J Neurooncol DOI 10.1007/s11060-007-9426-z
CLINICAL PATIENT STUDIES
Delay in radiotherapy shortens survival in patients with high grade glioma Chris Irwin Æ Martin Hunn Æ Gordon Purdie Æ David Hamilton
Received: 6 March 2007 / Accepted: 29 May 2007 Springer Science+Business Media B.V. 2007
Abstract Fractionated external beam radiotherapy is an important component of standard treatment for high grade glioma. Due to resource constraints, patients may experience delays in receiving treatment. The purpose of this study was to evaluate the effect of radiotherapy waiting time on survival in patients with high grade glioma. A retrospective analysis was performed of 172 patients with a histological diagnosis of WHO Grade 3 or 4 Astrocytoma who had undergone surgery at Wellington Hospital between 1993 and 2003, and who subsequently underwent radiotherapy. Time to radiotherapy after surgery varied from 7 days to over 16 weeks. Multiple Cox regression analysis showed that age, performance status, tumour grade, extent of surgical resection, radiotherapy dose, and time to radiotherapy from day of surgery were all independently related to survival. Every additional week of delay until the start of radiotherapy increases the risk of death (hazard ratio) by 8.9% (95%CI 2.0%–16.1%). A 6 week delay in starting radiotherapy (from 2 weeks postop to 8 weeks) reduces median survival by 11 weeks for a typical patient. Delay in radiotherapy results in a clinically
This study was approved by the Central Regional Ethics Committee, New Zealand. C. Irwin M. Hunn (&) Department of Neurosurgery, Wellington Hospital, Private Bag 7902, Wellington South, New Zealand e-mail:
[email protected] G. Purdie Department of Public Health, Wellington School of Medicine and Health Sciences, Wellington, New Zealand D. Hamilton Wellington Cancer Centre, Wellington Hospital, Wellington, New Zealand
significant reduction in survival. These findings have implications for resource allocation and for the design of clinical trials. Keywords Glioblastoma multiforme High grade glioma Radiotherapy Survival Treatment delay
Introduction Glioblastoma multiforme (GBM) is the most common primary brain tumour in adults. Until recently, standard treatment consisted of maximum safe surgical debulking followed by external beam radiotherapy. Recently, chemotherapy with Temozolomide has also become standard for newly-presenting patients with GBM. However, the prognosis for patients with GBM remains extremely poor, with a median survival of 14.6 months [1]. Commonly, radiotherapy starts 2–4 weeks post-operatively but the optimum timing of radiotherapy is not known with certainty and the literature on this subject is scarce. In our region, due to resource and staffing constraints the waiting time for these patients over the decade 1993–2003 fluctuated widely, and was often much longer than 4 weeks. This unfortunate situation has inadvertently provided us with an opportunity to retrospectively study the effect of delay to start of radiotherapy on survival.
Methods Patients were identified from the Wellington Hospital Pathology Department database. Histology reports of all patients operated on by the neurosurgical team at Wellington Hospital from 1st January 1993 to the 31st
123
J Neurooncol
December 2003 were reviewed. Any equivocal histopathology reports were reviewed by a pathologist and a consultant neurosurgeon to confirm a diagnosis of WHO Grade 3 or 4 Astrocytoma (Anaplastic Astrocytoma or Glioblastoma Mutiforme). Relevant data were obtained from hospital records. The variables collected for each patient are listed in Table 1. Patients younger than 15 years and those who had undergone any previous treatment (surgery or radiotherapy) for their brain tumour were excluded. Date of death was obtained from NZ Health Information Statisitics. Deaths before 1 February 2005 were ascertained. All known deaths that occurred were related to glioma. For patients not known to have died survival times were censored to the last date they were known to be alive. Radiotherapy was delivered at one of two radiotherapy centres in our region. Treatment protocols and personnel varied over time and between centres. In one of the treating centres dose prescribed was influenced by the patient’s post-operative prognostic score utilising the MRC index [2]. Patients with good prognostic factors would typically receive external beam radiotherapy by parallel opposed fields to the tumour bed plus a 3 cm margin to a dose of 60 Gray in 30 fractions. In the other centre, for part of the study period, a similar patient was typically treated with a lower dose of 45 Gray, sometimes receiving further radiotherapy at time of recurrence. In both centres, some patients with poor prognostic factors received lower doses in fewer fractions. Some patients did not complete the prescribed course of radiotherapy. For the purposes of this study we recorded only the maximum dose actually delivered to the tumour bed at the time of the initial radiotherapy treatment. Chemotherapy was used rarely during this study period. Typically, palliative chemotherapy (oral CCNU) was offered to younger patients at time of recurrence. Temozolomide was not available in this country during the study period. Performance status at presentation was assessed retrospectively for each patient using a combination of Table 1 Data collected • Age at diagnosis
medical and pre-operative nursing notes. If a patient received help in any ADL (Activity of Daily Living) during their pre-operative stay, a maximum Karnofsky performance score of 60 was given. Waiting time for radiotherapy was defined as the time from the day of surgery to the first day of radiotherapy. Data was entered into a Microsoft Access 97 database. Microsoft Excel was used to create the raw descriptive data and graphs. The statistical programme SAS V9 (SAS Institute Inc., Cary, NC, USA) was used for statistical analysis. Cox’s proportional hazards model was used for multivariate analysis. Our model treated Radiotherapy Dose as a continuous variable. The assumptions of proportionality and linear relationship were checked using the graphical and numerical methods of Lin et al. [3].
Results We identified 221 adult patients with a histological diagnosis of supratentorial Anaplastic Astrocytoma (AA) or Glioblastoma Multiforme (i.e. WHO Grade 3 or 4 Astrocytoma) who had undergone their first neurosurgical procedure during the defined study period, and who were known to have received any radiotherapy post-operatively. In one of the radiotherapy sites we found that radiotherapy records had not been stored systematically and complete data on radiotherapy treatment could not be obtained on 49 patients. In almost all of these cases, the missing data variable was radiotherapy start date. All subsequent analysis is restricted to the 172 patients with a complete data set. Descriptive statistics of the study population are shown in Table 2. Radiotherapy dose varied substantially (Fig. 1) for reasons described above. 53 patients received a dose of radiation of ‡60 Gray. Twenty three patients received some form of adjuvant chemotherapy, usually at time of recurrence. Figure 2 shows waiting time from first surgery to beginning of radiotherapy treatment. We examined whether waiting time was influenced by tumour grade, age or performance status. To assess the relationship between tumour grade (i.e. Glioblastoma or Anaplastic
• Sex • Histological diagnosis (Anaplastic Astrocytoma or GBM)
Table 2 Descriptive statistics of the study population Glioblastoma Multiforme
• Karnofsky Performance Status at presentation (KPS) • Date of first surgery • Extent of first surgery (Surgeon’s impression—complete resection/ partial resection/biopsy) • Start date of radiotherapy
Number
144
Age at diagnosis, median (range)
59 (16 to 83)
• Total radiotherapy dose
Sex, percentage of males
64%
• Other treatments received (e.g. chemotherapy, further resection)
Karnofsky at presentation, median (range)
60 (30 to 90)
• Date of death
123
Anaplastic astrocytoma 28 50 (17 to 73) 50% 70 (30 to 90)
J Neurooncol
Fig. 1 Radiotherapy dose
Median follow up for those not known to have died was 24 months with a range from 4 days to 6 years 1 month. Median survival was 34 weeks for patients with GBM, and 65 weeks for AA. The median survival for patients with GBM who received at least 60 Gray was 58 weeks. Survival curves for these groups are shown in Fig. 3 and 4. The results of the proportional hazards model are shown in Table 3. GBM (as opposed to Anaplastic Astrocytoma) was associated with a Hazard ratio for death of 3.74. Increasing age, increasing time to radiotherapy, and biopsy or partial resection (vs. complete resection) were also associated with Hazard ratios greater than 1, although the difference between partial and complete resection did not reach statistical significance. Increasing radiotherapy dose and performance status were found to positively influence survival (hazard ratios below 1). The proportional hazards model was used to estimate survival curves for hypothetical populations with constant attributes other than time to start of radiotherapy. This method was used to calculate the loss of median survival due to delay in radiotherapy for a typical patient, assuming that 2 weeks post-surgery is the earliest practicable time for starting treatment (Table 4).
Discussion
Fig. 2 Time from surgery to start of radiotherapy
Astrocytoma) and radiotherapy waiting time, the geometric means of the waiting times for the two groups were calculated. (The geometric mean was chosen because waiting times were found to have a log normal distribution). The geometric means were 4.93 weeks and 5.01 weeks for GBM and Anaplastic Astrocytoma respectively (t-test: P = 0.89). Thus there was no significant difference between the waiting times of the two histological groups. To assess the relationship between age and waiting time, and between performance status (KPS) and waiting time, we calculated the Pearson correlation co-efficient of the log of waiting time and the other two variables. This co-efficient relates to the fraction of the observed behaviour that can be explained by the given variables. The closer the value is to 1, the better the correlation. The correlation coefficient of age and WT was –0.03, P = 0.68. Thus age did not appear to influence waiting time. The correlation co-efficient of KPS and WT was 0.16, P = 0.032. Thus there was a correlation between the two, such that increasing KPS (i.e. better performance status) is associated with increased delay.
Our study suggests that delay in receiving radiotherapy after surgery decreases survival of patients with high grade glioma. Delay in radiotherapy has been found to be associated with reduced survival in some other forms of cancer, such as breast cancer [4], but the literature concerning the effect of timing of radiotherapy for GBM is scarce. Fazeny-Dorner et al. [5] found that the median survival of patients receiving combined radiotherapy and chemotherapy within 2 weeks of stereotactic biopsy for GBM was
Fig. 3 Survival curves of whole study population by tumour grade (GBM, AA)
123
J Neurooncol
Fig. 4 Survival curves of GBM patients by radiotherapy dose (
