Late-delayed cerebral involvement in systemic non-Hodgkin lymphoma

A second primary tumor or a tardy recurrence?




Central nervous system involvement is a well recognized complication of systemic non-Hodgkin lymphoma. Most central nervous system recurrences occur within the first 2 years after the initial diagnosis and are considered to represent clonally related recurrence of systemic disease. The authors attempted to investigate the clonal relation between the late-delayed central nervous system involvement and the original systemic tumor.


The authors studied archival, formalin fixed, paraffin embedded tissue samples from 8 patients with isolated cerebral involvement diagnosed > 3 years after their initial presentation with aggressive, systemic, B-cell non-Hodgkin lymphoma. The rearranged immunoglobulin heavy-chain variable region genes (VH) from both sites were amplified by polymerase chain reaction and were sequenced when necessary.


In three of five patients who had interpretable results, a distinct, monoclonal, VH family-specific band profile was obtained from the cerebral and systemic lymphoma. In the other two patients, a similar VH band pattern was observed and also was compared using direct sequencing, which demonstrated sequence differences between tumors from the two sites.


Clonal variance between the cerebral and systemic lymphoma in these patients suggested the possibility that some instances of late-delayed recurrence in the central nervous system represent a second, new B-cell lymphoma rather than a true recurrence of the original systemic tumor, a finding that may have significant clinical and biologic implications. Cancer 2004. © 2004 American Cancer Society.

Central nervous system (CNS) involvement is a well recognized extranodal complication of systemic non-Hodgkin lymphoma (NHL) with reported risks 5% in indolent variants and 30% in aggressive variants.1 Most CNS recurrences tend to occur in the first 14 months after the initial diagnosis, with only sporadic events encountered thereafter.2–4 Recent advances in the management of systemic NHL have produced a high rate of complete response (CR) and a long-term survival. Still, some patients remain at continued risk for a late systemic recurrence after a sustained remission of 24 months5, 6 and even as late as 26 years.7 Late-delayed involvement of the CNS also has been reported,6 but it is unknown whether it represents a true delayed recurrence of the original tumor or rather an emergence of a second new lymphoma.

Evaluation of the lymphoid origin and clonality of malignant and nonmalignant B-cell disorders is now possible by molecular analysis of the immunoglobulin heavy chain gene (IgH). During B-cell development, the repertoire of IgH diversity is generated by combinatorial permutation of germline variable (VH), diversity (DH), and joining (JH) gene segments, by addition or deletion of short coding sequences at VD and DJ joints, and by somatic hypermutation.8 Because malignant transformation usually involves expansion of a single progenitor cell, it yields a cell population that is arrested at a particular stage of differentiation and results in a proliferation of cells manifesting identical, clonal-specific pattern of V(D)J rearrangement forming the complementary determining region 3 (CDR3). Confirmation of a true recurrence, therefore, is possible by documenting the presence of a common clonal B-cell origin at the initial diagnosis and at recurrence. Conversely, different clonality may suggest the emergence of a new, clonally unrelated neoplasm. We used this strategy in patients with isolated cerebral lymphoma diagnosed > 3 years after their initial presentation with systemic B-cell NHL and analyzed the clonality of tumors from both sites by comparing their immunoglobulin VH gene rearrangement with a polymerase chain reaction (PCR)-based technique and direct sequencing.



Eight patients with a presumed, isolated cerebral recurrence of B-cell NHL that developed > 3 years after a sustained remission of their systemic disease were included in this study (Table 1). All 8 patients were treated at our Neuro-Oncology Center between 1990 and 1999, and 5 of them were referred from other hospitals.

Table 1. Clinical Details and Results of Variable Region, Family-Specific Immunoglobulin Heavy Chain Gene Rearrangement Analysis in Patients with Non-Hodgkin Lymphoma at Systemic and Cerebral Involvement
PatientAge (yrs)GenderSystemic lymphomaCerebral lymphoma
Initial siteHistologyStageaVH familyTime to recurrence (mos)bHistologyVH family
  • VH family: the rearranged immunoglobulin heavy chain variable region gene family; M: male; F: female.

  • a

    Clinical disease stage was determined according to the Ann Arbor system (see Carbone et al., 197131).

  • b

    The time to recurrence means the time between the initial diagnosis of systemic disease and the diagnosis of cerebral involvement.

146MBowelDiffuse largeIVH140Diffuse largeVH4
272MMaxillary sinusDiffuse largeIVH546Diffuse largeVH5
336FLymph nodesDiffuse largeIIIUnavailable58BurkittUninterpretable
439FKidney, boneDiffuse largeIVVH355Diffuse largeVH1
539MOrbit, lymph nodesMantleIIIVH3142Diffuse largeVH6
650FSoft tissueDiffuse largeIVH288Diffuse largeVH2
760MLymph nodeDiffuse largeIUninterpretable96Diffuse largeUninterpretable
873MTestisDiffuse largeIUninterpretable53Diffuse largeUninterpretable

The mean patient age at diagnosis of systemic lymphoma was 52 years (range, 36–73 years). Results of the CNS evaluation performed at initial staging were available in six patients. All six patients had negative cerebrospinal fluid (CSF) cytology, and three patients had unremarkable brain computed tomography (CT) scans. Systemic therapy included conventional combination chemotherapy regimens, and one patient underwent autologous bone marrow transplantation and received immunotherapy. Two patients received local irradiation, one to the maxillary sinus and the other to the orbit. The patient with a maxillary sinus lymphoma received prophylactic intra-CSF chemotherapy. Systemic therapy induced a CR in all 8 patients, but 2 patients subsequently developed recurrences at 10 months and at 1 month (Patients 3 and 5, respectively). Patient 3 had systemic recurrence and achieved a second CR after high-dose combination chemotherapy with autologous bone marrow transplantation, immunotherapy, and radiotherapy. Patient 5 developed isolated leptomeningeal seeding, which was treated with radiation therapy and intra-CSF methotrexate (MTX) injections through an implanted Ommaya device that brought about clinical improvement and clearing of malignant cells from the CSF. However, he had an asymptomatic recurrence 21 months later with a nearly continuous, asymptomatic, positive CSF cytology since then complicated by bilateral lymphomatous vitreitis. He was treated with systemic, intra-CSF, and intravitreous chemotherapy; bilateral vitrectomy; and total neuraxis irradiation that intermittently cleared the CSF of malignant cells. Eventually, after > 11 years, he developed parenchymal brain lymphoma. Because at no time after the initial diagnosis was there systemic involvement, and because his disease behaved like a primary CNS lymphoma, he was included in the study.

At the time they were diagnosed with cerebral lymphoma, all patients underwent reevaluation for evidence of possible systemic recurrence. This consisted of a serum lactate dehydrogenase level, whole-body CT scan, gallium scintigram, and bone marrow biopsy. Additional studies included whole-neuraxis (six patients) or cranial (two patients) magnetic resonance imaging, repeat CSF analysis (eight patients), and slit-lamp examination for possible lymphomatous vitreitis (eight patients). Response to treatment was defined according to the integrated results of clinical, neuroimaging, and CSF studies.9

The median time to recurrence in the CNS was 57 months (range, 40–142 months) from the diagnosis of systemic lymphoma. All eight patients had parenchymal cerebral involvement. Multifocal brain lesions were present in three patients, leptomeningeal seeding was present in two patients, and one patient had new intraocular involvement. All patients were seronegative for the human immunodeficiency virus, and none had evidence of systemic disease. Treatment varied but was based primarily on regimens with systemic, high-dose MTX combination chemotherapy and intra-CSF cytosine arabinoside delivered through an Ommaya device. This was followed by cranial irradiation once resistant disease was present.9 Regional application of chemotherapy by way of intraarterial injection with or without osmotic blood-brain barrier disruption became available in our center in 1996 and was given to 4 patients. Overall, five patients responded to treatment: three patients achieved a CR, two patients had a partial response, and three patients failed to respond. All 5 responding patients succumbed to a new recurrence 1–19 months after withdrawal of treatment. The median survival from diagnosis of cerebral involvement was 15 months (range, 4–40 months).

Pathologic Studies

Twelve tissue specimens were retrieved from the pathology archives of our institution, and 3 specimens kindly were provided by the referring hospitals. All tissues were obtained at diagnostic surgical intervention and were fixed in 10% neutral formalin, routinely processed, and embedded in paraffin. All of the histologic slides and immunohistochemical stains were reviewed by one pathologist (G.A.) and, if they proved inadequate, were repeated. CD20 and CD3 were assessed in all of the specimens. Additional stains included cyclin D1, CD79α, LN-1 and LN-2 (for B cells), and UCHL-1 and CD4 (for T cells).

VH Family-Specific PCR Analysis

We used formalin fixed and paraffin embedded tissues. Samples were digested by proteinase K, and DNA was extracted by using a QIAamp DNA blood minikit (QIAGEN GmbH, Hilden, Germany). The DNA was amplified using specific sense primers corresponding to 1 of the 6 conserved areas at the 5′-end of the leader sequence for each of the 6 human immunoglobulin VH gene families (VH1–VH6) and an antisense JH consensus primer.10 Due to their family specificity, these primers allow easy identification of the VH gene family usage in a given B-cell clone. The PCR conditions were as follows: 10 mM Tris-HCl, 50 mM KCl, 1.5 mM MgCl2, 0.1% Triton X-100, 0.2 mg/mL bovine serum albumin or gelatin, 0.2 mM of each dinucleotide triphosphate, and 1.3 U of Taq polymerase (Qbiogene, Illkirch, France) in a final volume of 50 μL. The primers were added to a final concentration of 1 μM each. After denaturation at 95 °C for 5 minutes, Taq polymerase was added at 86 °C. Fifty cycles of amplification were performed under the following conditions: 90 °C for 1 minute and either 61.5 °C for 1 minute (VH1 and VH4), or 58 °C for 1 minute (VH5 and VH6), or 62 °C for 1 minute (VH3); 72 °C for 1 minute. PCR products were resolved on 2% agarose gels. The expected size of the PCR product was ≈ 350 base pairs. Samples were analyzed in duplicate. Peripheral blood from healthy individuals and bone marrow tissue from patients who had NHL with a known rearrangement served as a negative and positive controls, respectively.

Sequencing of PCR Products

To obtain sufficient amounts of PCR products for the sequencing reaction, two rounds of amplification were performed. The design of universal primers was based on the multiple sequence alignment of the VH and JH regions that were obtained from the International Immunogenetics Database ( The two sets of degenerate primers were as follows: VH-ext 5′-GGGAAGGSSCTKGAGTGG-3′ (forward), JH-ext 5′-GACTCACCTGAGGAGACRGTGA-3′ (reverse), and VH-int 5′-ACGGGCYGYGTATTACTGT-3′ (forward); JH-int 5′-CCTGAGGAGACGGTGACC-3′ (reverse). The reagents of the nested PCR mixture are described above, except for the template of the second round, which was the amplification product of the first round diluted 1:10. The thermal cycling for both rounds was carried out by denaturation at 95 °C for 40 seconds, annealing at 55 °C for 40 seconds, and extension at 72 °C for 30 seconds. There were 20 cycles for the first round and 40 cycles for the second round. Each DNA sample was amplified and sequenced in two independent reactions to confirm reproducibility of the results. The specific amplification products were excised after resolving on 4% low-melting agarose. After purifying the DNA with the QIAquik gel extraction kit (QIAGEN GmbH), the products were sequenced directly with an antisense JH sequence primer (5′-ACCTGAGGAGACGGTGACCAGGGT-3′).


Pathologic Analysis

The systemic B-cell lymphoma was extranodal in six patients and nodal in the other two patients. Extranodal tissue specimens included bowel, maxillary sinus, femur, and testes. Seven of the eight systemic lymphomas were diffuse large cell lymphoma, and one was mantle cell lymphoma (Table 1). The histology of the cerebral lymphoma was diffuse large cell lymphoma in seven of eight patients, including the one patient with systemic mantle cell lymphoma (Table 1). The eighth patient had a Burkitt lymphoma of the brain, although her prior lymph node lymphoma was diffuse large cell lymphoma.

Molecular Analysis

One tissue sample from the initial diagnosis of systemic lymphoma could not be recovered. Five other samples failed control amplification and were considered uninterpretable. This most likely was the result of degradation of the DNA and often is observed in formalin fixed and paraffin embedded tissues.11 Amplification of the genomic DNA in the remaining 10 samples was performed in 6 separate reactions using primers that were specific for the leader sequence of each VH1 to VH6 family and a JH consensus primer. For all 10 samples, 1 dominant monoclonal PCR product was obtained, indicating absence of contamination by nonmalignant polyclonal B-cells (Table 1). A diverse clonal origin of the systemic tumor and the cerebral lymphoma was suggested in 3 patients, in whom a different VH band profile was recorded from the two sites of the disease (Fig. 1). Identical VH band patterns were obtained from both the systemic disease and the brain tumor in two other patients. This finding was not sufficient to confirm identical rearrangement due to the presence of multiple alleles for each VH family. To verify the clonal origin of these tumors, we analyzed them by direct sequencing.12, 13 The CDR3 sequences for each VH gene and the JH allele of tumors at the 2 sites revealed a diverse clonal origin despite the use of the same VH family (Fig. 2).

Figure 1.

Histologic findings and agarose-gel analysis of gene variable region immunoglobulin heavy chain (IgH) (VH) family-specific rearrangements detected by polymerase chain reaction analysis in Patient 1 (A), Patient 4 (B), and Patient 5 (C) from systemic and cerebral non-Hodgkin lymphoma. (A) Photomicrographs show diffuse large cell lymphoma of the ileum infiltrating the mesenteric fat (left; original magnification, × 400) and diffuse large cell lymphoma of the brain (right; original magnification, × 400). (B) Photomicrographs show diffuse large cell lymphoma of the femur infiltrating between fibers of striated muscle (left; original magnification, × 400) and diffuse large cell lymphoma of the brain (right; original magnification, × 400). (C) Photomicrographs show mantle cell lymphoma of a lymph node partially preserving the germinal center (left; original magnification, × 200) and diffuse large cell lymphoma of the brain (right; original magnification, × 400). All pathologic specimens were stained with hematoxylin and eosin.

Figure 2.

Comparative sequence alignment of the amplified polymerase chain reaction products from systemic and cerebral non-Hodgkin lymphoma in Patients 2 and 6. The sequences of the degenerate, immunoglobulin heavy chain genes (IgH) variable region (VH)-int primers are shown in boldface. The different rearranged IgH joining region (JH) gene alleles are indicated on the right of the alignments, and their sequence is underlined. Double dots indicate sequence identity.


In this study, we used comparative PCR-based molecular analysis of the immunoglobulin VH region in order to determine the clonal relation between the original systemic B-cell NHL and the late-delayed occurrence of cerebral lymphoma. The divergent VH families encountered in 3 of the 5 patients with interpretable results and the distinct CDR3 sequences in the other 2 patients indicate clonal variance between tumors from the 2 sites.

Late-delayed involvement of the CNS that develops after a prolonged remission of systemic NHL may derive from two distinct pathogenic pathways that have been described previously in systemic B-cell neoplasms.5, 6 First, it may represent a true recurrence of the original tumor. In that case, tumor cells may have resided dormant in an immunologically privileged sanctuary site after escaping therapy or may have invaded the CNS after the completion of therapy, while the systemic disease remained minimal or undetectable. In the second pathway, late-delayed CNS involvement represents a new neoplasm that develops in the setting of an inherent immunologic and genetic predisposition or secondary to prior therapy or an infectious agent.14–16 Our findings of disparity in the immunoglobulin VH gene between the original tumor and the recurrent neoplasm suggest that, in late-delayed CNS recurrences, the evolution of a new neoplasm may have taken place. Still, the existence of a biclonal population in the original tumor cannot be excluded. In this case, one clone obviously dominated the initial presentation but may have been eliminated after treatment; and the second, minor clone, which was not detected by our analysis of the systemic lymphoma, may be responsible for the delayed cerebral involvement.17 It is possible that the experience of Patient 5 may represent such an event, because his disease behaved more like a primary CNS lymphoma, inducing lymphomatous vitreitis and parenchymal mass, rather than a typical recurrence of systemic lymphoma. However, the first manifestation of his CNS involvement occurred early after the completion of systemic chemotherapy and remained resistant to therapies, unlike his systemic lymphoma. In this patient, a biclonal population is an attractive possible explanation.

Molecular analysis of clonality has been used previously to study recurrences of systemic B-cell lymphoma17–22 and leukemia,23 usually indicating recurrence of the original tumor. Identification of a second new neoplasm is less common and is reported in isolated patients with diffuse large cell lymphoma, mantle cell lymphoma, and follicular lymphoma17, 22 and in patients with mantle cell lymphoma, chronic lymphocytic leukemia, and follicular lymphoma transformed into Hodgkin disease or Burkitt lymphoma after autotransplantation.24, 25 Those works studied patients with systemic recurrence, and there are only two previous reports of recurrence in the CNS evaluated for its clonal origin. One of those reports briefly mentions a patient who developed brain involvement 66 months after his testicular diffuse large cell lymphoma was treated with chemotherapy and radiotherapy.26 It was concluded that both sites were related clonally by showing identical V(D)J complexes on sequencing the IgH gene. The other patient presented with leptomeningeal involvement 6 years after treatment for systemic mantle cell lymphoma with autotransplantation.27 He maintained clinical and molecular systemic remission. A common clonality of the leptomeningeal and systemic tumor was verified by identification of the same VH3 family and was confirmed by sequencing. Taken together, our findings indicate phenotypic heterogeneity of late-delayed CNS involvement in systemic NHL that may be responsible for different biologic and clinical behavior.

Because late and isolated CNS recurrences of systemic NHL are rare,28, 29 it is difficult to draw conclusions related to their clonal origin from the clinical and histologic characteristics alone. Molecular diagnosis can answer this question and may have important therapeutic and prognostic implications. A recent work evaluated PCR analysis for the identification of IgH rearrangement on formalin fixed, paraffin embedded biopsy material from primary CNS lymphoma.30 That study showed an overall monoclonal pattern in 77% of patients with lymphoma, similar to our findings. Those results, viewed in light of our findings, make it likely that, in the future, the DNA monoclonality of brain involvement will be identified on biopsy material. This will allow verification of whether the selection and development of new malignant clones occur in late recurring disease. It also may enable appropriate selection of therapeutic strategies based on the stratification of CNS involvement either as a new primary lymphoma or as a delayed recurrence of the systemic disease.