Glioblastoma multiforme (GBM) is one of the most rapidly growing and devastating neoplasms (1). The traditional management of this tumor involves a multi-modal strategy consisting of surgical resection, external beam radiotherapy (EBRT) and, in some cases, a nitrosoureas or procarbazine-based chemotherapy (2, 3).

The principal procedure in the treatment of GBM is the surgical attempt to radically remove the neoplastic tissue. On the other hand, radiotherapy has been demonstrated the most effective adjuvant therapy to surgery (4, 5). Many studies are in progress to enhance the efficacy of radiotherapy and, among different treatment modalities, brachytherapy, radiosurgery and intraoperative radiotherapy are the most frequently studied.

Radioimmunotherapy, as systemic (6, 7, 8, 9) or locoregional application (10, 11, 12, 13) has the potential to become an option in the management of GBM, complementing the above mentioned treatment regiments. 

The systemic application of monoclonal antibodies (MoAbs) for therapeutic purposes is restricted by many factors: high interstitial pressure inside the neoplastic tissue, limited blood supply to the tumor, inhomogeneous and inconstant antigen expression, possible presence of physiological barriers (necrosis and/or fibrosis), formation of immunocomplexes and catabolism of immunoglobulins. Although often impaired in tumor, the blood-brain barrier further hampers the accumulation of antibodies in the malignant tissue. The amount of immunoglobulin actually localized within neoplastic glial tissue has been measured to be less than 0,01% of i.v. administered MoAbs per gram of tumor tissue (6, 14, 15, 16).

Among strategies proposed to overcome these drawbacks and to improve the tumor-non tumor uptake, the 3-step pre-targeting method based on avidin-biotin system should be considered (17, 18). Our group has studied and applied this new method, both as systemic or locoregional radioimmunotherapy (LR-RIT), in glioblastoma patients (6, 8, 9, 19).

Different MoAbs have been utilized for radioimmunotherapy of glioblastoma. In many trials antitenascin MoAbs have been employed (20). Tenascin, a tumor associated extracellular matrix glycoprotein, is over-expressed in the stroma of GBM, but absent in the normal brain tissues. The locoregional application of anti-tenascin MoAbs allows their penetration through the neoplastic tissue, where they bind to their specific antigens (21). Additionally MoAbs, when labeled with high energy beta-emitting radionuclides, are able to destroy a large number of tumor cells in antigen negative areas, due to the “cross-fire effect”.

The results reported so far with LR-RIT indeed demonstrated its effectiveness in glial malignancies (10, 11, 12, 13, 19). In a previous, phase I study we concluded that pre-targeted locoregional radioimmunotherapy with ⁹⁰Y-biotin in glioma patients was a safe procedure. Median overall survival for glioblastoma patients, was 20 months and the maximal tolerated dose (MTD) per cycle, limited by neurotoxicity, was established within the range of 30 mCi (19).

Recently Temozolomide (TMZ), a novel alkylating agent with excellent properties of penetration into brain, was introduced as standard treatment for recurrent high-grade gliomas (22). This prompted us to add it to our LR-RIT in order to increase the efficacy of the treatment.

The role of combining locoregional radioimmunotherapy with Temozolomide in patients with recurrent glioblastoma multiforme is addressed in this analysis.

Over the past six years, more than one hundred GBM patients were treated with locoregional radioimmunotherapy in our Institute. Patients were required to have histologically-proven glioblastoma, immuno-histochemical demonstration of tenascin expression in tumor and a life expectancy of more than two months. A catheter connected with a subcutaneous Ommaya or Rickam reservoir had to be present into the surgical cavity to allow the injection of reagents.

Seventy three patients who were treated with a second surgical debulking (plus catheter implantation) due to recurrent or persistent disease and were given at least two cycles of LR-RIT, were included into this analysis.

The other 30 patients had catheter implantation and LR-RIT after first surgical debulking thus they are not included in the present study.

According to the received treatments, the patients were then divided into 2 groups: a group A (38 patients), who received only locoregional radioimmunotherapy (LR-RIT) and a group B (35 patients), who received locoregional radioimmunotherapy in association with Temozolomide (LR-RIT+TMZ).

Baseline evaluation was performed within 15 days before the first cycle of LR-RIT. All patients underwent a physical examination, in order to determine the Karnofsky performance status (KPS) and a brain cerebral MRI or CT scan with contrast enhancement.  In patients not operated in the study Centers (Milan or Rome) a central pathology review was performed to confirm the diagnosis of glioblastoma multiforme.

The evaluation was repeated at 2-3 month intervals, up to disease progression.

Complete blood count and blood chemistry was drawn every 2 weeks for the patients who received LR-RIT; for the patients who had Temozolomide in addition to LR-RIT, the complete blood count was checked weekly.

Objective response and toxicity were defined according to WHO criteria (23).

The study was performed after approval by the Ethical Committee of the European Institute of Oncology. All patients signed a consent form after receiving detailed information on the aim and potential risks of the study and agreed to the collection of data.

Locoregional radioimmunotherapy. Reagents (anti-tenascin monoclonal murine antibodies BC4, native avidin, ⁹⁰Y-biotin) and radiolabeling procedure were described in our previous report (19). Under sterile conditions, 2-5 mg of biotinylated anti-tenascin MoAb were injected into the surgical cavity, through the indwelling catheter. Eighteen to 24 hours later, 5-15 mg of native avidin was administered and, finally, 14-16 hours later, ⁹⁰Y-biotin was administered. The injected activity ranged from 5 mCi to 25 mCi per cycle, depending on the cavity volume (higher doses in higher volumes). Before each step, patients were pre-medicated with intravenous steroids and Mannitol solution.

Bremsstrahlung scintigraphic images were performed within 1 hour after ⁹⁰Y-biotin administration. Protocol scheduled at least 2 cycles of LR-RIT for each patient, with two month interval.

Chemotherapy. In 1999 our standard treatment was modified and systemic Temozolomide chemotherapy was added to LR-RIT. TMZ was administered orally at 150 mg/m²/d for 5 consecutive days, every 28 days, for the first cycle. Dose was increased to 200 mg/m²/d (only in naive-chemotherapy patients) if no hematological toxicity occurred. Prophylactic anti-emetics were routinely prescribed during chemotherapy. On average, two cycles of TMZ were given between two consecutive cycles of LR-RIT. Chemotherapy was continued unless unacceptable toxicity or disease progression was observed.

The potential association between baseline factors and treatment groups was tested using Student’s t-test, Wilcoxon’s test or Pearson Chi-square test. The various combinations of sites of the GBM were grouped into: “Occipital”, “Frontal”, “Parietal” and “Temporal”. When fitting the models, a proper comparison was possible only for the last three groups, as only 4 patients had a GBM located in the Occipital zone.

Progression Free Survival (PFS) and Overall Survival (OS) were estimated using the Kaplan-Meier method and for OS we considered time from first surgery to death or last follow up, while for PFS we considered time since evaluation at our institute to progression or last follow up. A Cox proportional hazard model (24) was fitted to adjust for a potential confounding effect of age and to assess treatment effects and associations with prognostic factors. When fitting such models: cavity diameter was grouped into 3-4 cm vs 5-5+ cm, depth of the tumor was grouped into cortical vs non-cortical. All the analyses were done in S-Plus (S-PLUS 2000 Professional Release 2, MathSoft, Inc. Seattle, WA).

We analyzed a sample of 73 patients, with an overall median age [min, Max] of 52 years [29, 77], treated with LR-RIT (52%) and LR-RIT+TMZ (48%). In the whole group, the location of GBM was: “Frontal” 36%, “Temporal” 33%, “Parietal” 26% and “Occipital” 5%, while the midline of the brain was crossed in 19% of the cases. Considering the depth of the tumor, we observed 42% cortical locations, 52% more deeply in the white matter and 6% with ventricular communication.

At our first evaluation, patients presented a median Karnofsky score of 70 [40, 100]. Instrumental imaging (MRI/CT) revealed clear evidence of persistent disease in 65 patients (among them, 29 had infiltration in the brain adjacent tissue), while in other 8 only dubious alterations were visible. Table 1 shows the distribution of patients’ characteristics. Apart from patients treated with LR-RIT+TMZ being slightly, but non significantly, older (Wilcoxon’s test p=0.30), and with a smaller, but non significantly, cavity diameter (Fisher’s Exact test: p-value=0.13) the two groups were comparable.

As standard treatment, all patients underwent a first surgery, where the tumor was radically removed in 31 patients and partially removed in 42. Subsequently, all the patients but one, received adjuvant external beam radiotherapy, with a median dose of 60Gy [min=54Gy, Max=72Gy] equally distributed in the 2 groups and 41% also received adjuvant chemotherapy. None of the non-radically operated patients achieved a complete response after adjuvant radio- or chemotherapy.

All patients underwent also a second surgery, that was radical in 11 cases. The median time interval between the first and the second surgery was of 6 months [1^st quartile=4, 3^rd quartile=8].

Regarding LR-RIT, the median number of cycles was 3 [min=2, Max=7], and a median cumulative activity was 50 mCi [1^st quartile=40, 3^rd quartile=60].

In patients who received TMZ in addition, the median number of cycles was 6 [min=4, Max=16].

Table 2, describes the overall treatment received in this group of patients since their diagnosis.

Radiological objective response occurred in 9 patients (3 PR, 6 MR). In a large number of patients (63%) a stabilization of disease was obtained. An example of long lasting progression free survival is shown in figure 1.  Eighteen patients remained in progression. Although statistically not significant, there was a tendency toward higher response rate in the LR-RIT+TMZ group (7 vs. 2) and  the median follow up time after LR-RIT was 14 months [1^st quartile=10, 3^rd quartile=18].

Overall, in this group of 73 patients, the estimated median PFS from the evaluation at our Department was 8 months (95%CI=[6 mm, 10mm]) and the estimated median OS from the first surgery was 21 months (95%CI=[19 mm, 25 mm]), figures 2 and 3.

In the group treated with LR-RIT alone, the overall survival and the progression free survival were 17.5 months (95%CI=[17 mm, 20mm]) and 5 months (95%CI=[4 mm, 8 mm]) respectively. Patients treated with LR-RIT+TMZ had a median overall survival and a progression free survival respectively of 25 months (95%CI=[23 mm, 30 mm]) and 10 months (95%CI=[9 mm, 18 mm]), figures 4 and 5.

In the final Cox model, we considered the factors in table 1 plus radicality of the first surgery and treatment received. These prognostic factors were all well balanced across the two treatment groups. When fitting the final model we found no evidence of a treatment by site interaction. In table 3 are reported the coefficients for the main effects of the final models for PFS and OS. We found that, after adjusting for age and Karnofsky score, radicality of first surgery was a very important prognostic factor, strongly reducing both the risk of disease progression and death.

We couldn’t find a significant association between temporal, parietal or frontal location and OS or RFS. Although non significant, the coefficient for the cavity suggests an increased risk for a diameter equal or greater than five centimeters; a similar consideration applies to a cortical versus a non-cortical location of the tumor. From the final row of table 3, we can see that the importance of the addition of TMZ to LR-RIT is confirmed by these data, as it strongly characterized the prognosis of these patients, together with the radicality of the first surgery (figures 4 and 5).

Four patients experienced a mild allergic reaction at second cycle of LR-RIT [during MoAbs (3 patients) and avidin (1 patient) administration]. Two more patients presented a brief attack of myoclonic epilepsy, within a few minutes after ⁹⁰Y-biotin injection.

In 3 patients, who received four cycles of LR-RIT a delayed, progressive exacerbation of haemiparesis occurred. A subsequent palliative surgical debulking demonstrated a prevalent part of necrotic tissue mixed with tumor. In 2 patients a skin infection around the reservoir was reported.

Twenty-two out of the 35 patients who received LR-RIT+TMZ presented hematological toxicity: a grade II-III lymphocytopenia (in 12/22) and/or a grade II-III thrombocytopenia (in 8/22). No hematological, renal or liver toxicity was reported in the remaining patients.

Since a definitive cure or a long-term remission for glioblastoma seems unlikely in the near future, the common goal of most experimental protocols is the attempt to control disease progression, prolong survival of patients and improve their quality of life. Although, a large number of treatments for recurrent GBM are reported in the literature, there is no substantial evidence to establish general guidelines in the case of GBM relapse.

Considering that more than 80% of recurrent GBM arise within 2 cm from the original margin of contrast-enhancing tumor (25, 26, 27, 28), in case of GBM relapse, locoregional therapeutic approaches, can be applied. They include reoperation and various types of radiotherapy (conformal conventionally fractionated radiotherapy, hypo-fractionated stereotactic radiotherapy, interstitial brachytherapy and radiosurgery). Reoperation may improve neurological status and prolong survival in some cases; however, radiation therapies can provide similar benefit in a less invasive manner (29).

Clinical experiences employing more innovative approaches, such as local hyperthermia, intra-tumoral injection of interleukin-2 activated lymphocytes, α-interferon, recombinant toxins and various chemothepeutic agents, have been reported over the last few years (30, 31, 32, 33). 

Locoregional radioimmunotherapy, with its ability to destroy a large number of tumor cells, was reported as a very effective and safe therapy. In our phase I study (19), involving a small group of patients with high-grade astrocytoma treated with pre-targeted LR-RIT, the median overall survival for GBM patients was 20 months. These results were in agreement with those from other LR-RIT trials and indicated that locoregional approach with radionuclides, can compete with other radiotherapy modalities (11, 12, 13, 20).

A major aim of the current retrospective study was to verify the efficacy, in terms of response and survival, of pre-targeted RIT performed in a larger group of recurrent glioblastoma patients. When we consider the whole group of 73 patients, objective response rate was 12.3%, while the overall and progression free survival were respectively 21 and 8 months. In the subgroup of 35 patients who underwent only locoregional RIT, the respective survival values were 17.5 (OS) and 5 (PFS) months, while in the subgroup of combine therapy (LR-RIT+TMZ) the respective estimates were 25 and 10 months.

Since this is not a randomized study, the comparison between two treatments’ efficacy cannot be regarded as conclusive. Given the limited number of patients and the potential bias of a retrospective analysis, the coefficients from the final models should be considered as a corroboration of the role of these factors, rather than a precise quantification of their effect. We are also aware, that due to differences in study protocols and changes in pathologic and response evaluation, imaging techniques and supportive care, a comparison between studies should be carefully made.

However a median overall survival of 90 weeks compares quite favorably with selected retrospective results. In the major clinical study evaluating survival in glioblastoma patients treated with surgery, with or without external beam radiotherapy, Walker et al reported median survival of 35 and 14 weeks respectively in patients treated with surgery followed by radiotherapy or best supportive care only (5). The most recent report of Brain Tumor Cooperative group NHI Trial 87-01 reported medial survival of 68.1 weeks in patients with GBM intensively treated with surgery, interstitial and external radiotherapy and carmustine (34). On the other hand in a review, which included more than 1400 patients with recurrent high-grade gliomas (anaplastic astrocytoma and glioblastoma multiforme) median time to progression was only 14 weeks (35). Our results, in recurrent patients treated with surgery, radiotherapy and LR-RIT, appear to be more favorable both in terms of overall and progression free survival.

The role of chemotherapy in glioblastoma, either in an adjuvant setting or at recurrence, has often been controversial. Recently, in relation with positive results assessed in pre-clinical and clinical trials (36, 37) the new alkylating drug, Temozolomide was approved for the treatment of relapsing GBM. Since then, Temozolomide has been studied in different treatment schedules both in primary and recurrent GBM (38). More recently, a combination with external beam radiotherapy was introduced (39).

The rationale for combining Temozolomide and radiotherapy is based on preclinical data suggesting additional or, at least, synergistic activity against GBM cell lines (40). In clinic, Stupp and coworkers recently reported a median survival of 16 months in 64 GBM patients treated with external radiotherapy and Temozolomide. At present, a randomized trial of European Organization for Research and Treatment of Cancer (EORTC)/National Cancer Institute of Canada (NCIC), comparing the combination of Temozolomide and radiotherapy to the radiotherapy alone, is ongoing.

From the beginning of 1999, in our Department, Temozolomide was proposed in association with LR-RIT to the new enrolled patients. In our opinion, the basis for combining LR-RIT and Temozolomide include spatial cooperation (Temozolomide should eliminate microscopic disease outside the radiation LR-RIT field), toxicity independence (the two treatments have different toxicity profiles) and the destructive effect of radiation on brain-blood barrier, with improved permeability for chemotherapeutic agents (41). The OS and PFS, of patients receiving the combined treatment, confirmed the hypothesis of better results with combined treatment. In fact, the effectiveness of these treatments was at least additional: both overall survival and progression free survival increased respectively from 17.5 and 5 months for the group treated with LR-RIT alone, to 25 and 10 months in combined treatment (LR-RIT+TMZ).

Regarding the safety, LR-RIT alone or combined with TMZ was very well tolerated. Remarkably, there is a low incidence of early and late neurotoxicity. Only three patients suffered from progressive neurotoxicity that required surgical debulking. In all cases histopatholology revealed a mixture of necrosis and tumor cells, with the prevalence of the former at the bounders of the resected mass. This frequency for radionecrosis is much lower than in brachytherapy or stereotactic radiotherapy series (42, 43). Grade II and III lymphocytopenia and/or thrombocytopenia were observed in 62% of patients treated with LR-RIT and TMZ that is comparable to studies applying TMZ and external radiotherapy (39, 44). Patients’ subjective tolerance of the catheter was very good: we observed only two complications (skin infection) connected with catheter implantation.

In conclusion, our study confirms the efficacy and safety of locoregional radioimmunotherapy in glioblastoma, with a significant increase in survival compared to the one obtained with surgery and external radiotherapy alone. In particular, this study shows that this improvement in survival can be further increased by the multimodal approach of combining LR-RIT with Temozolomide.

We are aware of possible bias in our retrospective study, such as in the selection of patients or in incomplete treatment in patients that were not included into the analysis. However, this results could represent the basis for further prospective trials, to asses timing and schedule of radioimmunotherapy in the therapeutic algorithm of glioblastoma. We can speculate that the best result may be obtained when LR-RIT will be applied as an adjunct to initial surgery. As the patients tolerate the catheter very well, it could be inserted already during the fist surgical intervention. The 2 to 4 week break, between surgical intervention and external radiotherapy, should be a very convenient period to start LR-RIT.

Table 1. 
Characteristics of the Patients at Baseline

Table 2. 
Overall treatment received, since diagnosis, in the two groups of patients

*Delivered to all but one patients after the first surgery

Table 3. 
Estimated Relative Hazard and 95% Confidence Interval from Cox PH model:
Overall Survival (OS) and Progression Free Survival (PFS)

Figure 1.

Male, 52 years of age, diagnosed with glioblastoma multiforme in right parietal lobe, treated with surgery and EBRT (60 Gy).

In June 1999 he was re-operated due to recurrence. A control MRI performed in September 1999 revealed dubious alterations in the surgical bed. The patient was enrolled for LR-RIT (3 cycles) in association with Temozolomide (12 cycles).

Further MRI control did not show Gadolinium enhancement until October 02.

Subsequently, a third neurosurgery was performed, as a salvage therapy, to palliate the symptom of cerebral edema. Histological examination documented necrotic tissue mixed with viable tumor cells.

1.