Central nervous system lymphoma: Patterns of incidence in Indian population and effect of steroids on stereotactic biopsy yield
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.128272
Source of Support: None, Conflict of Interest: None
Background: Epidemiology of primary central nervous system lymphoma (PCNSL) world-wide shows an increase in incidence linked to human immunodeficiency virus (HIV) pandemic. Materials and Methods: This retrospective review of case records analyzed the trends of hospital-based incidence of PCNSL over two decades (1991-2010), relation to immune status and effect of steroids on yield of stereotactic biopsy (STB). Results: A total of 76 cases of PCNSL were diagnosed over a period of two decades. Incidence of lymphomas amongst all biopsied lesions showed a gradual increase from 0.18% at the beginning of study period to 0.41% at the end of study period. Only 8.6% (3 of 35 tested) of the PCNSL patients were positive for HIV. The mean age of patients with HIV infection (31.3 ± 3.5 years) was significantly lower compared with those without HIV infection (44.7 ± 10.9 years) (P = 0.033). Diagnosis was obtained by open biopsy in 32 patients (42.1%) and STB in 44 patients (57.9%). Open biopsy yielded a histological confirmation of PCNSL in all cases. Among those who underwent STB, the incidence of negative biopsy with short duration of steroids (≤1 week) was 33.3% and increased to 57.1% with increasing duration of steroid treatment (>1 week). Conclusions: This study documented an increase in hospital based incidence of PCNSL in our institute, independent of HIV association. Steroid intake administration for more than a week prior to biopsy adversely affected the yield of STB in PCNSL.
Keywords: Central nervous system lymphoma, human immunodeficiency virus, incidence, stereotactic biopsy, steroids
Lymphomas of central nervous system (CNS) are rare tumors accounting for 3.1% of all primary brain tumors and 2-3% of non-Hodgkin's lymphomas. They may be either secondary to systemic lymphoma or primary without the evidence of systemic involvement. Since secondary spread of systemic lymphoma to CNS occurs late in the course of disease and is a relatively rare event in the absence of primary disease, almost all cases presenting initially with CNS involvement are considered primary central nervous system lymphomas (PCNSL). 
Epidemiology of PCNSL world-wide shows a progressive increase in incidence followed by stabilization in the last three decades. , This increasing incidence was partially linked to the human immunodeficiency virus (HIV) pandemic. The persistent increase among the immunocompetent however, cannot be rationalized and it remains unresolved whether this increasing incidence is a global phenomenon or is geographically limited. Intriguingly, CNS lymphomas in the Indian population seem to differ in epidemiology and clinical presentation from Western population. For instance, increasing incidence of CNS lymphoma in immunocompromised host has not been seen in Indian studies. , The trend in hospital-based incidence, association with immunocompromised state, histological subtypes of PCNSL in Indian population and modalities of diagnosis and management needs further evaluation in Indian scenario as published studies from India are few and are restricted to few case series and reviews. ,,,,,, The present study was proposed to analyze the trends of hospital-based incidence of PCNSL over two decades, the diversity of clinical symptomatology, relation to immune status and histopathological characteristics.
This retrospective study included all cases of PCNSL diagnosed histopathologically between 1991 and 2010 in our institute, a tertiary care neurosurgical center in South India. The slides were reviewed by the neuropathologist for confirmation of diagnosis. Cases not consistent with lymphoma on review, secondary CNS lymphoma, or with inadequate tissue to permit immunohistochemical characterization were excluded. The clinical and radiological data were obtained from the case records and radiology archives. All the patients were referred to an oncology center following tissue diagnosis for adjuvant therapy and received chemotherapy based on cyclophosphamide, Adriamycin, vincristine and prednisolone (CHOP) or rituximab added to CHOP (R-CHOP) regimen for PCNSL. They were further followed-up by the medical oncologists in an oncology center.
All sections were stained for routine hematoxylin-eosin stains. Immunophenotyping of tumor cells was carried out by indirect immunohistochemistry for T cell (CD3, monoclonal, 1:50, Biogenex, USA), B cell (CD20, monoclonal, 1:50, Biogenex, USA) and macrophage (CD68, monoclonal, 1:100, Biogenex, USA) markers on serial sections. Additional markers like glial fibrillary acidic protein (monoclonal, 1:100, Novocastra Labs, USA) were employed where essential. The morphological phenotype (centroblastic/immunoblastic) on routine stains and immunophenotype (B cell/T cell) were ascertained. Tissue response as in tumor necrosis, macrophage infiltration of tumor and reactive astrocytic response in adjacent infiltrated host parenchyma were noted.
The clinical, histopathological and radiological features of all the patients, influence of HIV status were analyzed. SPSS 15.0 for Windows (SPSS Inc., USA) was used for statistical analysis and appropriate tests were applied to determine significance.
Totally 76 cases of PCNSL were diagnosed during the study period of two decades (1991-2010). A progressive increase in the hospital-based incidence of CNS lymphoma was noted during the study period, with a gradual increase from 0.18% of all biopsied lesions at the beginning of the study period to 0.41% at the end [Figure 1]b and c. Males were predominant (76.3%) among the cohort. The age at presentation ranged from 18 to 75 years (mean: 46.80 ± 13.23 years). Most patients were in 5 th and 6 th decade [Figure 1]a. The median duration of clinical symptoms at presentation was 1.5 months (range: 3 days-2 years). The clinical features at presentation are given in detail in [Table 1]. Headache was the most common symptom (68.4%), followed by features of raised intracranial pressure (47/76, 61.8%). Hemiplegia/hemi paresis (44.7%), vomiting (43.4%), cognitive decline (39.5%) and seizures (22.4%) were the other common symptoms at presentation. Cognitive decline was reported as a symptom in only 30 (39.4%) patients, whereas incidence of cognitive decline was much higher on formal testing (40/76, 52.6%). Systemic symptoms such as generalized weakness, fever, weight loss etc., were seen in 14 (18.4%) patients. The cranial nerves most commonly affected were cranial nerve III (n = 5), VI (n = 1) and VII (n = 3).
Neuroimaging data was available in 72 patients [Figure 2]. On computed tomography/magnetic resonance imaging (MRI), 44 (61.1%) patients had a single lesion and 28 (38.9%) patients had multiple lesions. Lesions were located mainly in the periventricular regions (n = 23; 31.9%), lobar subcortical (n = 20, 27.7%), involving both periventricular and lobar regions (n = 24, 33.3%) or posterior fossa (n = 5, 6.9%). The most frequent pattern of enhancement on gadolinium administration was a homogenous enhancement (62.5%), followed by heterogeneous (25%), patchy (6.9%) and ring (4.2%) enhancement. The lesion was non-enhancing in one case. Most lesions had moderate perilesional edema.
Mode of biopsy
The tissue diagnosis was obtained by open biopsy in 32 (42.1%) patients and stereotactic biopsy (STB) in 44 (57.9%) patients. On analyzing factors influencing the decision for mode of biopsy, presentation with altered sensorium was the most frequent factor favoring open surgery, though not statistically significant (P = 0.586). Lobar location also prompted decision for open biopsy (P = 0.001) compared with periventricular location of lesions. Further, higher proportion of patients (37/57, 64.9%) underwent STB during 2001-2010 when compared with the period between 1991 and 2000 (7/19; 36.8%), which was statistically significant (P = 0.032) [Table 2]. This was due to advances in imaging allowing more precise recognition of lymphoma and easier access to stereotactic biopsies, thereby avoiding craniotomy and major microsurgical procedures.
Influence of steroids on biopsy yield
History of steroid treatment prior to biopsy was available in 72 of 76 patients. Forty-five (63.8%) patients did not receive any steroids, 17 (23.8%) patients received steroids for <1 week prior to surgery and 9 (12.5%) patients received steroids for >1 week prior to surgery.
All patients who underwent open biopsy (n = 32) had histological confirmation of PCNSL. Of the 42 patients who underwent STB for tissue diagnosis, five were inconclusive. Of these five, four had received steroid treatment for >1 week prior to biopsy. The pre-procedural imaging diagnosis in these four patients was demyelination (n = 2), acute demyelinating encephalomyelitis (n = 1). In one patient who was evaluated and treated at another center, the initial diagnosis was not known. In all these patients, the initial STB was inconclusive. All the patients demonstrated clinical improvement with steroids and were followed-up clinically and radiologically. Two patients had complete radiological resolution and two patients had reduction in size of the lesions at 1 month follow-up. One patient with posterior fossa lesion had increase in size of lesion, for which he underwent open surgery, which confirmed lymphoma. In patients with good response to steroids (n = 4), the lesion recurred after a mean duration of 5.75 months (range: 2-10 months). On reappearance of lesions, these patients underwent repeat STB which yielded a confirmatory diagnosis of CNS lymphoma.
Steroid use prior to surgery was divided arbitrarily into short duration (≤1 week) and long duration (>1 week). An attempt was made to analyze the effect of duration of steroid treatment on diagnostic yield [Table 3]. Of the nine patients with steroid use for long duration, two underwent open biopsy which yielded a diagnosis. The remaining seven patients underwent a STB, in whom four had a non-diagnostic initial STB (P < 0.001). Among 17 patients who received short duration steroids before biopsy, 14 had undergone open biopsy which yielded a confirmatory diagnosis. Biopsy was inconclusive in one out of the three patients who underwent STB. We noted that open biopsy yielded a confirmatory diagnosis, irrespective of steroid intake and duration of steroid treatment [Table 3]. Interestingly, a short duration of steroid intake demonstrated a trend toward inconclusive STB when compared with those who were not started on steroids (P = 0.086), while long duration of steroid intake had significant influence on diagnostic yield of STB (P < 0.001). The incidence of negative STB with short duration of steroids was 33.3%, which increased to 57.1% with a long duration of steroid treatment.
CNS lymphoma and HIV serology
HIV serology status was available for 35 patients among the total cohort. Many patients, especially in early part of the study period, were not tested for HIV status. Of the 35 cases tested, only 3 (8.6%) were positive. Among HIV negative patients, one had a low CD4 count (221/μL). No other cause for immunocompromised state such as post-transplant state, history to suggest congenital immunodeficiency states or long-term steroid usage was identified in any other patients.
We noted that the mean age of patients with HIV (31.3 ± 3.5 years) was significantly lower compared with patients without HIV (44.7 ± 10.9 years) (P = 0.033). Among the HIV positive patients, two presented with seizures and one patient presented with hemi paresis. The duration of symptoms was short in two patients (<1 month), while one had seizures for 1 year before he underwent evaluation. Imaging available in only two of these patients, revealed uniformly enhancing lesions with moderate perilesional edema. One patient had two lesions, a periventricular and a lobar; while the other had a single periventricular lesion. Diagnosis was obtained by STB (n = 2) and open surgery (n = 1).
Histopathology and immunohistochemistry
Detailed slide review and immunohistochemistry was performed in 69 patients in whom adequate tissue was available. All cases revealed features of a non-Hodgkin's lymphoma that were immunophenotypically diffuse, large B cell in phenotype. Histologically, the tumor was composed of densely cellular fragments of tightly cohesive cells dispersed in sheets. The individual tumor cells had centroblastic or less commonly immnuoblastic morphology [Figure 3]a. Mitotic activity was high with frequent apoptotic figures in presence of necrosis. Characteristic perivascular or angiocentric pattern splaying apart the reticulin framework of blood vessels was seen in almost all tumors [Figure 3]b and c, more common at the infiltrating borders of the tumor and hence more likely to be seen in resection specimens than STB unless the infiltrating margins were sampled. In cerebellar lymphoma, tumor was frequently noted to extend along Virchow Robin spaces as perivascular sleeves, reaching up to the meninges to flood the subarachnoid space.
Immunophenotyping revealed all cases to be CD20 expressing large B cell in phenotype [Figure 3]d. Admixture of mature CD3 expressing T cells and less frequently macrophages was seen to variable degree in all cases [Figure 3]e and f.
In those who received short duration of steroids prior to biopsy, early apoptotic changes were seen in tumor cells, in addition to macrophage infiltrates [Figure 4]a and b. CD20 immunolabeling was still retained in tumor cells [Figure 4]c Infiltrates of mature T cells were less frequent [Figure 4]d, while astrocytic response was mild and restricted to brain tumor interface [Figure 4]e.
In cases who received long duration of steroids, apoptosis was prominent with several tangible body macrophages with ingested apoptotic bodies conferring a Burkitt-like phenotype to sheets of foamy macrophages [Figure 4]f and g with numerous mature T cells infiltrating the tumor in parenchymal and perivascular distribution [Figure 4]i. CD20 immunoreactivity was patchy with several tumor cells appearing to lose immunoreactivity [Figure 4]h while glial response was exuberant [Figure 4]j. No case of T cell lymphoma, anaplastic large cell lymphoma or Hodgkin's lymphoma was seen in this series.
Differences in demographic and clinical presentation of CNS lymphoma have been reported by previously published studies from the Indian subcontinent compared with the Western studies. The current study, one of the largest series from the Indian subcontinent, confirms these findings and also makes some fresh observations. Patients in our population were a decade younger than that reported from Western studies (55 years). ,, Similar observations were made in previous studies from India. ,,, Many western studies have noted cognitive decline/behavioral changes as the common presenting symptom of PCNSL with a reported incidence of 73%. ,,, However, we noted that only 35% of patients had cognitive decline, which is similar to other Indian studies reporting a lower incidence of 12.5-23% cognitive symptoms. , Lack of a formal neuropsychological testing combined with some degree of under reporting by patient's relatives may be the cause for this observed difference.
There was a progressive increase in the hospital-based incidence of CNS lymphoma during the study period. Percentage of lymphomas in all biopsied lesions showed a gradual increase from 0.18% at the beginning of the study period to 0.41% at the end of the study period in our institute. This is similar to the epidemiological trends reported in western studies. , Interestingly, previous Indian studies failed to show any significant increase in incidence and opined that the increase in the incidence of CNS lymphomas over the last two decades was not a universal phenomenon. , Inability to demonstrate a statistically significant rise may be due to the smaller cohorts compared with western studies.
HIV positive patients carry a 3-9% lifetime risk of developing CNS lymphomas and this risk increases with longevity. , HIV positive patients with CNS lymphomas differ in clinical and radiological characteristics and prognosis, from immune competent CNS lymphoma patients. , The present study noted that HIV was associated in 8.6% of lymphoma patients. This is in line with the low incidence of association of CNS lymphomas in the HIV infected population in India noted in other Indian studies wherein only 0-8% cases of CNS lymphomas are HIV positive. ,, It reaffirms the observation, that increase in incidence of CNS lymphomas is independent of an increase in number of HIV positive patients in our population. The present study documented that the mean age of lymphoma patients with HIV was significantly less than those without HIV (P = 0.033).
Since surgical decompression does not improve survival in patients with CNS lymphoma,  STB is considered the investigation of choice. We noted that more number of patients in our study have undergone STB in the second half, which can be attributed to better MRI imaging and radiological diagnosis and increasingly easier access to guided stereotactic biopsies in cases of diagnostic difficulties. Inconclusive biopsy following steroid treatment in CNS lymphoma is a frequent observation. , Reactive gliosis with variable infiltrates of "B" and "T" lymphocytes and macrophages or a complete absence of tumor cell in routine histopathological examination and immunohistochemistry is reported in lymphomas following steroid treatment. , Selective toxicity of glucocorticoids to lymphoma cells and resultant rapid shrinkage has been postulated to be the cause of this finding. Choi et al., in their review of four cases of CNS lymphoma, where biopsy following steroid treatment was inconclusive found that molecular genetic methods may be useful in establishing a diagnosis in such cases.  Inconclusive biopsy is a known limitation of STB of any intracranial lesion. ,, However, no study has assessed the frequency of negative biopsy following steroid treatment, to the best of our knowledge. The current study documented 88.6% diagnostic rate of STB, close to the accuracy rate reported in other studies. ,, But patients who received steroids for long duration had 57.1% incidence of negative STB which is highly significant. Shorter duration of steroid therapy also resulted in negative biopsy, to a lesser extent (33.3%). We therefore suggest that whenever lymphoma is suspected on pre-operative imaging, steroids should be avoided prior to obtaining a tissue diagnosis as it decreases the diagnostic yield.
B-cell non-Hodgkin lymphomas constitute 92-98% of PCNSL, of which 95% are diffuse large B cell lymphoma (DLBCL). , All morphological variants of DLBCL such as ALK positive large B cell lymphoma, anaplastic variant of DLBCL, immunoblastic variant of DLBCL, Microvillous large B cell lymphoma and T cell/histiocyte rich B cell lymphoma may be found in the CNS. ,, However, in the current study, DLBCL was the only subtype.
The present study describes the peculiarities and patterns of CNS lymphoma in Indian population. A progressive increase in hospital-based incidence interestingly in immunocompetent individuals was documented. We identified that steroid treatment for more than a week significantly resulted in a higher rate of negative STB.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]
[Table 1], [Table 2], [Table 3]