AIDS-related primary central nervous system lymphoma: a Norwegian national survey 1989–2003

Background Primary central nervous system lymphoma (PCNSL) is a frequent complication in acquired immunodeficiency syndrome (AIDS). The objective of this survey was to investigate incidence, clinical features, radiological findings, histologic diagnosis, treatment and outcome for all patients with histologically verified AIDS-related PCNSL diagnosed in Norway in 1989–2003. Methods We identified the patients by chart review of all cases recorded as PCNSL in The Norwegian Cancer Registry (by law recording all cases of cancer in Norway) and all cases recorded as AIDS-related PCNSL in the autopsy registry at a hospital having 67% autopsy rate and treating 59% of AIDS patients in Norway, from 1989 to 2003. Histologic material and radiological images were reviewed. We used person-time techniques to calculate incidence rates of PCNSL among AIDS patients based on recordings on AIDS at the Norwegian Surveillance System for Communicable Diseases (by law recording all cases of AIDS in Norway). Results Twenty-nine patients had histologically confirmed, newly diagnosed AIDS-related PCNSL in Norway from 1989–2003. Only 2 patients had this diagnosis established while alive. AIDS patients had 5.5% lifetime risk of PCNSL. Their absolute incidence rate of PCNSL per 100 person-years was 1.7 (95%CI: 1.1–2.4) and decreased during the consecutive 5-year periods from 3.6, to 2.5, and to 0.4 (p < 0.001). Median survival from initial symptom of PCNSL was 2.3 months, but one patient was still alive 4 years after completed radiotherapy. Conclusion This is the first national survey to confirm decreasing incidence of AIDS-related PCNSL. Despite dismal survival in most patients, the possibility of long term survival should prompt more aggressive diagnostics in suspected PCNSL.


Methods
This study complied with the Helsinki Declaration. This study sample thus consisted of 29 AIDS patients with histologically verified PCNSL; 27 were diagnosed and treated at Ullevaal University Hospital. Clinical features, histologic diagnosis, radiological findings, treatment and outcome were registered. HAART was defined as the administration of three or more antiretroviral agents including a protease inhibitor or non-nucleoside reverse transcriptase inhibitor. One pathologist (AKG) re-examined histologic material of all these patients from biopsy (n = 2) or autopsy (n = 27) and reclassified it according to the current WHO-classification (2001). For immunohistochemistry, the antibodies CD3, CD20 and Ki67 were used as a minimum. Epstein-Barr virus hybridization was performed in 25 patients. Twenty-six patients had a complete autopsy of the entire body (excluding systemic lymphoma). The remaining 3 patients underwent biopsy (n = 2) or autopsy (n = 1) of the brain only; none of them had signs of lymphoma at chest-and abdomen imaging (n = 3) or bone marrow biopsy (n = 2). Cause of death was established from medical records, autopsy reports and death certificates. Number of new cases with AIDS and number of AIDS deaths in Norway each year were obtained from MSIS (The Norwegian Surveillance System for Communicable Diseases). Named AIDS reporting from laboratories and clinicians to this national surveillance system has been mandatory in Norway since 1983 [19].

Statistical analyses
We calculated absolute incidence rates of PCNSL among persons with AIDS based on numbers of newly diagnosed AIDS-related PCNSL and numbers of patients living with AIDS the same year (PYAR; Person-Years-At-Risk) (data from MSIS). Trend test and 95% confidence interval (CI) for incidence was calculated according to the Poisson distribution.
Duration of survival was calculated as time between date of first symptom attributable to PCNSL and date of death or January 1 st 2008. Survival curves were calculated and plotted using the method of Kaplan and Meier. All p-values in the analyses are two-sided. The analyses were performed with Stata 9.0 (College Station, TX) and SPSS 14.0 (Chicago, IL).
The two patients diagnosed while alive received whole brain radiation; no patients received cancer chemotherapy. HAART was given to 4/7 (57%) diagnosed after the introduction of HAART in Norway (August 1996). In 3 patients HAART was started mean (median, range) 4.3 (3.0, 2-8) months prior to first symptom of PCNSL and was continued after symptom debut. Only one of these had a favourable clinical and haematological response, with normalization of CD4 cell count; this response was first achieved after PCNSL was diagnosed. In one patient HAART was started after PCNSL was diagnosed and gave no favourable response. Toxoplasma therapy was given to 20/29 (69%) patients; 2 responded with clinical improvement and regress of imaging findings but both died rap-idly after their symptoms relapsed 1 and 15 months after initial toxoplasma therapy, respectively.
Median survival from first symptom attributable to PCNSL was 2.3 (95% CI: 1.5-3.2) months. Only 21% (6/ 29) and 7% (2/29) were alive at 6 months and 1 year, respectively ( Figure 3), but one patient has survived for 4.5 years. In this patient PCNSL was established by biopsy 3 months after debut of personality changes; he received HAART and radiotherapy and was alive > 4 years after completed radiotherapy. All deaths (n = 28) were caused by AIDS and PCNSL. Autopsy revealed cerebral toxoplasmosis in addition to PCNSL in 1 patient (not responding to toxoplasma therapy).

Discussion
Different from non AIDS-related PCSNL in Norway [18], the incidence of AIDS-related PCNSL in Norway decreased in 1989-2003. Previous single-and multicenter studies [8][9][10][11][12] have also found decreasing incidence of AIDS-related PCNSL in the era of HAART. However, to our knowledge, this is the first national survey to confirm these findings.
Our survey had both strengths and limitations. It was based on case reporting to a national cancer registry and a comprehensive search for non-reported cases in the autopsy registry at the hospital treating most AIDS patients in Norway. Furthermore, all cases were histologically verified. However, due to the small sample size and the very few patients diagnosed and treated while alive, we could not assess prognostic factors. Also, most radiological images from 1989-1993 were unavailable. The imaging results should still be quite representative for the years 1994-2003 since images were reviewed for 76% (13/17) of the patients from this time period.
Our figures for incidence rate (1.7 per 100 person-years) and lifetime risk (5.5%) of PCNSL among AIDS patients are in the upper range of figures in other studies [1,6,9,15,20]. Lower incidence rates were reported from the US [9,15]: the absolute incidence rate of brain lymphoma per 1000 person-years among AIDS patients Autopsy in all AIDS patients is needed to truly estimate the incidence of AIDS-related PCNSL. In our study, 93% (27/29) had the diagnosis of PCNSL first established by autopsy. None of these patients had a biopsy performed while alive, and 15% (4/27) of them had no focal brain AIDS-related PCNSL in Norway: imaging findings at CT and MRI lesion on imaging (Table 1), so PCNSL was unexpected. Furthermore, PCNSL was considered a likely differential diagnosis prior to death in only 35% (8/23) of those with focal brain lesions, whereas cerebral toxoplasmosis was considered likely in 91% (21/23).
The short median survival from first symptom attributable to PCNSL (2.3 months) was similar to survival times in other larger materials (1.0-2.8 months) [8,16,17]. However, lack of a prompt diagnosis of PCNSL may have excluded patients with very poor performance status from optimal treatment. Long time survival of more than 4.5 years was achieved in one patient treated with radiotherapy and HAART. Similar cases have been reported by others [7,21].
Good performance status, higher CD4 count, whole brain irradiation and HAART have favourable impact on survival in AIDS-related PCNSL [1,8,22,23]. Early diagnosis, to ensure optimal treatment before the performance status has declined, is thus essential. In our study, all but 4 of the 27 patients first diagnosed at autopsy had focal brain lesion(s) on CT/MRI, which should raise the suspicion of PCNSL. Furthermore, 17 of 19 patients with focal lesion(s) who received toxoplasma therapy, showed no clear response. None of these 17 patients underwent stereotactic or open biopsy which may be partly explained by the bad performance status (WHO 2-4) in most of them (77%, 13/17).
Delayed diagnosis of PCNSL is a concern with empiric toxoplasma therapy [1,22]. To reduce this delay, brain biopsy is advised when there is clinical or radiological deterioration during the first week of initial toxoplasma therapy or lack of clinical improvement within 2 weeks [1,24]. Also, PCNSL can be diagnosed without awaiting toxoplasma therapy or biopsy, based on CSF Epstein-Barr virus DNA and 201 Tl single-photon emission computed Survival curve for patients with AIDS-related PCNSL in Norway Figure 3 Survival curve for patients with AIDS-related PCNSL in Norway. Kaplan-Meier survival curve with 95% confidence interval for patients with AIDS-related PCNSL (n = 29). Median survival from first symptom attributable to PCNSL was 2.3 (95% CI: 1.5-3.2) months.
tomography (SPECT). Combining these two methods, a rapid, accurate and minimally invasive diagnosis may be achievable, although the accuracy of this diagnostic approach is not fully clarified in the era of HAART [2].
The present MRI/CT findings are similar to those in toxoplasmosis [1,25]. MRI and CT can seldom differentiate lymphoma from toxoplasmosis [1,20]. If available, diffusion or perfusion MRI, magnetic resonance spectroscopy, positron emission tomography and SPECT [1,25] might help to select patients for biopsy versus empiric toxoplasma therapy. So might PCR examination of the CSF to detect Ebstein-Barr virus DNA or Toxoplasma gondii DNA [1,2], but none of these tests are completely specific. Biopsy is often needed, but should be weighted against its relatively high procedural risk in AIDS patients [1]. Further research on the combination of less invasive diagnostic procedures is required to reduce the need for biopsy.