Controversies concerning the application of brachytherapy in central nervous system tumors

Abstract

Introduction

Brachytherapy (BRT) is defined as a therapy technique where a radioactive source is placed a short distance from or within the tumor being treated. Much expectation has been placed on its efficacy to improve the outcome for patients with central nervous system (CNS) tumors due to the initial promising results from single institution retrospective studies. However, these optimistic findings have been highly debated since the selection criteria itself is preferable to other therapeutic modalities. The fact that BRT demonstrated no significant survival advantage in two prospective studies, together with the emerging role of stereotactic convergence therapy as a promising alternative, has further decreased the enthusiasm for BRT. Despite all the negative aspects, BRT continues to be conducted for the management of CNS tumors including gliomas, meningiomas and brain metastases.

Material and methods

As many controversies have been aroused concerning the experience and future application of BRT, this article reviews the existing heterogeneities in terms of implants choice, optimal dose rate, targeting volume, timing of BRT, patients selection, substantial efficacy, BRT in comparison with stereotactic convergence therapy techniques and BRT in combination with other treatment modalities (data were identified by Pubmed searches).

Results and conclusion

Though it is inconvincible to argue for the routine use of BRT, BRT may provide a choice for patients with large recurrent or inoperable deep-seated tumors, especially with the Glia-site technique. Radiotherapies including BRT may hold more promise if biologic mechanisms of radiation could be better understand and biologic modifications could be added in clinical trials.

Introduction

Despite efforts made by neuro-oncologists over the past several decades, the management of central nervous system (CNS) tumors remains one of the most challenging tasks, with few progress concerning long-term survival especially in WHO grade III and IV malignancies because a great majority of tumors recur within or adjacent to the original tumor site. The failure to achieve local control is strongly correlated with pronounced neurologic deficits, resulting in eventual death in most patients with recurrent tumors.

Radiotherapy (RT) is thought to be the only treatment modality, which even in isolation is proved to substantially prolong the survival of patients with malignant gliomas (Walker et al. 1978). Many efforts have been made to improve the outcome upon these early findings. However, dose escalation with conventional fractionated techniques and various hyperfractionation schemes have shown little impact. Brachytherapy (BRT), RT delivered by placing radioactive sources directly into the local tumor site, was initially expected to be the solution to unsatisfactory local control. Initial single institution retrospective studies showed very promising results, which had created a keen interest in this technique as a favorable option to treat CNS malignancies (Table 1) (Gutin et al. 1984; Halligan et al. 1996; Malkin 1994; Mundinger et al. 1991; Scharfen et al. 1992). Nevertheless, two issues that (1) a high incidence (almost 100%) of steroid dependence caused by radiation necrosis and (2) a high rate (nearly 50%) of reoperation, had tempered these optimistic findings (Chamberlain et al. 1995; Gutin et al. 1984; Laperriere et al. 1998; Malkin 1994). Worse still, BRT had failed to show a significant survival advantage in two prospective randomized studies (Laperriere et al. 1998; Selker et al. 2002).

Table 1.

Outcomes of brachytherapy studies for malignant gliomas

Trial	No. of patients (P/R)	Histology		Technique	Median dose (Gy)	Median survival (months)	Rate of severe toxicityb (%)	Reoperation rate (%)
Temporary BRT
Gutin et al. 1984	31 (R)		GBM (13):AA (18)	125I + nitrosoureas	100	19	12.9	16.1
Gutin et al 1987b	41 (R)		GBM (18):AA (23)	125I + EBRT	88 (BRT) + 60 (EBRT)	18.5 (GBM only:13; AA only:38.3)	9.8	41
Gutin et al. 1991	63 (P)		GBM (34):AA (29)	125I + EBRT + procarbazine, lomustine, vincristine	60 (BRT) + 60 (EBRT)	GBM only:22; AA only:39.3	0	48
Scharfen et al. 1992	307 (P, R)		GBM (172):AA (135)	125I + EBRT (primary tumor)/125I (recurrent tumor)	56 (BRT) + 60 (EBRT)	Primary tumor:GBM only:22,AA only:35.5; recurrent tumor:GBM only:12.3, AA only:13	6	40
Zamorano et al. 1992	48 (P, R)		GBM:AA	125I + resection + EBRT	60 (BRT) + 50 (EBRT)	Primary tumor:15.1; recurrent tumor:10.5	NA	NA
Bernstein et al. 1994	44 (R)		GBM (32):AA (12)	125I + resection	70	11.5	11	26
Chamberlain et al. 1995	14 (R)		GBM (10):AA (3):AO (1)	125I + EBRT + cisplatin	50 (BRT) + 60 (EBRT)	9.5	7.1	64
Shrieve et al. 1995	32 (R)		GBM	125I + EBRT	50 (BRT) + NA (EBRT)	11.5	6.3	44
Sneed et al. 1998	112 (P)		GBM	125I + EBRT + oral hydroxyurea ± interstitial hyperthermia	60 (BRT) + 59.4 (EBRT)	16.8 (Without hyperthermia:19; with hyperthermia:20)	5 (Without hyperthermia); 10 (with hyperthermia)	58 (Without hyperthermia); 69 (with hyperthermia)
Laperriere et al. 1998 a	63 (P)		GBM	125I + EBRT	60 (BRT) + 50 (EBRT)	15.7	4.8	31
Kolotas et al. 1999	53 (R)		GBM (31):AA (22)	192Ir + resection + EBRT	40 (BRT) + 60 (EBRT)	8.8 (GBM only:7.3; AA only:13.3)	17.7	0
Selker et al. 2002 a	133 (P)		GBM (123):AA (10):AO (3):MMG (1)	125I + EBRT + BCNU	60 (BRT) + 60.2 (EBRT)	17 (GBM only:16)	NA	54
Tselis et al. 2007	84 (R)		GBM	192Ir + resection + EBRT	40 (BRT) + 60 (EBRT)	9.3	3.5	0
Permanent BRT
Zamorano et al. 1992	48 (P, R)	GBM:AA		125I + resection + EBRT	110 (BRT) + 50 (EBRT)	Primary tumor:14.6; recurrent tumor:10	NA	NA
Fernandez et al. 1995	58 (P)	GBM (18):AA (40)		125I + resection + EBRT	102 (BRT) + 50.4 (EBRT)	>31 (GBM only:>23; AA only:>31)	10.5	45
Halligan et al. 1996	22 (R)	GBM (18):AA (4)		125I + resection + EBRT	210 (BRT) + 60 (EBRT)	16.3 (GBM only:16)	5	0
Patel et al. 2000	40 (R)	GBM		125I + resection + EBRT	140 (BRT) + 60 (EBRT)	11.8	7.5	0
Larson et al. 2004	38 (R)	GBM		125I + resection + EBRT	300 (BRT) + 60 (EBRT)	13	24	11
Glia-site temporary BRT
Tatter et al. 2003	21 (R)	GBM (15):AA (5):AO (1)		125I + resection + EBRT	50 (BRT) + NA (EBRT)	12.7 (GBM only:8; AA only:17.9)	4.8	0
Chan et al. 2005	24 (R)	GBM		125I + resection	53.1	9.1	8	8
Gabayan et al. 2006	95 (R)	GBM (80):AA (9):AO (4):MMG (2)		125I + resection	60	9.1 (GBM only:9; non-GBM only:10.9)	2	2
Welsh et al. 2007	20 (P)	GBM		125I + resection + EBRT	50 (BRT) + 60 (EBRT)	11.4	14	NA

P primary tumor, R recurrent tumor, GBM glioblastoma multiforme, AA anaplastic astrocytoma, AO anaplastic oligodendroglioma, MMG malignant mixed glioma, EBRT external beam radiation, NA not available

^aProspective randomized studies

^bSerious complications defined as hemorrhage, infection, permanent neurological deficit

Recently rising interest in stereotactic convergence therapy (so-called “stereotactic radiosurgery” though there is actually no surgery, SRS) as a promising alternative for the treatment of malignant glial neoplasms, further decreased the enthusiasm for BRT. As many controversies have been aroused concerning the experience and future application of BRT, in this paper we reviewed both the pros and cons on detailed aspects of this technique, striving to define some directions for its future use.

Controversy 1: choice of radioactive sources

Many radionuclides are now available for clinical use in BRT. The most frequently used clinical radioisotopes are listed in Table 2. To guarantee the therapeutic effect of local radiation as well as the ability to protect health care personnel and patient families from radiation exposure, the choice of an optimal source for BRT should consider the half-life, the activity and energy of the isotope.

Table 2.

Isotopes

Isotope	Dose rate (cGy/h)	T1/2	Energy (MeV)	Half-value layers in tissue (mm)
Iodine-125	3–8 (Permanent) 40–70 (temporary)	60.2 days	0.028 (gamma rays)	20.0
Iridium-192	80–100 (LDR) 120–360 (HDR)	74.2 days	0.38 (gamma rays)	60.0
Cobalt-60	>1,000	5.27 years	1.25 (gamma rays)	100.0
Californium-252	>350	2.64 years	2.3 (neutrons)
P-32a		14.2 days	1.7 (beta particles)	0.8
Y-90a		2.7 days	2.3 (beta particles)	1.1

^aTypically used only for intracavitary BRT

Although variable isotopes were chosen by neuro-oncologists in absence of a gold standard for different clinical conditions, iodine-125 (125I) has become the preferred agent for intracranial BRT because of its easier preparation and relatively low energy (0.028–0.035 meV) of the emitted X-rays, which provides reduced normal tissue exposure and safety for others under the protection of lead aprons (Hughes et al. 1995; Krishnaswamy 1979). Intriguingly, the development of remote afterloading techniques has regenerated interest in isotopes with much higher energy such as 252Cf, 198Au and 60Co, as these devices eliminate the worry of excessive radiation exposure to surrounding people and make it possible to deliver high doses over minutes to hours or for the fractionation of BRT (Beach et al. 1990; Kolotas et al. 1999).

Though most BRT studies in the past have chosen 125I as radioisotope, it is still reasonable to believe that the alternatives of 252Cf, 198Au and 60Co, etc. may serve as other promising choices. However, reports on the comparison of effects of these isotopes are limited; therefore we need to evaluate the substantial benefit of new isotopes in randomized controlled trials.

Controversy 2: high-dose rate vs. low-dose rate

The so-called “dose-rate effect” refers to a phenomenon that the biological effect of a given dose of radiation and the side-effects both positively correlate with the dose rate predominantly (Hall 1988), which is most marked in the range between 1 and 100 cGy/min and more so in normal vs. tumor tissue. Thus, it is understandable that the repair of sublethal radiation damage especially in normal tissue is more efficient during low-dose rate (LDR, 0.4–2 Gy/h) exposures. Moreover, continuous LDR irradiation allows a buildup of damage by synchronizing tumor cells to radiosensitive G2-M phase (Gregg et al. 1979).

However, recent studies have reported that after careful comparison using a biologically equivalent dose value of both modalities, the equivalent effect of high-dose rate (HDR, >12 Gy/h) scheme could only be achieved with a much higher total doses of LDR (Tselis et al. 2007). This may account for the lower rate (6%) of moderate to severe complications in HDR studies (Tselis et al. 2007), compared to up to 26% in LDR studies and even higher when BRT was combined with hyperthermia (Table 1) (Bernstein et al. 1994; Halligan et al. 1996; Patel et al. 2000; Shrieve et al. 1995; Sneed et al. 1992). Additionally, HDR techniques can palliate patients with non-resectable lesions or poor functional scores without external beam radiation (EBRT), in order to reduce the overall hospitalization and treatment time (Micheletti et al. 1994, 1996) The authors concluded that together with decreased total treatment time (<7 days), the overall survival was equal or better than comparable patients received EBRT (Curran et al. 1993; Micheletti et al. 1994, 1996).

To date there is no solid evidence to favor either LDR or HDR in clinical use and currently the choice is probably made upon technical availabilities within each center. The comparison between the two strategies still need to be further addressed.

Controversy 3: optimal total dose for brachytherapy (dose–response effects)

The inverse-square law asserts that, for any point source of radiant energy, the dose absorbed is inversely proportional to the square of the distance away from the source (Vitaz et al. 2005). Besides, the efficacy of irradiation is not only limited by the inherent radioresistance of glioma cells, but also by the radiosensitivity of surrounding healthy brain tissue (Black 1991). Therefore, to provide a better control for the tumor and simultaneously better sparing the normal brain, moderate doses can be given to the margin while much higher doses can be delivered to the center of the lesion.

Tselis et al. (2007) suggested a dose of 30 Gy in minimum to achieve tumor control with a reasonably good chance of palliation and survival for recurrent glioblastoma (GBM) patients. On the other side, analysis of patients treated with increased doses suggests that doses above 40 Gy prescribed to the planning target volume surface confer no added survival benefit. The incidence of delayed radiation-induced complications increases significantly when doses exceed 60 Gy (Sheline et al. 1980). However, according to the recently developed mathematical model, dose escalation has an important predicted effect on survival; that is, a dose of 74 Gy given safely at 2 Gy/fraction would extend the survival of all patients (Burnet et al. 2006). This model holds much promise for assessing the potential benefit and estimating the necessary size in future RT trials.

In addition, it must be noticed that in most BRT trials, the BRT technique has been combined with EBRT, so that merely the dose used in BRT could not be directly compared with those without dose escalations by EBRT (Table 1). Therefore, the optimal total dose for BRT still remains debate depending on the regimens.

Controversy 4: temporary vs. permanent implants

The choice between temporary and permanent implants remains a controversy. Temporary implants keep their advantage in allowing delivery of higher dose rates (40–60 vs. 3–10 cGy/h). Besides, when both at lower dose rates, temporary implants shown lower morbidity than permanent implants (Kreth et al. 1995, 1997). However, patients receiving temporary implants had slightly higher infectious complication rates and prolonged hospitalization for 5–7 days compared to patients receiving permanent implants (Fernandez et al. 1995; Halligan et al. 1996; Patel et al. 2000; Zamorano et al. 1992). Much more importantly, radiation necrosis requiring surgical intervention occurs approximately 26–48% in temporary implants groups whereas it almost never occurs in those with permanent implants (Gutin et al. 1991; Halligan et al. 1996; Koot et al. 2000; Malkin 1994; Patel et al. 2000; Scharfen et al. 1992). Theoretically the tradeoff with lower necrosis rate would be relatively inefficient tumor control due to the relatively lower dose rates of permanent implants. But retrospective studies have failed to show a significant difference in median survival between temporary and permanent implantation (Table 1) (Fernandez et al. 1995; Halligan et al. 1996; Patel et al. 2000; Zamorano et al. 1992). The reason is elusive, yet some authors have boldly speculated that both techniques might be inadequate for controlling tumor progression (Vitaz et al. 2005).

As this bold speculation has not been proved yet, both techniques are worthy of further evaluation, especially the newly emerging Glia-Site temporary BRT technique which has shown very encouraging results in both recurrent and primary glioma treatment (Table 1) (Chan et al. 2005; Gabayan et al. 2006; Tatter et al. 2003; Welsh et al. 2007).

Controversy 5: treatment volumes

Size of the treatment volume and the safety margins varies in the reports of different centers’ experience. Advances in technology have replaced older cumbersome manual dosimetric calculations with more sophisticated computer generated paradigms and three-dimensional treatment planning (Hughes et al. 1995; Julow et al. 2000; Kolotas et al. 1999; Schupak et al. 1995). These technological improvements allow more optimal implants placement to overcome the limitations in standard treatment planning and prevent inhomogeneous tumor coverage as well as over-dosage to normal tissue.

Even though, there is still one problem that with using a focal form of therapy for an infiltrating glioma is defining the target. Taking the experience of SRS as a lesson, Graves et al. (2000) found a significant difference between patients with a low probability of a metabolic lesion outside the treatment volume vs. those with a high probability (P < 0.01), which revealed a significant increase in the relative contrast-enhancing volume, a decrease in time to further treatment, and a reduction in survival for patients with regions containing tumor-suggestive spectra outside the RT target. These data suggested that our current practice of defining the volume for treatment with standard post-contrast radiographic images is likely incorrect. A follow-up study showed that the degree of overlap between the treated volume and metabolically active tumor on magnetic resonance spectroscopy (MRS) could bring about a statistically significant difference in time to tumor progression and median survival rate, suggesting that RT may benefit from the inclusion of MRS in treatment planning process (McDermott et al. 2004). Positron emission tomography may be performed under the same purpose. Hopefully these additional techniques will help answer the question that exactly how much margins should be added in defining the target volume, especially in individualized treatment.

Controversy 6: timing of brachytherapy

Initially, BRT for malignant gliomas is performed after the completion of EBRT or at the time of disease recurrence; however, several recent studies have explored the administration of BRT as upfront treatment at the time of diagnosis prior to EBRT (Micheletti et al. 1994, 1996). Use of cytoreductive procedures prior to initiating treatment is the underlying rationale supported widely, though this rationale itself remains questionable with its benefit elusive from the literature for aggressive surgical resections (Lacroix et al. 2001; Simpson et al. 1993). Besides, the benefit of aggressive neurosurgeries in combination with BRT is even less clear because aggressive surgical resections themselves were positively correlated with better prognosis by some multifactorial analyses (Patel et al. 2000) while others failed to confirm this (Johannesen et al. 1999; Koot et al. 2000; Scharfen et al. 1992).

Additionally, inconsistency of the sequence to delivery BRT and EBRT adds further to the confusion of BRT effect. The optimal sequence is unknown, owing to unclear effects of the differences (BRT as a boost to EBRT vs. prior to EBRT vs. concurrently) on tumor control and risks of radiation necrosis (Fernandez et al. 1995; Zamorano et al. 1992).

In patients with low grade gliomas (LGG), timing of BRT is also highly debated and may impact outcomes significantly (Vitaz et al. 2005). Though some have supported the strategy of immediate postoperative RT with prolonged overall and progression-free survival (Shaw et al. 1989; Shibamoto et al. 1993), opposite observations existed (Grabenbauer et al. 2000). Currently, the notion to postpone RT until time to progression is accepted in many centers concerning severe side effects of RT.

No consistency has been reached on the timing of BRT concerning the aggressiveness of disease management and preference of therapeutic strategy within each center. However, it has to be emphasized that the standards to evaluate treatment effects may play a key role when addressing this issue in the future, for instance quality of life (QoL) may weigh over simply survival time for patients and families (Cheng et al. 2009).

Controversy 7: substantial efficacy of brachytherapy

Efficacy of brachytherapy on malignant gliomas in prospective randomized studies

To date only two prospective randomized studies have been conducted to evaluate the efficacy of BRT (Table 3). The first was a single center study from the University of Toronto performing EBRT alone or EBRT followed by BRT with temporary 125I among randomized patients with primary malignant astrocytomas, which ended with a conclusion that BRT did not provide significant survival benefit in the management of patients with malignant astrocytomas (Laperriere et al. 1998). There was neither difference of Karnofsky performance score (KPS) nor QoL between both groups during the first year follow-up in the same trial (Bampoe et al. 2000). Nonetheless, for both groups there were significant deteriorations in KPS and QoL during the first year compared to baseline scores.

Table 3.

Prospective randomized brachytherapy studies for primary malignant gliomas

Trial	Laperriere et al. 1998	Selker et al. 2002
Technique	125I + EBRT vs. EBRT	125I + EBRT + BCNU vs. EBRT + BCNU
No. of patients	71b vs. 69	133 vs. 137
KPS	70–100	50–100
Histology	GBM	GBM (123):AA (10):AO (3):MMG (1)
Tumor size/median (cm3)	2.5–158/42.3	1–117/21
Isotope (T/P)	125I (T)	125I (T)
Total dose (median, Gy)	57.22–67.69 (60)	60
Dose rate (median, Gy/h)	0.21–1.25 (0.7)	0.4
Adjuvant therapy	EBRT (50 Gy)	EBRT (60.2 Gy) + BCNU (1,500 mg/m2, intravenously)
Extent of resection
Subtotal resection	51	21
Partial resection	70	210
Biopsy	19	39
Median survival (weeks)	62.8 vs. 52.8	68.1 (GBM only:64.0) vs. 58.8
Necrosis in specimen (%)	65(92) vs. 60(87)	11 (15) vs. 15 (23)
Rate of severe toxicity related to implanta (%)	4.8	NA
Reoperation rate (%)	31 vs. 33	54 vs. 47
Favorable prognostic factors	Age ≤50, KPS ≥90, chemotherapy at recurrence, reoperation at the original tumor site	Age ≤54, KPS ≥70, non-GBM histology, male
Recurrence patterns (%)
Original site	82 vs. 93	NA
Original site + multifocal	7 vs. 0	NA
Alive at last observation (%)	8.5 vs. 5.8	NA

T temporary implant, P permanent implant, GBM glioblastoma multiforme, AA anaplastic astrocytoma, AO anaplastic oligodendroglioma, MMG malignant mixed glioma, EBRT external beam radiation, NA not available

^aSerious complications defined as hemorrhage, infection, permanent neurological deficit

^bOnly 63 of 71 patients actually received implant

The second randomized study was a multicenter effort by the Brain Tumor Cooperative Group (BTCG) assessing standard maximal treatment including surgery combined EBRT and carmustine (BCNU) for newly diagnosed GBM, followed with or without stereotactic placement of temporary 125I boost (Selker et al. 2002). Regrettably again, this randomized comparison demonstrated no survival advantage of BRT.

Although the results of the two prospective studies are discouraging, even prevent many centers from continuing to strive for improved BRT trials, it should be pointed out that both the prospective studies had design flaws which may have corrupted the conclusions.

Firstly, the Toronto group chose an EBRT dose of 50 Gy, which is lower than the current standard dose; therefore the overall outcome may be possibly influenced by the EBRT procedure other than BRT. Moreover, they did not use 3-D planning for radioisotope placement which is supposed to improve efficacy and reduce complications. Most noticeably, they randomized patients prior to EBRT and thus over 10% of the patients in the BRT arm never got implanted due to disease progression, medical complications or death. This is especially true because in the intent-to-treat analysis, there was no significant difference in the median survivals between the non-implant and the implant groups (13.2 vs. 13.8 months, P = 0.24); however, when only the actually implant patients were considered, the median survival was 15.7 months and multivariate analyses revealed a trend, though not significant, to improved survival in the implant group (Laperriere et al. 1998). It is reasonable to speculate that if the protocol could be restructured to randomize patients following the completion of EBRT, a survival advantage may have been proven.

Referring to the BTCG trial, an inconsistency with the treatment planning and implanting methodology at multicenters, an unconventional way of using adjuvant BRT before EBRT in cancer treatment (Schupak et al. 1995), and a rate of only 16% patients who underwent gross total resection, may partially be responsible for the discouraging outcome.

Additionally, both the two trials have used 125I seed implants, with which the inhomogeneity associated may in turn led to increased rates of radiation necrosis; and the seed implantation technique itself was technically challenging, usually requiring a second operation to place the seeds (Welsh et al. 2007).

Confounded effect of patient selection bias

The comparison of overall outcomes between studies using other adjuvant treatment modalities and BRT trials is largely confounded by the fact that most BRT trials are biased by selection of patients with favorable outcome predictors (i.e., younger age, improved KPS, smaller size of tumor, and more extensive surgical resection), according to which only a highly selected subgroup (approximately 10–30% of all) is suitable for this modality (Bernstein and Laperriere 1995).

A prospective study was conducted to evaluate the impact of patient selection on the survival of patients with malignant gliomas (Florell et al. 1992). Patients with newly diagnosed GBM and anaplastic astrocytoma were divided into two groups according to the most frequently used selection criteria for BRT: BRT eligible and ineligible. Then all patients were treated identically without the use of BRT, yet the median survival of the BRT eligible group is statistically longer than those from ineligible group (16.6 vs.9.3 months, P = 0.004). The authors concluded that selection alone was responsible for the outcome of BRT. More surprisingly, as indicated by Vitaz et al. (2005), when they use the value of 16.6 months from Florell et al.’s (1992) study as a comparison point with the results from BRT studies for malignant gliomas including those utilized HDR or fractionated techniques, it is clear that most of the BRT studies had similar or even worse outcomes (Table 1).

However, Videtic et al. (1999) had found improved survival even in the poorest prognostic categories, which suggested that selection factors alone did not account for the improved survival seen in their patient group. On the other hand, despite typical eligibility criteria which is much blamed for BRT’s suspicious efficacy, some authors had expanded the criteria to include atypically candidates with multifocal or nonsurgical lesions, diffuse margins, corpus callosum or leptomeningeal involvement, and KPS <70, and ended up with significantly improved survival (Tselis et al. 2007).

Proper endpoint in clinical trials

It has long been argued that since the median survival of patients with GBM is 12–15 months, QoL should been viewed as the most important outcome. Among QoL studies evaluating BRT-related cognitive function damage, Bampoe et al. (2000) had found no significant differences in QoL between the implant and non-implant groups. Later on, a prospective and longitudinal study investigated QoL in GBM patients treated with BRT as well as in their partners had shown a relatively high psychological distress compared to samples from a normal population and non-brain tumor patients (Koot et al. 2004). However, the results could only be interpreted as in agreement with the general high levels of psychological distress in brain tumor patients. Therefore, whether it is a tendency in brain tumor patients or in BRT patients particularly cannot be confirmed by this study since control group was not included. We recommended that in future trials, QoL assessment should be included during follow-up, which may provide us with more crucial evidence rather than merely survival.

Efficacy of brachytherapy on low grade gliomas

Brachytherapy had achieved equal or better overall survival compared to surgery plus RT in LGG patients amenable to implantation, with less surgical morbidity and radiation toxicity than those conducted for malignant gliomas (Kreth et al. 1995). Yet a majority of patients included in these studies had received additional treatments which may have confounded the isolated effect of BRT. As there is no prospective controlled study involving LGG patients, and therapy options (surgery vs. RT vs. surgery and RT) are applied much randomly without strict criteria, comparison between BRT and other treatment modalities is highly inconvincible. Moreover, the opponents have raised various concerns such as steroid dependence, progressive radiation necrosis, possibility of malignant transformation and difficulties in distinguishing disease progression from radiation necrosis (Vitaz et al. 2005).

Efficacy of brachytherapy on brain metastases

Data from several studies on 125I BRT for single brain metastasis had shown satisfactory control of local disease (Bogart et al. 1999; Ostertag and Kreth 1995). Recently Dagnew et al. (2007) proved that permanent 125I applied at the initial operation without whole-brain radiotherapy (WBRT) provided excellent local tumor control while avoiding long-term radiation-induced neurotoxicity; and patient survival rates were at least as good as those reported for resection plus WBRT. The advantage of using BRT could be explained by that tumors recurred predominantly at the site of resection in patients receiving WBRT. However, some argued that distant metastases occurred in patients receiving BRT because no initial WBRT was administered. In this situation, we do not believe that these patients could absolutely benefit from the initial use of WBRT either, given that the additional metastases may be cause by a new post-treatment “tumor shower” of metastatic cells. Additionally, in the only prospective phase II trial of BRT for single brain metastasis, the patients were given Glia-Site BRT (60 Gy) after resection of the tumors whilst no WBRT was given (Rogers et al. 2006). Extracranial metastasis, tumor size and radiation necrosis were significant factors affecting patient survival. The authors concluded that the local control rate, median patient survival time and duration of functional independence achieved with Glia-Site BRT were similar to those achieved with resection plus WBRT or SRS, while eliminating the treatment time and potential adverse effects of WBRT which can reduce QoL and cognitive function. However, it must be noted that patients in this trial were not randomized into two groups by which resection followed by BRT can be compared directly with resection followed by WBRT. Future trials should take randomization into consideration to test the substantial efficacy and tolerability of these techniques.

Controversy 8: brachytherapy in combination with other treatment modalities

It has long been expected that optimal tumor management could be achieved by the combination of multiple treatment modalities. As a delivery strategy of RT, many endeavors have been made to improve the outcome of BRT by using it in conjunction with several other modalities including hyperthermia, reoperation, chemotherapy and immunotherapy.

Hyperthermia was incorporated into the management of brain tumors based on that heat preferentially kills cells in an acidotic and hypoxic environment where radiation is less effective (Gerweck and Richards 1981); and that hyperthermia can be synergistic with the effects of LDR radiation (Harisladis et al. 1978). Though a mild survival advantage were found in some phase I and II studies of hyperthermia combined BRT, this therapeutic modality hold little promise in mainstream treatment protocols given the complexity of its administration (Sneed et al. 1998; Stea et al. 1992).

Reoperation performed either prior to or after BRT for recurrent GBM could lengthen survival compared to counterparts without reoperation (Bernstein et al. 1994; Gutin et al. 1987b, 1991; Halligan et al. 1996; Larson et al. 2004; Patel et al. 2000; Scharfen et al. 1992; Shrieve et al. 1995; Sneed et al. 1992).

Combining chemotherapy, Stupp et al. (2005) reported a median overall survival of 58.4 weeks for RT plus temozolomide in newly diagnosed GBM. However, Tselis et al.’s study which included only 20% of patients who received systemic chemotherapy, partly consisting of temozolomide during their HDR-BRT treatments, had achieved a median overall survival of 78 weeks (Tselis et al. 2007). It is conceivable that the additional survival benefit of 20 weeks is owing to the effect of BRT. Alternatively, the rest 80% of patients who had never received temozolomide apparently had no negative impact on survival (Tselis et al. 2007). On the other side, the combined approach of chemotherapy and RT is likely to increase side effects, especially in substances with strong radiosensitizing potential; and at least certain chemotherapeutic agents (i.e., BCNU, CCNU) were proved by ample evidence to increase the risk of radiation necrosis (Burger et al. 1979; Nelson et al. 1988). Additionally, when chemotherapy has failed, the role of RT as salvage therapy is unclear. Some claimed that this subset of patients represent with biologically more aggressive tumors and raised the question that whether chemotherapy renders the tumors radioresistance (Janss et al. 1995); while others failed to observe a reduced efficacy after chemotherapy.(Kortmann et al. 2000).

Radioimmunotherapy have also been developed to treat brain tumors, which includes two subtypes. One has focused on the delivery of a monoclonal antibody tagged with either 125I or 131I into a surgical resection cavity, which had been used in treating patients with newly diagnosed malignant gliomas (Reardon et al. 2002). The median survival achieved with 131I-labeled antitenascin compares favorably with either 125I BRT or SRS and is associated with a significantly lower rate of reoperation for radionecrosis (Reardon et al. 2002). The other treatment strategy is based on a similar philosophy of localized radiation delivery with radiolabeled thymidine analogs, but targets tumor cells which exhibit significantly higher rates of cell division compared with normal cells (Kassis et al. 1998). Though this technique is still in its infancy due to limited clinical data, animal models have shown promising results (Kassis et al. 1998).

Though some optimistic results had been acquired by the aforementioned combined therapy strategies, to date no comparison has ever been made to evaluate the efficacy and safety of different treatment selections and combinations. Therefore, which to choose (the modality), how to perform (the whole plan), and where to go (future investigation) are all questions to be answered.

Controversy 9: brachytherapy vs. other radiotherapy modalities

Stereotactic convergence therapy

The long existing problem of inadequate local control has not been solved by the once promising BRT. Malignant gliomas recur locally within 2 cm of the treatment volume in 50–80% of patients following BRT and an additional of 10–20% outside the field (Bernstein et al. 1994; Patel et al. 2000; Sneed et al. 1994). Due to its failure in controlling local disease, the invasiveness of this procedure with the antecedent risks of hemorrhage and infection, and the high rate of radiation necrosis, some oncologists have considered SRS as a preferable means to deliver a radiation boost.

The rationale for SRS is similar to BRT in that improved local control may improve survival except that a large conformal dose is delivered over several minutes instead of delivering a large dose over a prolonged period of time (Kondziolka et al. 1999). Sparing of surrounding normal structures is achieved via computer assisted tightly conformal planning (Kondziolka et al. 1999).

Several single institution studies evaluating the efficacy of SRS in malignant glioma management have shown survival similar to those obtained with BRT (Nwokedi et al. 2002; Sarkaria et al. 1995), but without the invasive procedure needed for BRT, resulting in lower rates of complications, especially radiation necrosis (5–22%).(Nwokedi et al. 2002; Sarkaria et al. 1995; Shrieve et al. 1995). Besides, SRS tends to be more cost effective since it can typically be performed on an outpatient basis (Shrieve et al. 1995).

However, a multicenter RTOG study using SRS as an upfront radiation boost for the treatment of GBM failed to demonstrate a survival advantage over the control group (Souhami et al. 2002). Besides, it is noteworthy that SRS is not recommended for lesions >40 mm in diameter with the purpose of limiting the risk for serious morbidity caused by increasing dose inhomogeneity and an increased exposure of normal brain to high doses (Shaw et al. 2000). Furthermore, the large single fraction generally administered frequently leads to necrosis and thus often precludes the employment of this technique in functionally significant areas such as the visual pathway (Kortmann et al. 2003). Taken together, these restrictions greatly limit patient eligibility to be enrolled in.

On the other side, dose rates from interstitial BRT commonly range from 0.4 to 0.6 Gy/h, as compared to 1.8–2 Gy/min with high energy linear accelerator system. To this point, BRT takes advantage of the favorable radiation biology of continuous LDR irradiation mentioned above. Besides, hypoxic cells traditionally thought of being resistant to radiation could be reoxygenated during the period of isotope implanting and thus may partially explain the improved efficacy of this form of radiation delivery (Ling et al. 1985).

The role of SRS in patients with LGG remain conflicting, since current results were from limited centers with patients representing a very inhomogeneous cohort (Grabb et al. 1996; Somaza et al. 1996). Therefore it is reasonable to assess a much longer follow-up considering the slow growth pattern of LGG and possible late toxicity, before a definite conclusion could be made.

In the literature of reirradiation for recurrent disease, the foremost demand is a sparing of normal structure and thus requires a precise and focused RT approach. Stereotactic approaches may be useful, but experience of this endeavor is scarce (Kortmann et al. 2003). Glia-Site BRT, a recently improved technique of BRT which implant an inflatable balloon catheter into the surgical resection cavity for delivery of homogeneous LDR radioactive sources, has been showed to increase survival in patients with recurrent gliomas (Chan et al. 2005; Gabayan et al. 2006). As stated by Shrieve et al. (1995), the effect of BRT is comparable to the outcome achieved by SRS in a subgroup of patients. And according to Tatter et al. (2003), the Glia-Site BRT procedure was carried out without any serious side effects. The rationale is easy to understand since Glia-Site BRT is a LDR approach and possesses all the advantages of continuous LDR irradiation. Moreover, as tumor mostly recurred in the tumor bed with well-defined margins, Glia-Site system could deliver its maximal dose uniformly to the tissues at greatest risk adjacent to the resection cavity; while the dose decreases rapidly with distance from the balloon surface due to the inverse-square law. These characteristics give Glia-Site BRT unique advantages in treating tumors with highest risk of recurrence. And this might partially explain why BRT is still relevant in the setting of prospective randomized studies which have failed to demonstrate a benefit in primary malignant gliomas, that recurrent gliomas may be best suited for BRT vs. other modalities (Bernstein et al. 1994; Chan et al. 2005; Combs et al. 2005, 2007; Gabayan et al. 2006; Larson et al. 2004; Shaw et al. 2000; Shrieve et al. 1995; Tatter et al. 2003, 2007; Welsh et al. 2007).

In addition to its application in glioma management, BRT has also been used for the treatment of many other lesions such as solitary and cystic brain metastases, recurrent atypical/malignant meningiomas, and recurrent, life-threatening extra-axial tumors of the skull base (Bogart et al. 1999; Dagnew et al. 2007; Gutin et al. 1987a; Jaaskelainen and Mantyla 1995; Kumar et al. 1991; Ostertag and Kreth 1995). Though SRS is the preferred method for delivering HDR to these lesions, some patients may not be candidates for this who either failed prior EBRT or SRS, or required craniotomy because of the size of the lesion.

Fractionated stereotactic convergence therapy

Fractionated stereotactic convergence therapy (FSRT) combines both the advantages of SRS and conventional EBRT, thus enables the precise application of RT to a defined target volume, while exploiting the radiobiologic advantage of fractionation and minimizing the risk for severe radiation-induced side effects (Schulz-Ertner et al. 2002).

Early results of this new technique were very encouraging as indicated by Dunbar et al. (1994) and Freeman et al. (1994), that clinical and radiologic improvement was achieved in all patients with previously untreated primary brain tumors. Combs et al. (2005) evaluated the efficacy of FSRT performed as reirradiation in 172 patients with recurrent low- and high-grade gliomas. Improved survival was achieved in WHO grade II, III, IV gliomas, with time to progression and histology influencing survival after reirradiation.

Fractionated stereotactic convergence therapy makes an endeavor to spare normal tissues by fractionation, and thus increases therapeutic ratio especially for relatively larger treatment volumes or tumors untreatable with SRS. Furthermore, another vital advantage of FSRT over BRT and SRS is its good tolerance and safety with regard to therapy-related side effects (Combs et al. 2005). Yet the major drawback of most reports on FSRT is the inclusion of smaller groups of several histologic subtypes, producing somewhat biased results, along with which survival rates may be higher than expected for a chosen histology (Combs et al. 2005).

Recently a new RT modality called hypofractionated stereotactic radiotherapy (H-FSRT) has been taken into investigation, which is characterized by potentially lower toxicity than SRS. When compared to FSRT, the risk for side effects seems increased because of the higher single doses (Combs et al. 2007). But a major advantage of the fewer treatment fractions is the reduction of overall treatment time which is an especially important issue in terminally ill patients, as possessed by BRT.

Simultaneously, a growing number of other newly developed RT techniques have been taken into consideration. With fractionated conformal RT, tumors of almost all sizes and shapes as well as those located in functionally significant areas can be treated with an equivalent geometric accuracy (Kortmann et al. 1999). Intensity modulated radiotherapy (IMRT) offers advantages for patients with complex tumors such as skull base or paraspinal tumors, while accompanies major down sides including increased dose inhomogeneity within the target volume and increased preparation time. Proton/carbon ion RT could provide high degree of dose conformity around the target, which may suggest a favorable application for larger, irregularly shaped tumors in functional areas (Hug et al. 2002), and for certain tumor entities such as radioresistant, hypoxic and/or slow-growing tumors (Schulz-Ertner et al. 2004, 2005) However, whether these approaches hold more promise than BRT/FSRT needs to be evaluated in the ongoing and future prospective clinical trials.

Conclusion

Brachytherapy for the treatment of CNS malignancies has been a subject of dispute ever since its implementation. Despite the detailed inconsistencies concerning its radiological aspects such as the choice of implants, optimal dose rate, timing of delivery and targeting volume, BRT merits more rigorous evaluation of its substantial efficacy. Those single center studies showing a survival advantage are limited by their retrospective nature and lack of a randomized control group. It is of special importance since the known selection bias within this treatment strategy had contributed to the better outcome. Considering all these and additionally the fact that two randomized prospective studies have demonstrated no significant survival advantage; it is inconvincible to argue in favor for the routine use of BRT for malignant gliomas. Yet BRT might still have a role as salvage therapy for a selected group of patients with large recurrent lesions, especially with the Glia-site technique which had made essential technical improvement than 125I seeds implantation used in the two prospective studies. Though it is not clear why many international medical centers do not offer multimodal therapy which includes surgery and BRT, for patients with malignant gliomas whose prognosis are so poor, a temporary or permanent implant for RT which performed concurrently with tumor resection may be severed as a choice. Moreover, the outcome of LGG and brain metastases treated with BRT still needs to be better investigated especially in randomized prospective studies before we draw a definite conclusion. For inoperable deep-seated tumors, BRT may offer a reasonable alternative since that many previously considered inoperable lesions have been encompassed due to advances in surgical techniques, intraoperative mapping and imaging. Most importantly, for each clinical situation, the decision of whether BRT, SRS should be applied, with the advantage of short treatment times but a higher risk of side effects, or whether FSRT, H-FSRT, etc. is chosen, with a lower toxicity risk yet a longer treatment course, should be weighed individually, taking such factors into account as KPS, size and location of the lesion, and previous therapies.

Radiation is believed to be the best adjuvant treatment for malignant gliomas (Walker et al. 1978). Nevertheless, the maximal dose of radiation that can be tolerated by the brain is still well below that required for long-term tumor control. Consequently, great efforts have been made to overcome this limitation. Radiosensitizers and hyperfractionation techniques have been exhaustively studied to end with disappointments. Conformal techniques including BRT and SRS, through which higher dose was strived to administer, have been of no avail in improving local control either. Therefore, what we can learn from these failures might be that dose escalation is fallible as a concept to improve tumor control, regardless of the conformal methodology (Gutin et al. 1991; Shrieve et al. 1999).

Actually, simply increases in dose or merely improvements in technology such as BRT, SRS, FSRT, H-FSRT and IMRT techniques, are unlikely to result in cures or even substantial advancement. In addition to more rigorous study design and more sophisticated technical improvement, more attention should also be paid to a better understanding of the biologic events involved in late radiation complications, or the molecular/genetic factors that is responsible for the radioresistance of CNS malignancies. Radiotherapies including BRT may hold more promise if biologic mechanisms of radiation could be better understand and biologic modifications could be added in clinical trials.