|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 25, No 11 (April 10), 2007: pp. 1310-1315 © 2007 American Society of Clinical Oncology. DOI: 10.1200/JCO.2006.08.4889 Randomized Trial of Two or Five Computed Tomography Scans in the Surveillance of Patients With Stage I Nonseminomatous Germ Cell Tumors of the Testis: Medical Research Council Trial TE08, ISRCTN56475197—The National Cancer Research Institute Testis Cancer Clinical Studies Group
From the Mount Vernon Cancer Centre, Middlesex; Royal South Hants Hospital, Southampton; Medical Research Council Clinical Trials Unit; The Middlesex Hospital, London; Beatson Oncology Centre, Glasgow; Royal Marsden Hospital, Sutton; Nottingham City Hospital, Nottingham; Cookridge Hospital, Leeds, United Kingdom; and Norwegian Radium Hospital, Oslo, Norway Address reprint requests to Gordon J. Rustin, MD, Mount Vernon Cancer Centre, Northwood, Middlesex HA62RN, United Kingdom; e-mail: grustin{at}nhs.net
Purpose Surveillance is a standard management approach for stage I nonseminomatous germ cell tumors (NSGCT). A randomized trial of two versus five computed tomography (CT) scans was performed to determine whether the number of scans influenced the proportion of patients relapsing with intermediate- or poor-prognosis disease at relapse. Methods Patients with clinical stage I NSGCT opting for surveillance were randomly assigned to chest and abdominal CT scans at either 3 and 12 or 3, 6, 9, 12, and 24 months, with all other investigations identical in the two arms. Three of five patients were allocated to the two-scan schedule. Four hundred patients were required. Results Two hundred forty-seven patients were allocated to a two-scan and 167 to five-scan policy. With a median follow-up of 40 months, 37 relapses (15%) have occurred in the two-scan arm and 33 (20%) in the five-scan arm. No patients had poor prognosis at relapse, but two (0.8%) of those relapsing in the two-scan arm had intermediate prognosis compared with 1 (0.6%) in the five-scan arm, a difference of 0.2% (90% CI, –1.2% to 1.6%). No deaths have been reported. Conclusion This study can rule out with 95% probability an increase in the proportion of patients relapsing with intermediate- or poor-prognosis disease of more than 1.6% if they have two rather than five CT scans as part of their surveillance protocol. CT scans at 3 and 12 months after orchidectomy should be considered a reasonable option in low-risk patients.
The majority of patients with nonseminomatous germ cell tumors present with stage I disease.1 There are several options for management of disease in these patients, including close surveillance, template retroperitoneal lymph node dissection (RPLND), and adjuvant chemotherapy.2-5 Whichever option is chosen, the cure rate should be greater than 99%. The selection of the most appropriate management depends on several factors such as risk of relapse, immediate and delayed toxicity of the treatment, patient preferences, and local preferred practice. Adjuvant therapy may be most appropriate for those with a higher chance of having occult metastatic disease.4,6 The great majority of low-risk and many high-risk patients opt for close surveillance. The nature and frequency of follow-up investigations varies widely across the world. A Medical Research Council (MRC) Clinical Trials Unit (CTU) survey of surveillance practice carried out across the United Kingdom and Oslo (Norway) in 1997 showed that the frequency of computed tomography (CT) scans varied from one to seven during the first year of follow-up and between zero and three during the second year (S. Stenning, personal communication, July 1995). The rationale for more frequent scans is that they should lead to patients' having better-prognosis disease at relapse. In the MRC prospective surveillance study, centers could opt for different scan schedules.2 Although there was a trend toward those who had two or fewer scans during the first year having larger abdominal masses at relapse than did those who had three or more scans (mean diameter, 4.8 v 2.7 cm), this was of borderline significance, and the comparison may have been confounded by other factors that varied across centers, such as CT slice width (S. Stenning, personal communication, July 1995). A trend toward fewer scans was prompted by a study of just 46 patients that showed that all relapses detected after the 3-month CT scan were first noted by clinical examination, serum markers, or chest x-ray.7 The possible benefits of more frequent CT scanning need to be weighed against their cost and the risk of radiation. A whole-trunk CT scan will produce a radiation dose of 10 to 30 mSv.8 Current recommendations are to limit occupational doses to 100 mSv over 5 years.9 Although the risks of diagnostic radiation are based on the hypothesis of a linear nonthreshold dose-effect relation, there are suggestions that the risks from low doses of radiation have been overestimated.10 However, a recent study of cancer risk after occupational exposure suggested that a cumulative exposure of 100 mSv would lead to a 9.7% increased mortality from all cancer, excluding leukemia, and a 19% increase from leukemia.11 It could be postulated that five whole-body CT scans could induce one second cancer in every 200 patients, which is equivalent to the mortality rate from stage I testis cancer. This suggests that we should reduce the number of CT scans to the minimum number required to safely detect relapsing disease. We therefore initiated a trial comparing just two with five CT scans during surveillance. Our aim was to determine whether there was a difference between the groups with respect to prognostic group defined by the International Germ Cell Collaborative Group (IGCCG)12 at relapse, and survival, and to determine which investigations first detected relapse.
Study Design This was a two-arm randomized trial between a surveillance policy of two CT scans over 1 year (at 3 and 12 months after orchidectomy) or five CT scans over 2 years (at 3, 6, 9, 12, and 24 months after orchidectomy).
Patient Inclusion Criteria
Random Assignment
Follow-Up Schedule Minimum CT requirements were agreed to cover the chest and abdomen. Slice widths of 8 or 10 mm were mandated. In addition, if a nonspiral slice-by-slice technique was employed, the optimum slice width should have been 8 mm with 10 mm increments;: a 10 mm slice with 15 mm increments was the upper acceptable limit. All patients should have received oral contrast, and intravenous contrast should have been considered where uncertainty existed with regard to the retroperitoneum. Pelvic CT scans were not recommended unless part of local practice. Further investigations including additional CT scans and tumor marker measurements depended on there being suspicion of relapse because of suspicious nodes on scans or masses on clinical examination, opacities on chest x-ray, or elevated tumor markers. Treatment of relapse was at the clinician's discretion.
Statistical Methods
In the MRC prospective surveillance study,2 only 1% of all patients relapsed with intermediate- or poor-prognosis disease (approximately 5% of all relapsed patients). A difference of 3% in the proportion of patients relapsing with intermediate- or poor-prognosis disease was thought to be clinically important. Hence, using the method of Makuch and Simon,13 this study was powered to exclude an increase of 3% in the proportion of patients relapsing with intermediate- or poor-prognosis disease with 90% power and 5% significance level (one sided). Three of every five patients were allocated the two-scan schedule to facilitate entry of patients from centers for whom two CTs was standard. The primary outcome was assessed via the upper limit of the 90% CI for the difference in the proportions relapsing with IGCCG intermediate or poor prognosis. Relapse-free survival curves were produced using the Kaplan-Meier method, and compared using the log-rank test. The investigations first indicating relapse were compared using the An independent data monitoring committee (IDMC) reviewed the accrual, compliance, and outcome data annually. No formal stopping rules were prespecified; the IDMC recommended continued accrual at each review. The MRC funded the trial center and made a contribution toward local data collection.
Between February 1998 and April 2003, a total of 414 patients were randomly assigned from 32 centers in the United Kingdom, Norway, Australia, and New Zealand to receive two (n = 247) or five CT scans (n = 167). Patient characteristics at random assignment are presented in Table 1. The median time from orchidectomy to random assignment was 6 weeks in both arms, with 83% randomly assigned within 8 weeks of orchidectomy. This analysis was performed on data collected just over 2 years from the date of entry of the last patient, when there was an overall median follow-up of 40 months (40 months in the two-scan arm and 41 months in the five-scan arm). Eighty seven percent of patients have been followed to relapse or for a minimum of 2 years.
Patients were considered to be compliant with their allocated scan schedule if their scan occurred within 1.5 months of their allocated scan time (Table 2). Eighty-two percent of patients in the two-scan arm received just the two abdominal scans, whereas 87.5% of patients on the five-scan arm received all or all but one of their allocated abdominal scans up to relapse or last follow-up. Seventy relapses have been reported, 37 (15%) in the two- and 33 (20%) in the five-scan arm. The relapse-free rate by allocated schedule is shown in Figure 1. Although the relapse-free rate at 2 years was 84.7% (95% CI, 79.5% to 88.8%) in the two-scan arm and 79.6% (95% CI, 72.6% to 85.1%) in the five scan arm, there is no significant difference between the groups (two-sided log-rank P = .21). Vascular invasion was reported in 42 (10%) of all patients registered. Patients with vascular invasion had a 2-year relapse-free rate of 67.9% (95% CI, 46% to 82%) in the 2 scan and 63.6% (95% CI, 36% to 82%) in the five-scan arms compared with 86.7% (95% CI, 81% to 91%) and 81.4% (95% CI, 74% to 87%) respectively in those without vascular invasion.
No patients relapsed with poor-prognosis disease, but two (5.6% of relapses) in the two-scan arm and one (3.0% of relapses) in the five-scan arm had intermediate prognosis at relapse (Table 3). Considering all randomly assigned patients, the percentage relapsing with intermediate-risk disease was 0.8% (90% CI, 0.14% to 2.5%) in the two-scan arm and 0.6% (90% CI, 0.03% to 2.8%) in the five-scan arm, a difference of 0.2% (90% CI, –1.2% to 1.6%). The mean diameter of abdominal mass at relapse was 2.1 cm in the two-scan arm and 2.2 cm in the five-scan arm, with a range of 1 to 4 cm and 1 to 4.5 cm in the two respective arms. In those relapsing after 3 months, the median size for the two- and five-scan groups are 2.1 (range, 1.7 to 2.8 cm) versus 1.9 (range, 1.3 to 2.0 cm), respectively.
The only factor making three patients intermediate risk was an LDH level greater than 1.5x the upper limit of normal (ULN). The first indication of relapse in one patient in the two-scan arm was an HCG of 99 U/L, with LDH 1.87x ULN at 3 months after orchidectomy. His 2-month follow-up visit had shown raised markers and prompted the 3-month CT scan to be performed 2 weeks early, which showed an abdominal mass of 3 cm. The other two-scan patient relapsed 8 months after orchidectomy first detected by chest x-ray and had an LDH level 2.27x ULN (LDH was < 1.5x ULN on all visits before relapse). The intermediate-risk relapse patient in the five-scan arm had his relapse detected by an HCG of 860 U/L, with LDH 1.64x ULN 5 months after his orchidectomy.
A total of nine relapses were picked up by the 12-month CT scan, with only two relapses after that time. Elevated markers picked up both of these, and the patients were in the good-prognosis group. No relapses were picked up by the 24-month CT scan. The proportion of patients whose first indication of relapse was elevated tumor markers or mass on scan differed between the two schedules, as shown in Table 4 (
In the two randomized arms combined, there were a total of 11 patients with lung metastases (five marker negative and with no other involved sites, three marker positive but with no other involved sites, two marker negative but with other involved sites, and 1 marker positive and with other involved sites) and three with mediastinal disease at relapse, all of whom had other signs to confirm relapse. Thus, chest x-ray (three patients, all in the two-scan arm) or CT (two patients, both in the five-scan arm) were the only sign of relapse in five of these patients. No patient had his first and only indication of relapse in the pelvis. There was virtually no difference between the two schedules in the chemotherapy administered at relapse, with 91.5% receiving three or four courses of BEP (bleomycin, etoposide, and cisplatin). Among those who relapsed, a residual mass after chemotherapy was found in 35% in the two-scan arm and 36% in the five-scan arm. Seven of 13 in the two-scan arm and two of 12 in the five-scan arm had surgical resection of the masses. No viable tumor was seen in any of the resected specimens, but differentiated teratoma was seen in two patients. No deaths have yet been reported. One patient developed a second testis primary.
This study can exclude with 95% probability an increase in the proportion of patients on surveillance relapsing with intermediate- or poor-prognosis disease of more than 1.6% if they undergo two as opposed to five CT scans as part of their surveillance for stage I nonseminomatous germ cell tumors. Only 10% of the patients were considered high risk on the basis of vascular invasion. With such small numbers, it is unclear whether only a 3- and 12-month CT scan during surveillance is appropriate in this group. Fewer CT scans will reduce radiation exposure and reduce costs at no risk to these patients. Crawford7 suggested that scans after 3 months were unnecessary, because all relapses after that time were first noted by clinical examination, serum markers, or chest x-ray. This larger study has shown that nine relapses were first detected by the 12-month CT scan, demonstrating its necessity. The fact that no relapses were detected first by the 24-month CT suggests that the slightly lower relapse rate in the two-scan arm is unlikely to result from relapses being missed on that arm. The continued, slightly higher relapse rate in the five-scan arm is most likely caused by chance, and the curves would be expected to come together with longer follow-up and more relapses. It is of interest that, in all intermediate-risk relapses, the only intermediate risk factor was an elevated (> 1.5x ULN) LDH level that varied from 1.64x to 2.27x the ULN. Although LDH has use as a prognostic marker, it is debatable whether it actually contributes to the follow-up because it did not lead to earlier detection of relapse in any patient in this trial, and indeed was not permitted as the only indicator of relapse. This confirms a recent report14 that highlighted the lack of value of serial LDH measurements during surveillance and demonstrated that 5% of elevated levels were false positives. There has been considerable debate as to the need for chest CT scans and chest x-rays during surveillance. Harvey et al15 found only one patient of 168 with relapse isolated in the thorax, which was detected by chest x-ray, and proposed that chest CT scans are not necessary during follow-up. Other studies have suggested that chest x-rays are not necessary if chest CT scans are performed.16-18 The radiation dose of a chest CT scan is about 600-fold that of a chest x-ray.19 Before replacing chest CT scans with chest x-rays, it is important to determine the risks of omitting chest CT scans. There was one patient on the two-scan arm in whom one could postulate that a 6-month chest CT scan could have given a lead time of 2 months, and possibly prevented his relapsing in the intermediate-risk group. Is it justified to do 3- and 12-month abdominal CT scans plus 6-month abdominal and chest CT scans on all 414 patients to possibly prevent one patient's having a worse prognosis? We suggest not. Reassuringly, there have to date been no deaths in either arm. The first indicators of relapse, as shown in Table 4, clearly show that regular measurement of HCG and AFP together with chest x-rays and two abdominal CT scans are necessary for a surveillance program. Only one patient had a palpable mass as the first indicator of relapse, which was confirmed on CT scan. This suggests that clinical examination is of little importance. Currently, the risk criteria in use to guide management of stage I nonseminoma patients are relatively crude. Vascular invasion alone distinguishes between the 50% of patients with a 10% to 15% risk of relapse, and the 50% with a 35% to 40% risk of relapse.6 It is not currently possible to predict the site of relapse from analyzing histologic factors,2 and the discordance between preorchidectomy marker positivity and relapse marker positivity limits the value of preorchidectomy markers in determining the appropriate components of a surveillance policy. A recent prospective evaluation of a baseline [18]fluorodeoxyglucose positron emission tomography scan's ability to identify patients at low risk of relapse has clearly shown that it is unsuitable for this purpose.20 In conclusion, this large randomized trial supports a lack of benefit to more frequent CT scans in the surveillance of patients with stage I nonseminoma, and indirectly supports the omission of chest CTs. Adoption of just two abdominal CT scans and omission of chest and pelvic CT scans during surveillance will lead to a lower radiation dose and a reduced risk of radiation-induced second tumors.
The authors indicated no potential conflicts of interest.
Conception and design: Gordon J. Rustin, Graham M. Mead, Sally P. Stenning, Robert A. Huddart Administrative support: Sally P. Stenning, Sarah J. Kirk Provision of study materials or patients: Gordon J. Rustin, Graham M. Mead, Paul A. Vasey, Nina Aass, Robert A. Huddart, Michael P. Sokal, Jonathan K. Joffe, Stephen J. Harland Collection and assembly of data: Sally P. Stenning, Sarah J. Kirk Data analysis and interpretation: Gordon J. Rustin, Graham M. Mead, Sally P. Stenning, Sarah J. Kirk Manuscript writing: Gordon J. Rustin, Graham M. Mead, Sally P. Stenning, Nina Aass, Robert A. Huddart, Michael P. Sokal, Jonathan K. Joffe, Sarah J. Kirk Final approval of manuscript: Gordon J. Rustin, Graham M. Mead, Sally P. Stenning, Paul A. Vasey, Nina Aass, Robert A. Huddart, Michael P. Sokal, Jonathan K. Joffe, Stephen J. Harland, Sarah J. Kirk
This study was initiated by the Chief Investigator, Gordon Rustin, on behalf of the Medical Research Council Testicular Tumor Working Party (now the National Cancer Research Institute Testicular Cancer Clinical Studies Group), and designed and conducted with the MRC Clinical Trials Unit. The trial was funded by the MRC. We thank all the participating clinicians and their colleagues. Trial Participants UK (382). Royal South Hants Hospital: G.M. Mead (n = 78), P. Simmonds (n = 7); Mount Vernon Hospital: G.J. Rustin (n = 55); Beatson Oncology Centre: S. Kaye (n = 15), P. Vasey (n = 14), J. Russell (n = 3); Nottingham City: M. Sokal (n = 23); Royal Marsden Hospital: R. Huddart (n = 16), A. Horwich (n = 5), D. Dearnaley (n = 2); Cookridge Hospital/St James Leeds: W. Jones and J. Joffe (n = 19), P. Selby (n = 1); The Middlesex Hospital: S.J. Harland (n = 17); North Middlesex Hospital: J. Bridgewater (n = 13); Queen Elizabeth Hospital, Birmingham: M. Cullen (n = 10); Velindre: J. Barber (n = 8), M.D. Mason (n = 2);Bristol Oncology Centre: J, Graham (n = 8), S. Falk (n = 1); Newcastle General Hospital: J. Roberts (n = 7), I. Pedley (n = 1), R. McMenemin (n = 1); Weston Park Hospital: R. Coleman (n = 7), M. Hatton (n = 2); United Lincolnshire Hospitals: T. Sreenivasan (n = 8); Southend General Hospital: A. Robinson (n = 7), R.C. Leonard (n = 1); Clatterbridge Hospital: E. Marshall (n = 4), J. Green (n = 3), P. Clark (n = 1); Leicester Royal Infirmary: F. Madden (n = 4), A. Benghiat (n = 3); Kent Cancer Centre: S. Beesley (n = 5); Guy's Hospital: P. Harper (n = 4); Yeovil District Hospital: S. Falk (n = 3); Scunthorpe General: T. Sreenivasan (n = 3); Royal Devon & Exeter: A. Hong (n = 3); Ipswich Hospital: J. Le Vay (n = 3); Derbyshire Royal Infirmary: P. Chakraborti (n = 3); Raigmore Hospital: D. Whillis (n = 2); Ysbyty Gwynedd: N. Stuart (n = 1); Walsgrave Hospital: A. Stockdale (n = 1); James Cook University Hospital: A. Rathmell (n = 1). Norway (n = 29). Norwegian Radium Hospital, Oslo: N. Aass (n = 29). New Zealand (n = 2). Auckland Hospital: V. Harvey (n = 2). Australia (n = 1). Box Hill Hospital: J. McKendrick (n = 1). DMC members. Chair: Judith Bliss (Institute of Cancer Research, Sutton), Hans-Joachim Schmoll (Universität Halle-Wittenburg), Graham Read (Royal Preston Hospital). TSC members (MRC Urological Cancer Trial Steering Committee). Chair: David Guthrie (Derby); Helena Earl to 2005 (Addenbrookes Hospital, Cambridge); Stan Dische from 2005 (Mount Vernon Hospital); Colin McArdle to 2005 (Glasgow Royal Infirmary); John Schofield from 2005 (Nottingham). MRC CTU staff. Trial manager: Lisa McDonald (previously Eric Lallemand, Rachel Seston, Sharon Naylor, Neil Kelk, Rob Owens, Montse Wells); data manager: Philip Pollock (previously Sharon Naylor, Neil Kelk, Rob Owens, Montse Wells, Lisa McDonald); statisticians: Sarah Kirk, Sally Stenning.
Presented at the 42nd Annual Meeting of the American Society of Clinical Oncology, June 2-6, 2006, Atlanta, GA. For a list of trial participants, please see the online-only Appendix. Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
1. Bhardwa JM, Powles T, Berney D, et al: Assessing the size and stage of testicular germ cell tumours: 1984-2003. BJU Int 96:819-821, 2005[CrossRef][Medline] 2. Read G, Stenning SP, Cullen MH, et al: Medical Research Council prospective study of surveillance for stage I testicular teratoma. J Clin Oncol 10:1762-1768, 1992 3. Foster RS, Donohue JP: Retroperitoneal lymph node dissection for the management of clinical stage I nonseminoma. J Urol 163:1788-1792, 2000[CrossRef][Medline] 4. Cullen MH, Stenning SP, Parkinson MC, et al: Short-course adjuvant chemotherapy in high-risk stage I Nonseminomatous germ cell tumors of the testis: A Medical Research Council report. J Clin Oncol 14:1106-1113, 1996 5. Link R, Allaf M, Pili R, et al: Modeling the cost of management options for stage I nonseminomatous germ cell tumors: A decision tree analysis. J Clin Oncol 23:5762-5773, 2005 6. Vergouwe-Yvonne, Steyerberg-Ewout-W, Eijkemans-Marinus-J-C, et al: Predictors of occult metastasis in clinical stage I nonseminoma: A systematic review. J Clin Oncol 21:4092-4099, 2003 7. Crawford SM, Rustin GJ, Begent RH, et al: Safety of surveillance in the management of stage I anaplastic germ cell tumours of the testis. Br J Urol 61:250-253, 1988[Medline] 8. Rehani M, M, Berry M: Radiation doses in computed tomography : The increasing doses of radiation need to be controlled. BMJ 320:593-594, 2000 9. ICRP: International Commission on Radiological Protection: ICRP: Recommendations of the International Commission on Radiological Protection, ICRP publication 60. Oxford, United Kingdom, Pergamon Press, 1991 10. Feinendegen LE: Evidence for beneficial low level radiation effects and radiation hormesis. Br J Radiol 78:3-7, 2005 11. Cardis E, Vrijheid M, Blettner M, et al: Risk of cancer after low doses of ionising radiation: Retrospective cohort study in 15 countries. BMJ 331:77-80, 2005 12. Group IGCCC: International Germ Cell Consensus Classification: A prognostic factor-based staging system for metastatic germ cell cancers. J Clin Oncol 15:594-603, 1997 13. Makuch R, Simon R: Sample size requirements for evaluating a conservative therapy. Cancer Treat Rep 62:1037-1040, 1978[Medline] 14. Ackers C, Rustin G: The use of lactate dehydrogenase (LDH) as a marker for relapse in patients on surveillance for stage I germ cell tumours. Br J Cancer 94:1231-1232, 2006[CrossRef][Medline] 15. Harvey ML, Geldart TR, Duell R, et al: Routine computerised tomographic scans of the thorax in surveillance of stage I testicular non-seminomatous germ-cell cancer: A necessary risk? Ann Oncol 13:237-242, 2002 16. Gels M, Hoekstra H, Sleijfer D, et al: Detection of recurrence in patients with clinical stage I nonseminomatous testicular germ-cell tumors and consequences for further follow-up: A single-center 10-year experience. J Clin Oncol 13:1188-1194, 1995[Abstract] 17. White P, Adamson D, Howard G, et al: Imaging of the thorax in the management of germ-cell testicular tumours. Clin Radiol 54:207-211, 1999[CrossRef][Medline] 18. Sharir S, Jewett M, Sturgeon JF, et al: Progression detection of stage I nonseminomatous testis cancer on surveillance: Implications for the followup protocol. J Urol 161:472-476, 1999[CrossRef][Medline] 19. Collie D, Wright A, Williams J, et al: Comparison of spiral-acquisition computed tomography and conventional computed tomography in the assessment of pulmonary metastatic disease. Br J Radiol 67:436-444, 1994 20. Huddart R, O'Doherty M, Padhani A, et al: A prospective study of 18FDG PET in the prediction of relapse in patients with high risk stage 1 non-seminomatous germ cell cancer: MRC study TE22. J Clin Oncol 24:222s, 2006 (suppl; abstr 4520) Submitted July 31, 2006; accepted December 1, 2006.
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2007 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
2 test. It should be noted that this comparison does not include all randomly assigned patients. Although this comparison is potentially subject to bias if the trial interventions affect the number of relapses reported at the time of the analysis, this is minimized by the mature follow-up reported herein. 
