Intravenous methotrexate for primary central nervous system non-Hodgkin's lymphoma in AIDS

Objective: To evaluate high-dose intravenous methotrexate in primary central nervous system (CNS) lymphoma in HIV-infected patients.

Design: An uncontrolled pilot trial.

Setting: An infectious diseases department in Paris, France.

Patients: All consecutive AIDS patients with primary CNS lymphoma attending the same unit from August 1994 to March 1996.

Interventions: Methotrexate was intravenously administered at a dose of 3 g/m² every 14 days with leucovorin rescue. A maximum of six cycles was planned. Steroids were given to all patients and haematological growth factors were administered as required.

Main outcome measures: Rate of response, time to response and survival.

Results: Fifteen patients (10 with histological documentation) were recruited. The median time since clinical onset was 27 days (range, 7-69 days), median Karnofsky score was 51 (range, 30-70), and mean CD4+ cell count was 30 ± 19 × 10⁶/l (range, 7-69 × 10⁶/l). Complete responses, defined as clinical improvement and disappearance of contrast-enhancing brain abnormalities on computed tomography or magnetic resonance imaging, were obtained in seven out of 15 patients (three out of 10 patients with histological diagnosis and four out of five patients without histological confirmation). The Karnofsky score of these seven patients improved to 80 ± 10 (range, 70-100). The mean time taken to respond was 62 ± 20 days (range, 45-90 days). One patient relapsed at 6 months. Six patients failed to respond, and two died of severe sepsis on days 15 and 45. The median survival time was 290 days (range, 11-570 days): 73 days (range, 11-570 days) in the 10 patients with histological diagnosis, and 347 days (range, 286-409 days) in the five patients without histological confirmation. Side-effects occurred in 10 patients, with gastrointestinal disorders in five, mucositis and skin rash in two, and fever in three patients; however, these events were mild and did not require cycle postponement or dose changes. No cognitive dysfunction occurred.

Conclusion: Methotrexate appears to be an attractive alternative to radiation therapy for primary CNS lymphoma and is associated with a far greater improvement in quality of life relative to historical series of radiation therapy.

Introduction

The frequency of non-Hodgkin's lymphoma (NHL), an AIDS-defining disease, has increased in recent years. Primary central nervous system lymphoma (PCNSL) is a frequent manifestation of NHL and is the second most common cause of cerebral mass lesions in AIDS. Therapy is currently based on experience with PCNSL in HIV-seronegative patients, and whole-brain radiation therapy (WBRT) with steroids is generally the first-line treatment. Although this treatment effectively reduces the size of tumours, the overall prognosis remains very poor, with a median survival time of 2.5-4.8 months [^1-5]. In addition, WBRT is associated with progressive impairment of cognitive function and neurological status related to chronic leukoencephalopathy.

Several chemotherapy regimens have been tested in uncontrolled studies as alternatives or adjuncts to radiation therapy in HIV-seronegative patients [^6-9]. Intravenous methotrexate (a folic acid antagonist) is attractive in this setting, given its activity against lymphoma cells, its good penetration into the brain parenchyma (even when the blood-brain barrier is intact), and its relatively good tolerability (notably the low risk of leukoencephalopathy). Recently, Tosi et al. [¹⁰] reported encouraging results of the combination of methotrexate plus high-dose zidovudine in HIV-infected patients with non-cerebral NHL [¹⁰]. However, intravenous methotrexate has never been prospectively assessed in AIDS-related PCNSL, except in a study in which chemotherapy was followed by radiation [¹¹]. We report a prospective evaluation of intravenous methotrexate in this setting.

Patients and methods

Patients

This prospective study was conducted between August 1994 and March 1996. All HIV-infected patients with confirmed or presumptive PCNSL attending the same unit (Service des Maladies Infectieuses, Hôpital Rothschild, Paris, France) were enrolled during this period. PCNSL was confirmed by histological examination of tumour samples obtained by stereotaxic biopsy. Presumptive PCNSL was defined as follows: contrast-enhancing mass lesion(s) on computed tomography (CT) or magnetic resonance imaging (MRI), or both, that were unresponsive to a minimum of 7 days of antitoxoplasmic therapy. All the enrolled patients gave their informed consent.

Treatment

Intravenous methotrexate (Laboratoire Roger Bellon, Neuilly sur Seine, France) was administered at a dose of 3 g/m² every 14 days, with leucovorin rescue. A maximum of six cycles was planned and treatment was discontinued in the event of a complete response. Methotrexate was diluted in 750 ml 5% glucose and infused over 1 h after parenteral hyperhydration and alkalinization to obtain a minimal urinary output of 100 ml/h with a pH above 7.0. Calcium leucovorin rescue (20 mg intravenously) was begun 4 h after the end of the methotrexate infusion and repeated every 6 h until the methotrexate plasma level fell below 5 × 10^-8mol/l. Patients received steroids (methyl prednisolone 120 mg daily) at each of the first three cycles for 4-7 days. In case of disease progression or relapse, WBRT was given at a total dose of 20-30 Gy.

Evaluation during therapy

The endpoints of the study were the response rate, time to obtain a response, and survival. The clinical response was assessed after each cycle. Repeat CT or MRI was performed every two cycles during treatment, and then every 2 months after the end of chemotherapy. Responses were defined as a clinical improvement with disappearance of contrast-enhancing abnormalities on brain CT or MRI (complete response), or with regression of more than one-half of the larger diameter of the contrast-enhancing abnormalities (partial response). Those responses had to be persisting for at least 1 month at the final categorization. Non-response and progressive disease were defined as clinical deterioration or unchanged or increased size of radiographic abnormalities. All CT and MRI films were independently reviewed by a radiologist. Recurrence was defined as clinical deterioration and reappearance of contrast-enhancing radiographic abnormalities, or new lesions. The time to response was measured from the first cycle of methotrexate, and survival was measured from the date of diagnosis.

Statistical analysis

Results are expressed as medians and means ± SD. Quantitative values were compared by means of analysis of variance. The Kaplan-Meier method was used to draw response and survival curves. Univariate analysis (log-rank test) was used to identify factors predictive of response and survival.

Results

PCNSL was diagnosed in 15 (4.3%) out of 349 AIDS patients seen during the study period. The diagnosis was confirmed histologically in 10 patients (group 1) and was presumptive in five patients (group 2; biopsy was contraindicated in two patients with thrombocytopenia, and refused by another three patients). The clinical and biological characteristics of the patients at diagnosis are shown in Table 1. All the patients were profoundly immunosuppressed, with a mean CD4+ cell count of 23 ± 37 × 10⁶/l (range, 11-117 × 10⁶cells/l). Maintenance therapies for cytomegalovirus or mycobacterial diseases were continued. The mean duration of antitoxoplasmic therapy before diagnosis was 27 ±18 days (range, 14-63 days) in group 1, and 12 ±8 days (range, 7-23 days) in group 2. Radiological abnormalities are also shown in Table 1. The Epstein-Barr virus (EBV) genome was detected by polymerase chain reaction (PCR) in cerebrospinal fluid of seven out of eight patients in group 1 and one out of three in group 2. Two group 2 patients were seronegative for Toxoplasma gondii. All group 1 patients had malignant B-cell lymphomas and are referred to as having large-cell or immunoblast-rich PCNSL. EBV encoded RNA hybridization was positive in five out of the seven patients studied.

Table 1

Results were analysed separately in the two groups. Because the patients' characteristics did not differ statistically between the two groups, data on the response to therapy, survival, and adverse effects were pooled for analysis.

Efficacy

Complete responses were obtained in seven out of 15 patients (three out of 10 in group 1 and four out of five in group 2). The mean time to response was 62 ±20 days (range, 45-90 days). The mean Karnofsky score of these seven patients increased from 50 ± 20 (range, 40-70) at diagnosis, to 80 ± 10 (range, 70-100) after completion of treatment. One of the seven responders belonging to the group 2 relapsed at month 6; this patient had received only three cycles of methotrexate because of a complete response on day 45. He was subsequently retreated with methotrexate followed by WBRT. Only partial response was obtained and the patient died 6 months later.

Six of the 15 patients (five patients of group 1 and one patient of group 2) failed to respond to chemotherapy 38 ± 27 days (range, 11-85 days) after the first cycle. Four patients died after 51 ± 33 days (range, 11-91 days). Two patients subsequently received WBRT: one failed to improve and died on day 67 after diagnosis, and the second had a partial response but died on day 286 of disseminated lymphoma.

Two patients of group 1 died of severe sepsis (pulmonary aspergillosis and Pseudomonas aeruginosa septicaemia), one after 15 days (response to chemotherapy was unassessable), and one after 45 days (partial response on day 30; Tables 2 and 3).

Table 2

Table 3

The median survival time was 290 days (range, 11-575 days): 73 days (range, 11-575 days) in patients of group 1, and 383 days (range, 290-409 days) in patients of group 2 (Fig. 1). The median survival time of the seven responders was 570 days (range, 170-575 days). Despite a marked overall difference between the two groups, the survival time was similar in responders from group 1 (570 days; range, 170-575 days) and group 2 (383 days; range, 336-409 days). Five of the seven responders subsequently received an antiretroviral regimen containing a protease inhibitor (indinavir or ritonavir; Table 2). Five of the responders were alive and had no evidence of disease 170-575 days after therapy, whereas the remaining two died, one from AIDS-related infections at month 19 (no evidence of PCNSL relapse), and the other at month 12 (PCNSL relapse diagnosed at month 6). Autopsies were not performed.

Fig. 1

None of the following parameters selected for univariate analysis was predictive of response or survival: clinical manifestations, Karnofsky score, weight loss, single or multiple focal abnormalities, concomitant infections, duration of empirical antitoxoplasmic treatment and CD4+ cell count.

Tolerability

Adverse events are listed in Table 4. Neutropenia was the most frequent form of haematological intolerance and required administration of haematopoeitic growth factors to six patients, although it was not treatment-limiting. All other side-effects were mild and required neither dose adjustment nor postponement of chemotherapy. No cognitive dysfunction occurred, even in long-term survivors.

Table 4

Discussion

The main findings of this study are as follows: (i) 47% of patients showed a complete response to high-dose intravenous methotrexate, with a median survival time of 19 months; (ii) side-effects were common but not treatment-limiting; (iii) the rate of relapse in responders was 14%; and (iv) no clinical signs of neurological impairment were seen in responders.

Although the study was not controlled, the results were very encouraging relative to published uncontrolled studies of other PCNSL therapies [^11,12]. Indeed, the partial response rate to radiation therapy was below 50%, and the median overall survival time varied from 2.1 to 5.5 months in recent series [^1-5]. Although PCNSL was not histologically proven in five patients (group 2), other diagnoses were highly unlikely given the use of stringent clinical criteria. Moreover, the fact that four of these five patients responded supports the diagnosis of PCNSL.

The response rate was comparable with the results obtained by Forsyth et al. [¹¹], who evaluated a triple-drug regimen including methotrexate at the same dose and obtained a similar response rate (four out of seven, 57%). On the other hand, the antiretroviral regimen containing protease inhibitors given to the responders certainly contributed to the long survival time observed in this subset of patients. It may also explain the low incidence of PCNSL relapse. However, the risk of relapse may still have been reduced by administering the full six cycles to all responders, because the only patient who relapsed received only three cycles.

The lack of late neurological toxicity contrasts strikingly with the major impairment of neurological status seen in all HIV-infected patients treated with WBRT alone [¹³] or WBRT followed by chemotherapy [¹¹]. The regimen we used did not appear to worsen the prognosis of HIV infection, which was the main reason why use of WBRT has been preferred to aggressive multidrug therapy. Indeed, the survival times and causes of death in the two responders who died did not differ from the events observed in cohorts of patients with very low CD4+ cell counts. Similarly, the two patients who died early in the course of intravenous methotrexate therapy developed infectious complications commonly seen in such patients. A negative effect of chemotherapy cannot be ruled out, however, even though these patients were in very poor overall condition when methotrexate therapy was started.

The regimen we used included glucocorticoids, which are known to reduce the oedema associated with these tumours and to have a lymphotoxic effect. Initial responses to steroids occur in 30-70% of cases [¹³], but are usually short-lived and thus cannot account for the mid-term and long-term benefits that we observed.

Univariate analysis failed to identify factors predictive of short-term responses or longer survival. This may have been due to the small number of patients and to the fact that they were relatively homogeneous with respect to the extent of the tumour(s), the severity of immune deficiency, and overall condition. Indeed, prognostic factors in AIDS-related lymphoma have been determined only in systemic forms, in which patients are far more heterogeneous, especially with regard to immune deficiency (previous opportunistic events and CD4+ cell counts) [¹⁴].

This study indirectly shows the benefit of differential diagnosis of brain lesions in HIV-infected patients, even when they are profoundly immunodeficient. Central nervous system masses are generally treated empirically as presumed toxoplasmosis for approximately 2 weeks followed by stereotaxic biopsy when there is no clinical or radiological improvement. However T. gondii-seronegative individuals should benefit from early brain biopsy. This policy is not always applied, either because neurosurgeons are reluctant to biopsy or because treatment of PCNSL has so far been perceived as only palliative [¹⁵].

We feel the need to stress the potential benefits of thoroughly assessing brain mass lesions. In particular, the diagnostic algorithm could be improved by two non-invasive procedures: both negative toxoplasmic serology [¹⁶] and detection of EBV genes in CSF by means of PCR [¹⁷] are predictive of PCNSL, and call for brain biopsy instead of empirical antitoxoplasmic treatment, especially in patients who are receiving primary toxoplasmic prophylaxis.

These results suggest the effectiveness of methotrexate alone in PCNSL. Because of the small size of this pilot study, the true response rate to this agent was not clearly assessed. A controlled trial of intravenous methotrexate versus WBRT is theoretically needed to confirm these encouraging results. However, the implication of a controlled trial is debatable from an ethical standpoint given the high response rate to intravenous methotrexate and the lack of neurological sequelae. The potent effect of antiretroviral regimens containing protease inhibitors on the survival of even profoundly immunodeficient HIV-infected patients encourages a switch of the management of PCNSL from palliative to curative care.