- Email: [email protected]
The impact of enrollment in clinical trials on survival of patients with glioblastoma
Journal of Clinical Neuroscience 19 (2012) 1530–1534
Contents lists available at SciVerse ScienceDirect
Journal of Clinical Neuroscience journal homepage: www.elsevier.com/locate/jocn
Clinical Study
The impact of enrollment in clinical trials on survival of patients with glioblastoma Tal Shahar a, Erez Nossek a, David M. Steinberg b, Uri Rozovski c, Deborah T. Blumenthal d, Felix Bokstein d, Razi Sitt a, Sigal Freedman a, Benjamin W. Corn e, Andrew A. Kanner a, Zvi Ram a,⇑ a
Department of Neurosurgery, Tel Aviv Medical Center, 6 Weizman Street, Tel Aviv 64239, Israel Department of Statistics and Operations Research, Raymond and Beverly Sackler Faculty of Exact Sciences, Tel Aviv University, Tel Aviv, Israel c Department of Hematology and Bone Marrow Transplantation, Tel Aviv Medical Center, Tel Aviv, Israel d Oncology Division, Tel Aviv Medical Center, Tel Aviv, Israel e Institute of Radiotherapy, Tel Aviv Medical Center, Tel Aviv, Israel b
a r t i c l e
i n f o
Article history: Received 31 March 2012 Accepted 4 April 2012
Keywords: Clinical trials Glioblastoma Neuro-oncology Neurosurgery Survival analysis
a b s t r a c t The impact of enrollment in a clinical study on the survival of patients with glioblastoma has not been established. We retrospectively analyzed 564 patients with newly diagnosed glioblastoma treated between 1995 and 2008. They were divided into those enrolled in a clinical trial and randomized to a treatment or control arm, and those not enrolled and who received best standard of care (BSC). The three groups were matched for age and Karnofsky performance scale (KPS) score at presentation, and included only patients who underwent at least one tumor resection. Survival analysis was performed and multivariate Cox proportional hazards model and recursive partitioning analysis (RPA) identified predictors of survival. Following the matching process, 261 patients remained to form the final cohort. Of the 124 patients enrolled in a study, 81 (31.0%) were randomized to the treatment and 43 (16.5%) to the control arms. The overall median survival for the BSC (n = 137), control, and treatment groups was 11.57 months (95% confidence interval [CI], 10.41–12.73), 16.27 months (95% CI, 14.10–18.43) and 16.10 months (95% CI, 14.34–17.86), respectively (p = 0.002). Participation in a clinical trial, regardless of the arm, was a significant predictor of survival, as were age and KPS at diagnosis. The RPA also demonstrated a favorable impact of participation in a clinical trial. Additional tumor resections and various treatment modalities were administered with significantly higher frequency among patients enrolled in clinical studies. Thus, enrollment in a clinical study carried a significant survival advantage for patients with glioblastoma, raising practical and ethical issues regarding the quality of care of patients who receive ‘‘standard’’ therapy. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction Glioblastoma is the most common primary malignant brain tumor in adults. Despite multimodality treatment, the prognosis remains poor, with a median survival of approximately 1 year following diagnosis.1,2 Prognosis of glioblastoma patients is not uniform.3 A wide range of prognostic factors and their relative importance in influencing outcome have been studied by multivariate analyses, with age and Karnofsky performance scale (KPS) score on admission constituting the most significant predictors of survival for patients with glioblastoma.4 Histological characteristics,5,6 de novo glioblastoma or secondary transformation7 and tumor location8 also affect survival. Current treatment strategies for glioblastoma include surgery, radiation therapy and chemotherapy.2 Patients who have undergone gross total tumor resection have a survival advantage compared to those who underwent partial resection or biopsy only.9 ⇑ Corresponding author. Tel.: +972 3 697 4273; fax: +972 3 697 4680. E-mail address: [email protected] (Z. Ram). 0967-5868/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jocn.2012.04.005
Response to radiation therapy and chemotherapy were also predictive factors for survival.2,4,10 Recent advances in understanding the molecular events that take part in the pathogenesis of glioma have led to investigation of many new therapeutic options in clinical trials. A search of our brain tumor database revealed 10 clinical trials11–19 that had been conducted in patients diagnosed with glioblastoma between March 1995 and May 2008 (Table 1). Outcome analyses of such studies are based on comparison of the treatment group to either a control group receiving ‘‘best standard of care’’ (BSC) therapies or, in fewer cases, comparison to historical controls. The actual impact of being enrolled in a clinical trial on survival has not been fully investigated. 2. Methods 2.1. Study design The current study is a retrospective analysis of prospectively collected data from the Tel Aviv Medical Center (TAMC) brain tumor database. Approval from the TAMC institutional review board
T. Shahar et al. / Journal of Clinical Neuroscience 19 (2012) 1530–1534 Table 1 Clinical trials conducted in patients diagnosed with glioblastoma between March 1995 and May 2008 Clinical trialref
No. patients enrolled
Cerepro13 Gliadel14 GTI 10711 GTI 11511 IL1312a Novocure17 RTOG 052531 RTOG 062532 Taxol15 Transmid16
31 14 2 10 26 14 18 6 9 10
Note: the total number of patients in the table is greater than the number in our group of patients who were enrolled into clinical trials due to the fact that some were enrolled into more than one clinical trial. a IL13 includes three separate clinical trial studies.
(Number: 003108) was obtained. We evaluated data from all patients with de novo glioblastoma who were treated in our institution between 1995 and 2008. The evaluated variables were age and KPS on admission, number of tumor resection surgeries, histopathological evaluation of the tumors, date of diagnosis (determined as the date of the first tissue sampling), date of death, calculated survival period (from diagnosis until death) and participation in a clinical trial. The neurosurgical procedures and all nonsurgical treatment modalities were also recorded. The patients were divided into three groups according to treatment modalities: those who were not enrolled in a study and had received BSC, and those enrolled in at least one clinical trial and randomized to either the control or experimental treatment arms. The non-trial patients (BSC) received therapy considered ‘‘standard’’ at the time of their treatment. From 1995 until the end of 2004, standard therapy for newly diagnosed glioblastoma in Israel was tumor resection surgery, followed by radiotherapy. As of the beginning of 2005, standard of care for glioblastoma patients in our institute was tumor resection surgery, followed by concomitant treatment with radiotherapy plus temozolomide (Stupp protocol).2 In order to overcome the possible selection bias used in the process of recruiting patients into clinical trials, we have selected patients who had met, or would have met, the standard inclusion criteria for the various clinical trials. We matched the groups (BSC, treatment and controls) for age and KPS at presentation and included only patients who had undergone at least one tumor resection surgery. Thus, patients with a KPS of <60, older than 70 years of age and who were not considered candidates for tumor resection surgeries (that is, biopsy only) were excluded. Survival analysis was calculated for each of the groups, and multivariable analyses were performed to identify independent clinical prognostic factors for survival. 2.2. Measurement of study parameters Clinical and therapeutic factors were analyzed using the univariate Student’s t-test or chi-squared (v2) test for determining their prognostic impact on overall survival. Time to event distribution (for example, survival) was assessed by the Kaplan–Meier survival curves and compared using the Log Rank (Mantel–Cox) test. The importance of predictor variables was assessed using Cox proportional hazards regressions. The analysis was done with the R statistical environment and the survival library by Therneau.20 Survival experience was also analyzed by recursive partitioning analysis (RPA) using the R statistical environment and the Recursive PARTitioning (RPART) library by Therneau and Atkinson.21 Patients who were alive on the last day of follow-up were considered to have censored survival times.
1531
3. Results 3.1. Patient demographics Between March 1995 and May 2008, 716 patients diagnosed with glioblastoma were identified in our database. Those with a previous history of low-grade glioma (LGG) and a secondary transformation to glioblastoma were excluded from the study (n = 42), as were those whose medical records were incomplete or who were lost to follow-up (n = 110). The remaining 564 patients formed the study cohort. Following the matching process, 261 patients remained to form the final study group. The 303 patients who were not included in the analysis were patients older than 70 years, had a KPS < 60, or were patients whom had undergone biopsy only. A total of 137 patients (52.5%, mean age 54.6 ± 11.7 years) were non-trial patients who had received BSC, while the remaining 124 patients were enrolled in at least one clinical trial. Of the clinical trial patients, 81 patients (31.0%, mean age 55.5 ± 9.4 years) had been randomized to the treatment arm and 43 (16.5%, mean age 55.7 ± 9.8 years) to the control arm. Patient characteristics are listed in Table 2 and reflect the homogeneity between the groups for the important prognostic variables. As demonstrated in Table 2, patients were divided into two functional groups based on their KPS score: those able to carry out normal activity (KPS 80–100) and those able to live at home and care for most of their personal needs with varying amount of assistance (KPS 60–70).
3.2. Survival Calculations of overall median and mean survival of the groups indicated that enrollment into a clinical trial carried a major survival advantage for patients with primary glioblastoma regardless of the study arm to which they had been enrolled. The median survival time for the entire group was 13.80 months (95% confidence interval [CI], 12.52–15.08). The median survival time for each of the three groups was 11.57 months (BSC: 95% CI, 10.41–12.73), 16.27 months (control: 95% CI, 14.10–18.43) and 16.10 months (treatment: 95% CI, 14.34–17.86). There was no significant difference in survival between the treatment and control patients (Fig. 1). The Cox regression model compared the study groups adjusting for age, KPS and year of diagnosis. We considered diagnosis from the beginning of 2005 as a covariate, as these patients received concomitant treatment with radiotherapy plus temozolomide (Stupp protocol).2 Including Stupp protocol as a covariate in the Cox model was important to identify a possible bias related to the introduction of the protocol in early 2005 (which is characterized by improved survival for patients with glioblastoma2) and for a higher percentage of patients joining clinical trials after the introduction of the Stupp protocol (Table 3). Multivariate analysis, including the entire study group, showed that participation in a clinical trial, regardless of whether in the treatment (p = 0.0009) or control (p = 0.0095) arm carried a significant survival advantage for patients with primary glioblastoma (Table 4). The hazard ratio (HR) for death in the control group compared to the BSC group was 0.599 (95% CI, 0.406–0.882), and the HR for death in the treatment group compared to the BSC group was 0.596 (95% CI, 0.438–0.809). This represents a 40.1% and 40.4% relative reduction in the risk of death for the control and treatment groups, respectively, compared to the BSC group. Since the Stupp protocol was found to have a favorable impact on survival in our cohort (HR, 0.672), we performed an additional survival analysis comparing the study groups from the time of the introduction of Stupp protocol in 2005, and included only the 96 most recent patients (Table 3). A survival
1532
T. Shahar et al. / Journal of Clinical Neuroscience 19 (2012) 1530–1534
Table 2 Classification of patients based on mode of therapy Best standard of care
Clinical trial Control Arm
Treatment Arm
Total
No. of patients (%) Mean age (years)
137 (52.5%) 54.6 ± 11.7
43 (16.5%) 55.7 ± 9.8
81 (31.0%) 55.5 ± 9.4
261 (100%)
KPS distribution 60–70 80–100
34 (24.8%) 103 (75.2%)
15 (34.9%) 28 (65.1%)
23 (28.4%) 58 (71.6%)
72 (27.6%) 189 (72.4%)
KPS = Karnofsky performance scale score.
Table 3 Distribution of patients in clinical trials before and after the introduction of the Stupp protocol in 2005 Year
Enrolled in clinical trial Treatment
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Fig. 1. Kaplan–Meier survival curves of the three groups showing a survival advantage for primary glioblastoma patients enrolled in a clinical trial, regardless of the study arm, compared to the BSC group. BSC = best standard of care.
advantage was demonstrated again for patients being enrolled in a clinical trial compared to the patients receiving BSC regardless of the arm (p = 0.037). The median survival time for each of the three groups was 11.60 months (BSC: 95% CI, 11.12–17.59), 17.90 months (control: 95% CI, 14.22–23.46) and 17.00 months (treatment: 95% CI, 14.59–20.81). The overall median survival for the entire 96 most recent patients was 16.10 (95% CI, 14.88–19.49). There was no significant difference in survival between the patients receiving treatment and controls (Fig. 2). The HR of 0.65 and 0.45 were calculated for the patients enrolled to the treatment arm and for those enrolled to the control arm compared to the BSC group, respectively. Despite a smaller sample size the effect still reaches statistical significance. Young age and good performance status on admission were also found to have a favorable impact on survival (Table 4). Survival was also analyzed by recursive partitioning. The possible predictors were age, KPS on admission and group (BSC, treatment arm and control arm). The group variable was entered as a ‘‘categorical variable’’. The first two splits separated out ‘‘special’’ age groups, those aged 70 years (terminal node 1) included 11 subjects who had poor performance, and those aged 40 years and under, including 23 subjects, who had good performance (terminal node 2). The remaining 227 subjects (aged 41–69 years) were split by the mode of therapy with the patients enrolled in a clinical trial, regardless of the arm, separated from the BSC patients (terminal node 3). The BSC group was then split by KPS to 85 patients with a KPS of 80–100 (terminal node 4) and to 28 patients with a KPS of 60–70 (terminal node 5). No additional variables were found to affect survival of the trial patients.
BSC
Total
Control
2 0 2 5 3 2 8 3 7 9
0 0 6 7 3 0 0 1 2 4
4 12 11 2 7 8 10 19 10 18
6 12 19 14 13 10 18 23 19 31
Subtotal 2005 2006 2007 2008
41 3 16 18 3
23 0 12 8 0
101 17 8 9 2
165 20 36 35 5
Subtotal Total
40 81
20 43
36 137
96 261
BSC = best standard of care.
These data indicated that, following age, participation in a clinical trial is the most significant prognostic factor for survival. Patients aged 640 years had the best survival advantage, regardless of whether or not they had been enrolled into a clinical trial. For patients aged 41 years to 69 years, a Kaplan–Meier analysis of overall survival revealed a survival advantage for the patients enrolled in a clinical trial, regardless of the arm, compared to the BSC group. The median survival time for patients aged from 41 years to 69 years among the BSC, control and treatment groups was 11.57 months (95% CI, 10.25–12.89), 15.63 months (95% CI, 12.95–18.32) and 15.57 months (95% CI, 14.06–17.07), respectively. 3.3. History of clinical management To identify possible mechanisms that could lead to extended survival in patients enrolled in a clinical trial, we compared the clinical management history of all matched patients. The rates of tumor resection surgeries, non-surgical oncological treatment, outpatient clinic visits and follow-up MRI studies were evaluated. The rate of these covariates represented the number of events per month of a patient’s life. The conversion to rates should eliminate any obvious dependence of these covariates on survival time and so give a ‘‘cleaner’’ covariate for use as a prognostic factor. Patients enrolled in clinical trials tended to be followed-up more closely compared to patients receiving BSC, as demonstrated by higher rates of outpatient clinic visits and follow-up MRI studies. This trend did not, however, reach a level of significance (p = 0.438 for visits and p = 0.192 for follow-up MRI, using the Kruskal–Wallis test).
1533
T. Shahar et al. / Journal of Clinical Neuroscience 19 (2012) 1530–1534 Table 4 The impact on survival of patients with primary glioblastoma of the evaluated clinical prognostic factors Variable
HR
95% CI
p value
Age KPS score Concomitant treatment with radiotherapy plus temozolomide (diagnosis from early 2005 onwards)
1.025 0.707 0.672
1.012–1.037 0.524–0.954 0.502–0.899
0.0002 0.0234 0.0075
0.406–0.882 0.438–0.809
0.0095 0.0009
Mode of therapy BSC Clinical trial Control arm vs. BSC Treatment arm vs. BSC
1 0.599 0.596
BSC = best standard of care, CI = confidence interval, HR = hazard ratio, KPS = Karnofsky performance scale, vs. = versus.
Fig. 2. Kaplan–Meier survival curves of the patients diagnosed since 2005 onwards showing a survival advantage for primary glioblastoma patients enrolled in a clinical trial, regardless of the study arm, compared to the BSC group.
As for non-surgical oncological treatment, patients enrolled into clinical trials were more likely to receive second-line or third-line chemotherapy (procarbazine, lomustine, and vincristine, carboplatin, bevacizumab [Avastin; Genentech/Roche, Nutley, NJ, USA] and polifeprosan 20 with carmustine implant [Gliadel wafers, Eisai, Tokyo, Japan]) or additional radiation therapy compared to patients receiving BSC. This was demonstrated by significantly higher rates of additional non-surgical oncological treatments for the clinical trial group (p = 0.019, Kruskal–Wallis test). For the analysis of the tumor resection surgery rate, the first operation was not included in the computation. This more accurately reflects the rate of additional surgeries since the first operation was the point of entry to the study. The rates of additional tumor resection surgeries among the three groups were significantly different (p = 0.039, Kruskal–Wallis test): 0.038 ± 0.088 for the BSC group, 0.043 ± 0.085 for the control group, and 0.046 ± 0.060 for the treatment group. 4. Discussion We had noted improved outcome in our patients enrolled in clinical trials and conducted this investigation to identify why they did better than our non-trial patients. A recent systematic review from the Cochrane library database by Vist et al. suggested that there is no survival benefit of enrolment in a clinical trial. Their conclusion was that ‘‘Participation in randomized clinical trials is associated with similar outcomes to receiving the same treatment outside randomized clinical trials’’.22 We compared the outcomes of patients with primary glioblastoma who were eligible for, but
did not participate in, a clinical trial and who received clinical interventions similar to those given to patients who were enrolled in clinical trials. Our results demonstrated a significant survival benefit for patients enrolled in clinical trials regardless of whether they had been randomized to the treatment or control arms. This observation persisted following adjustment for patient characteristics in each of the three groups to meet the common criteria required for enrollment in clinical trials. Multivariate analyses also revealed that enrollment in a clinical trial carries a survival advantage for patients with primary glioblastoma with a significant reduction in relative risk of death for this group. This survival advantage was demonstrated regardless of whether the clinical trial was before or after the introduction of the Stupp protocol in early 2005. There are several possible reasons for a survival advantage for patients with glioblastoma who are enrolled in a clinical trial.23– 25 The experimental treatment may offer a superior therapeutic effect for the trial patients compared to those receiving standard therapy.25 This potential advantage may be translated into a survival benefit for the patients randomized to the treatment arm, but our data demonstrated that the trial patients had a survival advantage over the non-trial patients regardless of whether they were randomized to the treatment or control arms. As such, the overall experimental therapeutic effect did not have a major role in improving their outcome. Differences in patients’ baseline characteristics may also influence the survival outcome.26 Clinical trial participants often represent a prognostically favorable subset of patients (younger and with a better KPS on presentation), which may translate into an inherent survival advantage.27,28 We therefore matched the studied groups for age and KPS on presentation and were left with statistically homogeneous and comparable groups (Table 2) and adjusted for these factors in our statistical analysis. It is well documented that certain population subgroups, such as those with low socioeconomic status, are underrepresented in clinical research.29,30 Differences in socioeconomic status are reflected by differences in survival, with patients with a lower socioeconomic status having a poorer prognosis possibly due to ineffective screening, inferior primary health care, lack of health education or differences in proximity to a major health care center,31 all of which may result in delayed presentation and differences in treatment. We were unable to determine our patient’s socioeconomic class from their medical records, but socioeconomic status reportedly has more of a role in less aggressive tumors where treatment can be more effective and the prognosis is better. Differences in survival outcomes of aggressive tumors that are characterized by having a poor prognosis, such as glioblastoma, cannot be attributed to socioeconomic differences.32 Moreover, national health care mandated by law in Israel makes such an explanation unlikely. Finally, and probably of greatest significance, is the participation effect or ‘‘trial effect’’. As Peppercorn et al. had suggested, a participation effect exists when participants in the control group
1534
T. Shahar et al. / Journal of Clinical Neuroscience 19 (2012) 1530–1534
perform better than individuals who were not enrolled in a clinical trial.23 Braunholtz et al. identified the components of the trial effect as including a protocol effect (the result of following detailed and rigid guidelines that are carefully described in clinical trial protocols), a care effect (the extra follow-up and extra nursing care that accompanies clinical trials), the Hawthorne effect (the change in patient–physician interaction in a trial setting due to both being observed) and a placebo effect.25 The trial effect could, however, be beneficial to one and harmful to another. For example, some patients may find the consent procedure to enroll in a clinical trial intimidating and not in their best interest, while others may welcome the opportunity to receive a new form of treatment and extra nursing and physician follow-up. This, in turn, may influence the outcome of the study negatively or positively, respectively. In addition, physicians treating patients enrolled in clinical trials are more likely to adhere to the strict clinical guidelines demanded by the trial protocol. There are several limitations to our study, including the design, the size of the group and the heterogeneity of standard of care and clinical trial treatment protocols. This is a single-institute retrospective analysis with relatively few patients in the final matched group. Although we started with a relatively large number of patients (716), only 564 patients were diagnosed with primary glioblastoma and had complete follow-up records and so were suitable for inclusion. Following the matching process, we were left with 261 patients that formed the final matched group. There were changes in the standard care of glioblastoma patients during the study period and, by the nature of clinical studies that use different treatments and different protocols, there is an innate heterogeneity of the studied cohort. Although we addressed these issues by using multivariate analyses and by taking into account the most significant modification in glioblastoma patient’s treatment since the early 2000s, the Stupp protocol, we feel that a shorter study period with a more homogeneous patient population and management protocols would have allowed us to produce ‘‘cleaner’’ results. In conclusion, identifying the improved outcome of patients enrolled in clinical trials and the likelihood of the presence of a ‘‘trial effect’’ raises practical and ethical issues related to the quality of care given to patients with glioblastoma who receive BSC, which may not necessarily be the best care possible. The survival advantages for patients enrolled in clinical trials may be due to more aggressive follow-up and oncological treatment. The findings of this study also highlight the need to avoid using historical controls as a comparative arm in clinical studies and suggest that a prospective multicenter study is needed for further validation of these results.
Conflict of Interest/Disclosures The authors declare that they have no financial or other conflicts of interest in relation to this research and its publication. Acknowledgment We thank Esther Eshkol, MA, the institutional medical and scientific copyeditor, for editing the manuscript. References 1. DeAngelis LM. Brain tumors. N Engl J Med 2001;344:114–23. 2. Stupp R, Mason WP, van den Bent MJ, et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med 2005;352:987–96. 3. Gilbert H, Kagan AR, Cassidy F, et al. Glioblastoma multiforme is not a uniform disease! Cancer Clin Trials 1981;4:87–9.
4. Curran Jr WJ, Scott CB, Horton J, et al. Recursive partitioning analysis of prognostic factors in three Radiation Therapy Oncology Group malignant glioma trials. J Natl Cancer Inst 1993;85:704–10. 5. Homma T, Fukushima T, Vaccarella S, et al. Correlation among pathology, genotype, and patient outcomes in glioblastoma. J Neuropathol Exp Neurol 2006;65:846–54. 6. Weller M, Felsberg J, Hartmann C, et al. Molecular predictors of progressionfree and overall survival in patients with newly diagnosed glioblastoma: a prospective translational study of the German Glioma Network. J Clin Oncol 2009;27:5743–50. 7. Mineo JF, Quintin-Roue I, Lucas B, et al. Glioblastomas: clinical study and search for prognostic factors. Neurochirurgie 2002;48:500–9. 8. Lamborn KR, Chang SM, Prados MD. Prognostic factors for survival of patients with glioblastoma: recursive partitioning analysis. Neuro-oncol 2004;6:227–35. 9. Sanai N, Berger MS. Glioma extent of resection and its impact on patient outcome. Neurosurgery 2008;62:753–64 discussion 264–6. 10. Mineo JF, Bordron A, Baroncini M, et al. Prognosis factors of survival time in patients with glioblastoma multiforme: a multivariate analysis of 340 patients. Acta Neurochir (Wien) 2007;149:245–52 discussion 52–3. 11. Rainov NG. A phase III clinical evaluation of herpes simplex virus type 1 thymidine kinase and ganciclovir gene therapy as an adjuvant to surgical resection and radiation in adults with previously untreated glioblastoma multiforme. Hum Gene Ther 2000;11:2389–401. 12. Kunwar S, Prados MD, Chang SM, et al. Direct intracerebral delivery of cintredekin besudotox (IL13-PE38QQR) in recurrent malignant glioma: a report by the Cintredekin Besudotox Intraparenchymal Study Group. J Clin Oncol 2007;25:837–44. 13. Raty JK, Pikkarainen JT, Wirth T, et al. Gene therapy: the first approved genebased medicines, molecular mechanisms and clinical indications. Curr Mol Pharmacol 2008;1:13–23. 14. Westphal M, Hilt DC, Bortey E, et al. A phase 3 trial of local chemotherapy with biodegradable carmustine (BCNU) wafers (Gliadel wafers) in patients with primary malignant glioma. Neuro-oncol 2003;5:79–88. 15. Lidar Z, Mardor Y, Jonas T, et al. Convection-enhanced delivery of paclitaxel for the treatment of recurrent malignant glioma: a phase I/II clinical study. J Neurosurg 2004;100:472–9. 16. Comparison of TransMID vs standard treatment of cancerous brain tumors; 2004. Available from: http://clinicaltrials.gov/ct2/show/NCT00088400. 17. Kirson ED, Schneiderman RS, Dbaly V, et al. Chemotherapeutic treatment efficacy and sensitivity are increased by adjuvant alternating electric fields (TTFields). BMC Med Phys 2009;9:1. 18. RTOGÒ0525. Phase III trial comparing conventional adjuvant temozolomide with dose intensive temozolomide in patients with newly diagnosed glioblastoma. The Radiation Therapy Oncology GroupÒ. Available from: http://www.rtog.org/ClinicalTrials/ProtocolTable/StudyDetails.aspx?study=0525. 19. RTOGÒ0625. A randomized phase II trial of bevacizumab with irinotecan or bevacizumab with temozolomide in recurrent glioblastoma. The Radiation Therapy Oncology GroupÒ. Available from: http://www.rtog.org/ClinicalTrials/ ProtocolTable/StudyDetails.aspx?study=0625. 20. R Development Core Team. R: a language and environment for statistical computing program. Vienna, Austria; 2009. 21. Therneau TM, Atkinson EJ. An introduction to recursive partitioning using the RPART Routines Technical Report Series No. 61. Rochester, MN: Section of Biostatistics, Mayo Clinic; 1997. 22. Vist GE, Bryant D, Somerville L, et al. Outcomes of patients who participate in randomized controlled trials compared to similar patients receiving similar interventions who do not participate. Cochrane Database Syst Rev 2008;3:MR000009. 23. Peppercorn JM, Weeks JC, Cook EF, et al. Comparison of outcomes in cancer patients treated within and outside clinical trials: conceptual framework and structured review. Lancet 2004;363:263–70. 24. ECRI. Should I enter a clinical trial? A patient reference guide for adults with a serious or life-threatening illness. A Report by ECRI Commissioned by AAHP; 2002. 25. Braunholtz DA, Edwards SJ, Lilford RJ. Are randomized clinical trials good for us (in the short term)? Evidence for a ‘‘trial effect’’. J Clin Epidemiol 2001;54:217–24. 26. Stiller CA. Centralised treatment, entry to trials and survival. Br J Cancer 1994;70:352–62. 27. Antman K, Amato D, Wood W, et al. Selection bias in clinical trials. J Clin Oncol 1985;3:1142–7. 28. Rahman ZU, Frye DK, Buzdar AU, et al. Impact of selection process on response rate and long-term survival of potential high-dose chemotherapy candidates treated with standard-dose doxorubicin-containing chemotherapy in patients with metastatic breast cancer. J Clin Oncol 1997;15:3171–7. 29. Murthy VH, Krumholz HM, Gross CP. Participation in cancer clinical trials: race, sex-, and age-based disparities. JAMA 2004;291:2720–6. 30. Sateren WB, Trimble EL, Abrams J, et al. How sociodemographics, presence of oncology specialists, and hospital cancer programs affect accrual to cancer treatment trials. J Clin Oncol 2002;20:2109–17. 31. Hillner BE, Smith TJ, Desch CE. Hospital and physician volume or specialization and outcomes in cancer treatment: importance in quality of cancer care. J Clin Oncol 2000;18:2327–40. 32. Kogevinas M, Marmot MG, Fox AJ, et al. Socioeconomic differences in cancer survival. J Epidemiol Community Health 1991;45:216–9.
Contents lists available at SciVerse ScienceDirect
Journal of Clinical Neuroscience journal homepage: www.elsevier.com/locate/jocn
Clinical Study
The impact of enrollment in clinical trials on survival of patients with glioblastoma Tal Shahar a, Erez Nossek a, David M. Steinberg b, Uri Rozovski c, Deborah T. Blumenthal d, Felix Bokstein d, Razi Sitt a, Sigal Freedman a, Benjamin W. Corn e, Andrew A. Kanner a, Zvi Ram a,⇑ a
Department of Neurosurgery, Tel Aviv Medical Center, 6 Weizman Street, Tel Aviv 64239, Israel Department of Statistics and Operations Research, Raymond and Beverly Sackler Faculty of Exact Sciences, Tel Aviv University, Tel Aviv, Israel c Department of Hematology and Bone Marrow Transplantation, Tel Aviv Medical Center, Tel Aviv, Israel d Oncology Division, Tel Aviv Medical Center, Tel Aviv, Israel e Institute of Radiotherapy, Tel Aviv Medical Center, Tel Aviv, Israel b
a r t i c l e
i n f o
Article history: Received 31 March 2012 Accepted 4 April 2012
Keywords: Clinical trials Glioblastoma Neuro-oncology Neurosurgery Survival analysis
a b s t r a c t The impact of enrollment in a clinical study on the survival of patients with glioblastoma has not been established. We retrospectively analyzed 564 patients with newly diagnosed glioblastoma treated between 1995 and 2008. They were divided into those enrolled in a clinical trial and randomized to a treatment or control arm, and those not enrolled and who received best standard of care (BSC). The three groups were matched for age and Karnofsky performance scale (KPS) score at presentation, and included only patients who underwent at least one tumor resection. Survival analysis was performed and multivariate Cox proportional hazards model and recursive partitioning analysis (RPA) identified predictors of survival. Following the matching process, 261 patients remained to form the final cohort. Of the 124 patients enrolled in a study, 81 (31.0%) were randomized to the treatment and 43 (16.5%) to the control arms. The overall median survival for the BSC (n = 137), control, and treatment groups was 11.57 months (95% confidence interval [CI], 10.41–12.73), 16.27 months (95% CI, 14.10–18.43) and 16.10 months (95% CI, 14.34–17.86), respectively (p = 0.002). Participation in a clinical trial, regardless of the arm, was a significant predictor of survival, as were age and KPS at diagnosis. The RPA also demonstrated a favorable impact of participation in a clinical trial. Additional tumor resections and various treatment modalities were administered with significantly higher frequency among patients enrolled in clinical studies. Thus, enrollment in a clinical study carried a significant survival advantage for patients with glioblastoma, raising practical and ethical issues regarding the quality of care of patients who receive ‘‘standard’’ therapy. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction Glioblastoma is the most common primary malignant brain tumor in adults. Despite multimodality treatment, the prognosis remains poor, with a median survival of approximately 1 year following diagnosis.1,2 Prognosis of glioblastoma patients is not uniform.3 A wide range of prognostic factors and their relative importance in influencing outcome have been studied by multivariate analyses, with age and Karnofsky performance scale (KPS) score on admission constituting the most significant predictors of survival for patients with glioblastoma.4 Histological characteristics,5,6 de novo glioblastoma or secondary transformation7 and tumor location8 also affect survival. Current treatment strategies for glioblastoma include surgery, radiation therapy and chemotherapy.2 Patients who have undergone gross total tumor resection have a survival advantage compared to those who underwent partial resection or biopsy only.9 ⇑ Corresponding author. Tel.: +972 3 697 4273; fax: +972 3 697 4680. E-mail address: [email protected] (Z. Ram). 0967-5868/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jocn.2012.04.005
Response to radiation therapy and chemotherapy were also predictive factors for survival.2,4,10 Recent advances in understanding the molecular events that take part in the pathogenesis of glioma have led to investigation of many new therapeutic options in clinical trials. A search of our brain tumor database revealed 10 clinical trials11–19 that had been conducted in patients diagnosed with glioblastoma between March 1995 and May 2008 (Table 1). Outcome analyses of such studies are based on comparison of the treatment group to either a control group receiving ‘‘best standard of care’’ (BSC) therapies or, in fewer cases, comparison to historical controls. The actual impact of being enrolled in a clinical trial on survival has not been fully investigated. 2. Methods 2.1. Study design The current study is a retrospective analysis of prospectively collected data from the Tel Aviv Medical Center (TAMC) brain tumor database. Approval from the TAMC institutional review board
T. Shahar et al. / Journal of Clinical Neuroscience 19 (2012) 1530–1534 Table 1 Clinical trials conducted in patients diagnosed with glioblastoma between March 1995 and May 2008 Clinical trialref
No. patients enrolled
Cerepro13 Gliadel14 GTI 10711 GTI 11511 IL1312a Novocure17 RTOG 052531 RTOG 062532 Taxol15 Transmid16
31 14 2 10 26 14 18 6 9 10
Note: the total number of patients in the table is greater than the number in our group of patients who were enrolled into clinical trials due to the fact that some were enrolled into more than one clinical trial. a IL13 includes three separate clinical trial studies.
(Number: 003108) was obtained. We evaluated data from all patients with de novo glioblastoma who were treated in our institution between 1995 and 2008. The evaluated variables were age and KPS on admission, number of tumor resection surgeries, histopathological evaluation of the tumors, date of diagnosis (determined as the date of the first tissue sampling), date of death, calculated survival period (from diagnosis until death) and participation in a clinical trial. The neurosurgical procedures and all nonsurgical treatment modalities were also recorded. The patients were divided into three groups according to treatment modalities: those who were not enrolled in a study and had received BSC, and those enrolled in at least one clinical trial and randomized to either the control or experimental treatment arms. The non-trial patients (BSC) received therapy considered ‘‘standard’’ at the time of their treatment. From 1995 until the end of 2004, standard therapy for newly diagnosed glioblastoma in Israel was tumor resection surgery, followed by radiotherapy. As of the beginning of 2005, standard of care for glioblastoma patients in our institute was tumor resection surgery, followed by concomitant treatment with radiotherapy plus temozolomide (Stupp protocol).2 In order to overcome the possible selection bias used in the process of recruiting patients into clinical trials, we have selected patients who had met, or would have met, the standard inclusion criteria for the various clinical trials. We matched the groups (BSC, treatment and controls) for age and KPS at presentation and included only patients who had undergone at least one tumor resection surgery. Thus, patients with a KPS of <60, older than 70 years of age and who were not considered candidates for tumor resection surgeries (that is, biopsy only) were excluded. Survival analysis was calculated for each of the groups, and multivariable analyses were performed to identify independent clinical prognostic factors for survival. 2.2. Measurement of study parameters Clinical and therapeutic factors were analyzed using the univariate Student’s t-test or chi-squared (v2) test for determining their prognostic impact on overall survival. Time to event distribution (for example, survival) was assessed by the Kaplan–Meier survival curves and compared using the Log Rank (Mantel–Cox) test. The importance of predictor variables was assessed using Cox proportional hazards regressions. The analysis was done with the R statistical environment and the survival library by Therneau.20 Survival experience was also analyzed by recursive partitioning analysis (RPA) using the R statistical environment and the Recursive PARTitioning (RPART) library by Therneau and Atkinson.21 Patients who were alive on the last day of follow-up were considered to have censored survival times.
1531
3. Results 3.1. Patient demographics Between March 1995 and May 2008, 716 patients diagnosed with glioblastoma were identified in our database. Those with a previous history of low-grade glioma (LGG) and a secondary transformation to glioblastoma were excluded from the study (n = 42), as were those whose medical records were incomplete or who were lost to follow-up (n = 110). The remaining 564 patients formed the study cohort. Following the matching process, 261 patients remained to form the final study group. The 303 patients who were not included in the analysis were patients older than 70 years, had a KPS < 60, or were patients whom had undergone biopsy only. A total of 137 patients (52.5%, mean age 54.6 ± 11.7 years) were non-trial patients who had received BSC, while the remaining 124 patients were enrolled in at least one clinical trial. Of the clinical trial patients, 81 patients (31.0%, mean age 55.5 ± 9.4 years) had been randomized to the treatment arm and 43 (16.5%, mean age 55.7 ± 9.8 years) to the control arm. Patient characteristics are listed in Table 2 and reflect the homogeneity between the groups for the important prognostic variables. As demonstrated in Table 2, patients were divided into two functional groups based on their KPS score: those able to carry out normal activity (KPS 80–100) and those able to live at home and care for most of their personal needs with varying amount of assistance (KPS 60–70).
3.2. Survival Calculations of overall median and mean survival of the groups indicated that enrollment into a clinical trial carried a major survival advantage for patients with primary glioblastoma regardless of the study arm to which they had been enrolled. The median survival time for the entire group was 13.80 months (95% confidence interval [CI], 12.52–15.08). The median survival time for each of the three groups was 11.57 months (BSC: 95% CI, 10.41–12.73), 16.27 months (control: 95% CI, 14.10–18.43) and 16.10 months (treatment: 95% CI, 14.34–17.86). There was no significant difference in survival between the treatment and control patients (Fig. 1). The Cox regression model compared the study groups adjusting for age, KPS and year of diagnosis. We considered diagnosis from the beginning of 2005 as a covariate, as these patients received concomitant treatment with radiotherapy plus temozolomide (Stupp protocol).2 Including Stupp protocol as a covariate in the Cox model was important to identify a possible bias related to the introduction of the protocol in early 2005 (which is characterized by improved survival for patients with glioblastoma2) and for a higher percentage of patients joining clinical trials after the introduction of the Stupp protocol (Table 3). Multivariate analysis, including the entire study group, showed that participation in a clinical trial, regardless of whether in the treatment (p = 0.0009) or control (p = 0.0095) arm carried a significant survival advantage for patients with primary glioblastoma (Table 4). The hazard ratio (HR) for death in the control group compared to the BSC group was 0.599 (95% CI, 0.406–0.882), and the HR for death in the treatment group compared to the BSC group was 0.596 (95% CI, 0.438–0.809). This represents a 40.1% and 40.4% relative reduction in the risk of death for the control and treatment groups, respectively, compared to the BSC group. Since the Stupp protocol was found to have a favorable impact on survival in our cohort (HR, 0.672), we performed an additional survival analysis comparing the study groups from the time of the introduction of Stupp protocol in 2005, and included only the 96 most recent patients (Table 3). A survival
1532
T. Shahar et al. / Journal of Clinical Neuroscience 19 (2012) 1530–1534
Table 2 Classification of patients based on mode of therapy Best standard of care
Clinical trial Control Arm
Treatment Arm
Total
No. of patients (%) Mean age (years)
137 (52.5%) 54.6 ± 11.7
43 (16.5%) 55.7 ± 9.8
81 (31.0%) 55.5 ± 9.4
261 (100%)
KPS distribution 60–70 80–100
34 (24.8%) 103 (75.2%)
15 (34.9%) 28 (65.1%)
23 (28.4%) 58 (71.6%)
72 (27.6%) 189 (72.4%)
KPS = Karnofsky performance scale score.
Table 3 Distribution of patients in clinical trials before and after the introduction of the Stupp protocol in 2005 Year
Enrolled in clinical trial Treatment
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Fig. 1. Kaplan–Meier survival curves of the three groups showing a survival advantage for primary glioblastoma patients enrolled in a clinical trial, regardless of the study arm, compared to the BSC group. BSC = best standard of care.
advantage was demonstrated again for patients being enrolled in a clinical trial compared to the patients receiving BSC regardless of the arm (p = 0.037). The median survival time for each of the three groups was 11.60 months (BSC: 95% CI, 11.12–17.59), 17.90 months (control: 95% CI, 14.22–23.46) and 17.00 months (treatment: 95% CI, 14.59–20.81). The overall median survival for the entire 96 most recent patients was 16.10 (95% CI, 14.88–19.49). There was no significant difference in survival between the patients receiving treatment and controls (Fig. 2). The HR of 0.65 and 0.45 were calculated for the patients enrolled to the treatment arm and for those enrolled to the control arm compared to the BSC group, respectively. Despite a smaller sample size the effect still reaches statistical significance. Young age and good performance status on admission were also found to have a favorable impact on survival (Table 4). Survival was also analyzed by recursive partitioning. The possible predictors were age, KPS on admission and group (BSC, treatment arm and control arm). The group variable was entered as a ‘‘categorical variable’’. The first two splits separated out ‘‘special’’ age groups, those aged 70 years (terminal node 1) included 11 subjects who had poor performance, and those aged 40 years and under, including 23 subjects, who had good performance (terminal node 2). The remaining 227 subjects (aged 41–69 years) were split by the mode of therapy with the patients enrolled in a clinical trial, regardless of the arm, separated from the BSC patients (terminal node 3). The BSC group was then split by KPS to 85 patients with a KPS of 80–100 (terminal node 4) and to 28 patients with a KPS of 60–70 (terminal node 5). No additional variables were found to affect survival of the trial patients.
BSC
Total
Control
2 0 2 5 3 2 8 3 7 9
0 0 6 7 3 0 0 1 2 4
4 12 11 2 7 8 10 19 10 18
6 12 19 14 13 10 18 23 19 31
Subtotal 2005 2006 2007 2008
41 3 16 18 3
23 0 12 8 0
101 17 8 9 2
165 20 36 35 5
Subtotal Total
40 81
20 43
36 137
96 261
BSC = best standard of care.
These data indicated that, following age, participation in a clinical trial is the most significant prognostic factor for survival. Patients aged 640 years had the best survival advantage, regardless of whether or not they had been enrolled into a clinical trial. For patients aged 41 years to 69 years, a Kaplan–Meier analysis of overall survival revealed a survival advantage for the patients enrolled in a clinical trial, regardless of the arm, compared to the BSC group. The median survival time for patients aged from 41 years to 69 years among the BSC, control and treatment groups was 11.57 months (95% CI, 10.25–12.89), 15.63 months (95% CI, 12.95–18.32) and 15.57 months (95% CI, 14.06–17.07), respectively. 3.3. History of clinical management To identify possible mechanisms that could lead to extended survival in patients enrolled in a clinical trial, we compared the clinical management history of all matched patients. The rates of tumor resection surgeries, non-surgical oncological treatment, outpatient clinic visits and follow-up MRI studies were evaluated. The rate of these covariates represented the number of events per month of a patient’s life. The conversion to rates should eliminate any obvious dependence of these covariates on survival time and so give a ‘‘cleaner’’ covariate for use as a prognostic factor. Patients enrolled in clinical trials tended to be followed-up more closely compared to patients receiving BSC, as demonstrated by higher rates of outpatient clinic visits and follow-up MRI studies. This trend did not, however, reach a level of significance (p = 0.438 for visits and p = 0.192 for follow-up MRI, using the Kruskal–Wallis test).
1533
T. Shahar et al. / Journal of Clinical Neuroscience 19 (2012) 1530–1534 Table 4 The impact on survival of patients with primary glioblastoma of the evaluated clinical prognostic factors Variable
HR
95% CI
p value
Age KPS score Concomitant treatment with radiotherapy plus temozolomide (diagnosis from early 2005 onwards)
1.025 0.707 0.672
1.012–1.037 0.524–0.954 0.502–0.899
0.0002 0.0234 0.0075
0.406–0.882 0.438–0.809
0.0095 0.0009
Mode of therapy BSC Clinical trial Control arm vs. BSC Treatment arm vs. BSC
1 0.599 0.596
BSC = best standard of care, CI = confidence interval, HR = hazard ratio, KPS = Karnofsky performance scale, vs. = versus.
Fig. 2. Kaplan–Meier survival curves of the patients diagnosed since 2005 onwards showing a survival advantage for primary glioblastoma patients enrolled in a clinical trial, regardless of the study arm, compared to the BSC group.
As for non-surgical oncological treatment, patients enrolled into clinical trials were more likely to receive second-line or third-line chemotherapy (procarbazine, lomustine, and vincristine, carboplatin, bevacizumab [Avastin; Genentech/Roche, Nutley, NJ, USA] and polifeprosan 20 with carmustine implant [Gliadel wafers, Eisai, Tokyo, Japan]) or additional radiation therapy compared to patients receiving BSC. This was demonstrated by significantly higher rates of additional non-surgical oncological treatments for the clinical trial group (p = 0.019, Kruskal–Wallis test). For the analysis of the tumor resection surgery rate, the first operation was not included in the computation. This more accurately reflects the rate of additional surgeries since the first operation was the point of entry to the study. The rates of additional tumor resection surgeries among the three groups were significantly different (p = 0.039, Kruskal–Wallis test): 0.038 ± 0.088 for the BSC group, 0.043 ± 0.085 for the control group, and 0.046 ± 0.060 for the treatment group. 4. Discussion We had noted improved outcome in our patients enrolled in clinical trials and conducted this investigation to identify why they did better than our non-trial patients. A recent systematic review from the Cochrane library database by Vist et al. suggested that there is no survival benefit of enrolment in a clinical trial. Their conclusion was that ‘‘Participation in randomized clinical trials is associated with similar outcomes to receiving the same treatment outside randomized clinical trials’’.22 We compared the outcomes of patients with primary glioblastoma who were eligible for, but
did not participate in, a clinical trial and who received clinical interventions similar to those given to patients who were enrolled in clinical trials. Our results demonstrated a significant survival benefit for patients enrolled in clinical trials regardless of whether they had been randomized to the treatment or control arms. This observation persisted following adjustment for patient characteristics in each of the three groups to meet the common criteria required for enrollment in clinical trials. Multivariate analyses also revealed that enrollment in a clinical trial carries a survival advantage for patients with primary glioblastoma with a significant reduction in relative risk of death for this group. This survival advantage was demonstrated regardless of whether the clinical trial was before or after the introduction of the Stupp protocol in early 2005. There are several possible reasons for a survival advantage for patients with glioblastoma who are enrolled in a clinical trial.23– 25 The experimental treatment may offer a superior therapeutic effect for the trial patients compared to those receiving standard therapy.25 This potential advantage may be translated into a survival benefit for the patients randomized to the treatment arm, but our data demonstrated that the trial patients had a survival advantage over the non-trial patients regardless of whether they were randomized to the treatment or control arms. As such, the overall experimental therapeutic effect did not have a major role in improving their outcome. Differences in patients’ baseline characteristics may also influence the survival outcome.26 Clinical trial participants often represent a prognostically favorable subset of patients (younger and with a better KPS on presentation), which may translate into an inherent survival advantage.27,28 We therefore matched the studied groups for age and KPS on presentation and were left with statistically homogeneous and comparable groups (Table 2) and adjusted for these factors in our statistical analysis. It is well documented that certain population subgroups, such as those with low socioeconomic status, are underrepresented in clinical research.29,30 Differences in socioeconomic status are reflected by differences in survival, with patients with a lower socioeconomic status having a poorer prognosis possibly due to ineffective screening, inferior primary health care, lack of health education or differences in proximity to a major health care center,31 all of which may result in delayed presentation and differences in treatment. We were unable to determine our patient’s socioeconomic class from their medical records, but socioeconomic status reportedly has more of a role in less aggressive tumors where treatment can be more effective and the prognosis is better. Differences in survival outcomes of aggressive tumors that are characterized by having a poor prognosis, such as glioblastoma, cannot be attributed to socioeconomic differences.32 Moreover, national health care mandated by law in Israel makes such an explanation unlikely. Finally, and probably of greatest significance, is the participation effect or ‘‘trial effect’’. As Peppercorn et al. had suggested, a participation effect exists when participants in the control group
1534
T. Shahar et al. / Journal of Clinical Neuroscience 19 (2012) 1530–1534
perform better than individuals who were not enrolled in a clinical trial.23 Braunholtz et al. identified the components of the trial effect as including a protocol effect (the result of following detailed and rigid guidelines that are carefully described in clinical trial protocols), a care effect (the extra follow-up and extra nursing care that accompanies clinical trials), the Hawthorne effect (the change in patient–physician interaction in a trial setting due to both being observed) and a placebo effect.25 The trial effect could, however, be beneficial to one and harmful to another. For example, some patients may find the consent procedure to enroll in a clinical trial intimidating and not in their best interest, while others may welcome the opportunity to receive a new form of treatment and extra nursing and physician follow-up. This, in turn, may influence the outcome of the study negatively or positively, respectively. In addition, physicians treating patients enrolled in clinical trials are more likely to adhere to the strict clinical guidelines demanded by the trial protocol. There are several limitations to our study, including the design, the size of the group and the heterogeneity of standard of care and clinical trial treatment protocols. This is a single-institute retrospective analysis with relatively few patients in the final matched group. Although we started with a relatively large number of patients (716), only 564 patients were diagnosed with primary glioblastoma and had complete follow-up records and so were suitable for inclusion. Following the matching process, we were left with 261 patients that formed the final matched group. There were changes in the standard care of glioblastoma patients during the study period and, by the nature of clinical studies that use different treatments and different protocols, there is an innate heterogeneity of the studied cohort. Although we addressed these issues by using multivariate analyses and by taking into account the most significant modification in glioblastoma patient’s treatment since the early 2000s, the Stupp protocol, we feel that a shorter study period with a more homogeneous patient population and management protocols would have allowed us to produce ‘‘cleaner’’ results. In conclusion, identifying the improved outcome of patients enrolled in clinical trials and the likelihood of the presence of a ‘‘trial effect’’ raises practical and ethical issues related to the quality of care given to patients with glioblastoma who receive BSC, which may not necessarily be the best care possible. The survival advantages for patients enrolled in clinical trials may be due to more aggressive follow-up and oncological treatment. The findings of this study also highlight the need to avoid using historical controls as a comparative arm in clinical studies and suggest that a prospective multicenter study is needed for further validation of these results.
Conflict of Interest/Disclosures The authors declare that they have no financial or other conflicts of interest in relation to this research and its publication. Acknowledgment We thank Esther Eshkol, MA, the institutional medical and scientific copyeditor, for editing the manuscript. References 1. DeAngelis LM. Brain tumors. N Engl J Med 2001;344:114–23. 2. Stupp R, Mason WP, van den Bent MJ, et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med 2005;352:987–96. 3. Gilbert H, Kagan AR, Cassidy F, et al. Glioblastoma multiforme is not a uniform disease! Cancer Clin Trials 1981;4:87–9.
4. Curran Jr WJ, Scott CB, Horton J, et al. Recursive partitioning analysis of prognostic factors in three Radiation Therapy Oncology Group malignant glioma trials. J Natl Cancer Inst 1993;85:704–10. 5. Homma T, Fukushima T, Vaccarella S, et al. Correlation among pathology, genotype, and patient outcomes in glioblastoma. J Neuropathol Exp Neurol 2006;65:846–54. 6. Weller M, Felsberg J, Hartmann C, et al. Molecular predictors of progressionfree and overall survival in patients with newly diagnosed glioblastoma: a prospective translational study of the German Glioma Network. J Clin Oncol 2009;27:5743–50. 7. Mineo JF, Quintin-Roue I, Lucas B, et al. Glioblastomas: clinical study and search for prognostic factors. Neurochirurgie 2002;48:500–9. 8. Lamborn KR, Chang SM, Prados MD. Prognostic factors for survival of patients with glioblastoma: recursive partitioning analysis. Neuro-oncol 2004;6:227–35. 9. Sanai N, Berger MS. Glioma extent of resection and its impact on patient outcome. Neurosurgery 2008;62:753–64 discussion 264–6. 10. Mineo JF, Bordron A, Baroncini M, et al. Prognosis factors of survival time in patients with glioblastoma multiforme: a multivariate analysis of 340 patients. Acta Neurochir (Wien) 2007;149:245–52 discussion 52–3. 11. Rainov NG. A phase III clinical evaluation of herpes simplex virus type 1 thymidine kinase and ganciclovir gene therapy as an adjuvant to surgical resection and radiation in adults with previously untreated glioblastoma multiforme. Hum Gene Ther 2000;11:2389–401. 12. Kunwar S, Prados MD, Chang SM, et al. Direct intracerebral delivery of cintredekin besudotox (IL13-PE38QQR) in recurrent malignant glioma: a report by the Cintredekin Besudotox Intraparenchymal Study Group. J Clin Oncol 2007;25:837–44. 13. Raty JK, Pikkarainen JT, Wirth T, et al. Gene therapy: the first approved genebased medicines, molecular mechanisms and clinical indications. Curr Mol Pharmacol 2008;1:13–23. 14. Westphal M, Hilt DC, Bortey E, et al. A phase 3 trial of local chemotherapy with biodegradable carmustine (BCNU) wafers (Gliadel wafers) in patients with primary malignant glioma. Neuro-oncol 2003;5:79–88. 15. Lidar Z, Mardor Y, Jonas T, et al. Convection-enhanced delivery of paclitaxel for the treatment of recurrent malignant glioma: a phase I/II clinical study. J Neurosurg 2004;100:472–9. 16. Comparison of TransMID vs standard treatment of cancerous brain tumors; 2004. Available from: http://clinicaltrials.gov/ct2/show/NCT00088400. 17. Kirson ED, Schneiderman RS, Dbaly V, et al. Chemotherapeutic treatment efficacy and sensitivity are increased by adjuvant alternating electric fields (TTFields). BMC Med Phys 2009;9:1. 18. RTOGÒ0525. Phase III trial comparing conventional adjuvant temozolomide with dose intensive temozolomide in patients with newly diagnosed glioblastoma. The Radiation Therapy Oncology GroupÒ. Available from: http://www.rtog.org/ClinicalTrials/ProtocolTable/StudyDetails.aspx?study=0525. 19. RTOGÒ0625. A randomized phase II trial of bevacizumab with irinotecan or bevacizumab with temozolomide in recurrent glioblastoma. The Radiation Therapy Oncology GroupÒ. Available from: http://www.rtog.org/ClinicalTrials/ ProtocolTable/StudyDetails.aspx?study=0625. 20. R Development Core Team. R: a language and environment for statistical computing program. Vienna, Austria; 2009. 21. Therneau TM, Atkinson EJ. An introduction to recursive partitioning using the RPART Routines Technical Report Series No. 61. Rochester, MN: Section of Biostatistics, Mayo Clinic; 1997. 22. Vist GE, Bryant D, Somerville L, et al. Outcomes of patients who participate in randomized controlled trials compared to similar patients receiving similar interventions who do not participate. Cochrane Database Syst Rev 2008;3:MR000009. 23. Peppercorn JM, Weeks JC, Cook EF, et al. Comparison of outcomes in cancer patients treated within and outside clinical trials: conceptual framework and structured review. Lancet 2004;363:263–70. 24. ECRI. Should I enter a clinical trial? A patient reference guide for adults with a serious or life-threatening illness. A Report by ECRI Commissioned by AAHP; 2002. 25. Braunholtz DA, Edwards SJ, Lilford RJ. Are randomized clinical trials good for us (in the short term)? Evidence for a ‘‘trial effect’’. J Clin Epidemiol 2001;54:217–24. 26. Stiller CA. Centralised treatment, entry to trials and survival. Br J Cancer 1994;70:352–62. 27. Antman K, Amato D, Wood W, et al. Selection bias in clinical trials. J Clin Oncol 1985;3:1142–7. 28. Rahman ZU, Frye DK, Buzdar AU, et al. Impact of selection process on response rate and long-term survival of potential high-dose chemotherapy candidates treated with standard-dose doxorubicin-containing chemotherapy in patients with metastatic breast cancer. J Clin Oncol 1997;15:3171–7. 29. Murthy VH, Krumholz HM, Gross CP. Participation in cancer clinical trials: race, sex-, and age-based disparities. JAMA 2004;291:2720–6. 30. Sateren WB, Trimble EL, Abrams J, et al. How sociodemographics, presence of oncology specialists, and hospital cancer programs affect accrual to cancer treatment trials. J Clin Oncol 2002;20:2109–17. 31. Hillner BE, Smith TJ, Desch CE. Hospital and physician volume or specialization and outcomes in cancer treatment: importance in quality of cancer care. J Clin Oncol 2000;18:2327–40. 32. Kogevinas M, Marmot MG, Fox AJ, et al. Socioeconomic differences in cancer survival. J Epidemiol Community Health 1991;45:216–9.