et al:
radiotherapy (1970-1985). Int J Radiat Oncol Biol
20. Kuban DA, Thames HD, Levy LB: Radia- Errors in histological grading by prostatic needle
Phys 26:203-210, 1993
biopsy specimens: Frequency and predisposing
6. Read G, Pointon RC: Retrospective study tion for prostate cancer: Use of biochemical
factors. J Pathol 192:229-233, 2000
of radiotherapy in early carcinoma of the pros- failure as an endpoint following radiotherapy.
34. Chan TY, Chan DY, Stutzman KL, et al:
tate. Br J Urol 63:191-195, 1989 World J Urol 21:253-264, 2003
Does increased needle biopsy sampling of the
7. Kearsley JH: High dose radiotherapy for 21. Horwitz EM, Uzzo RG, Hanolon AL, et al:
prostate detect a higher number of potentially
localized prostatic cancer: An analysis of treat- Modifying the American Society for Therapeutic
insignificant tumors? J Urol 166:2181-2184,
ment results and early complications. Med J Radiology and Oncology definition of biochemi-
2001
Aust 144:624-628, 1986 cal failure to minimize the influence of backdat-
8. Preston CI, Duncan W, Kerr GR: Radical 35. Bauer JJ, Zeng J, Zhang W, et al: Lateral
ing in patients with prostate cancer treated with
treatment of prostatic carcinoma of the prostate. biopsies added to the traditional sextant prostate
3-dimensional conformal radiation therapy alone.
Br J Urol 54:723-735, 1982 biopsy pattern increases the detection rate of
J Urol 169:2153-2157, 2003
9. Kupelian PA, Reddy CA, Klein EA, et al: prostate cancer. Prostate Cancer Prostatic Dis
22. D’Amico AV, Moul JW, Carroll PR, et al:
Short course intensity-modulated radiotherapy 3:43-46, 2000
Surrogate end point for prostate cancer-specific
(70 Gy at 2.5 Gy per fraction) for localized pros- 36. Brundage M, Lukka H, Crook J, et al: The
mortality after radical prostatectomy or radiation
tate cancer: Preliminary results on late toxicity use of conformal radiotherapy and the selection
therapy. J Natl Cancer Inst 95:1376-1383, 2003
and quality of life. Int J Radiat Oncol Biol Phys of radiation dose in T1 or T22 low or intermediate
23. Sandler HM, DeSilvio ML: Surrogate end risk prostate cancer: A systematic review. Ra-
51:988-993, 2001 points for prostate cancer: What is prostate-
10. Brenner DJ: Toward optimal external- diother Oncol 64:239-250, 2002
specific antigen telling us? J Natl Cancer Inst 37. Zelefsky MJ, Cowen D, Fuks Z, et al: Long
beam fractionation for prostate cancer. Int J
95:1352-1353, 2003 term tolerance of high dose three-dimensional
Radiat Oncol Biol Phys 48:315-316, 2000
24. Brenner DJ, Martinez AA, Edmundson GK, conformal radiotherapy in patients with localized
11. Brenner DJ, Hall EJ: Fractionation and
et al: Direct evidence that prostate tumors show prostate carcinoma. Cancer 85:2460-2468, 1999
protraction for radiotherapy of prostate carci-
high sensitivity to fractionation (low alpha/beta 38. Zelefsky MJ, Fuks Z, Hunt M, et al: High
noma. Int J Radiat Oncol Biol Phys 43:1095-
ratio), similar to late responding normal tissue. dose intensity modulated radiation therapy for
1101, 1999
Int J Radiat Oncol Biol Phys 52:6-13, 2002 prostate cancer: Early toxicity and biochemical
12. Duchesne GM, Peters LJ: What is the
alpha/beta ratio for prostate cancer? Rationale 25. Fowler J, Ritter M, Chappell R, et al: What outcome in 772 patients. Int J Radiat Oncol Biol
for hypofractionated high-dose-rate brachyther- hypofractionated protocols should be tested for Phys 53:1111-1116, 2002
apy. Int J Radiat Oncol Biol Phys 44:747-748, prostate cancer? Int J Radiat Oncol Biol Phys 39. Kupelian PA, Reddy CA, Carlson TP, et al:
1999 56:1093-1104, 2003 Preliminary observations on biochemical relapse-
13. American Society for Therapeutic Radiol- 26. Baumann M: Accelerated radiation ther- free survival rates after short-course intensity-
ogy and Oncology Consensus Panel: Consensus apy in non-small cell lung cancer. Radiother modulated radiotherapy (70 Gy at 2.5 Gy/fraction)
Statement: Guidelines for PSA following radia- Oncol 52:97-99, 1999 for localized prostate cancer. Int J Radiat Oncol Biol
tion therapy. Int J Radiat Oncol Biol Phys 37: 27. Fowler J, Chappell R, Ritter M: Is alpha/ Phys 53:904-912, 2002
1035-1041, 1997 beta for prostate tumors really low? Int J Radiat 40. Pollack A, Zagars GK, Starkschall G, et al:
14. Newcombe RG: Interval estimation for the Oncol Biol Phys 50:1021-1031, 2001 Prostate cancer radiation dose response: Re-
difference between independent proportions: 28. Nahum A, Movsas B, Horwitz E, et al: sults of the M. D. Anderson phase III ran-
Comparison of eleven methods. Stat Med 17: Incorporating clinical measurements of hypoxia domized trial. Int J Radiat Oncol Biol Phys
873-890, 1998 into tumor local control modeling of prostate 53:1097-1105, 2002
Downloaded from jco.ascopubs.org on March 25, 2015. For personal use only. No other uses without permission.
Copyright © 2005 American Society of Clinical Oncology. All rights reserved.