Background
Diffuse large B-cell lymphoma (DLBCL) is the most common lymphoma subtype and accounts for 30–40% of adult non-Hodgkin lymphoma (NHL) [
1]. Primary central nervous system (CNS) DLBCL is defined as DLBCL arising from the brain, spinal cord, leptomeninges and eye. Primary CNS DLBCL is a rare subtype of B-cell lymphoma, which represents less than 1% of all NHL and 2.4–3% of all brain tumors [
2], and is classified as a distinct entity in the WHO classification of tumors in hematopoietic and lymphoid tissues [
1]. Statistics shows a peak incidence of patients aged 50 to 60 years (median 56 years old); however, in the recent two decades, there is an increasing incidence of reported patients over the age of 60 years old with primary CNS DLBCL [
3]. As it is difficult to make an early diagnosis and because targeted therapy is lacking, primary CNS DLBCL usually shows poorer prognosis than extraneural systemic DLBCL [
3].
DLBCL, not otherwise specified (DLBCL, NOS) is morphologically diverse and is generally divided into 3 morphological variants: centroblastic, immunoblastic, and anaplastic [
1]. The anaplastic variant of DLBCL (A-DLBCL) is an uncommon morphologic variant that represents approximately 3.4% of all DLBCL cases [
4,
5] and is characterized by large to very large polygonal cells with bizarre pleomorphic nuclei that resemble tumor cells of anaplastic large cell lymphoma (ALCL) [
1]. Recently, we reported that A-DLBCL had a high frequency of
TP53 mutations and concurrent
MYC and
BCL2 and/or
BCL6 genetic abnormalities, indicating that the clinicopathologic features and aggressive behavior of A-DLBCL are distinct from the common DLBCL [
6].
Because of the rarity, to our knowledge, there are only two case reports describing primary CNS A-DLBCL in the literature [
7,
8]. In this study, we reported on 3 patients with primary CNS A-DLBCL and studied their clinicopathologic and genetic features to provide further information for diagnostic and prognostic assessments as well as treatments of this distinctive type of DLBCL.
Discussion
Primary CNS DLBCL is a rare entity with poor prognosis and needs a better understanding of the genetic characteristics and prognostic markers [
13]. In terms of morphology, primary CNS A-DLBCL is an extremely uncommon lymphoma that has been described in only two other case reports to date but without intact clinicopathologic characteristics [
7,
8]. In the 3 cases of this study, the characteristic features of perivascular space infiltration with bizarre-shaped tumor cells led to the diagnosis of primary CNS A-DLBCL after excluding systemic DLBCL with CNS involvement. It is important to realize the existence of primary CNS A-DLBCL to avoid being misdiagnosed as ALCL or undifferentiated carcinoma. The former is positive for CD30 and/or ALK and negative for B-cell markers, whereas primary CNS A-DLBCL expresses B-cell markers with negativity for ALK. Negativity for cytokeratin is helpful in distinguishing this type of DLBCL from undifferentiated carcinoma [
14].
In our previous study of 35 patients with A-DLBCL, we defined A-DLBCL based on morphology and recognized the distinctiveness of this neoplasm from ordinary DLBCLs in terms of genetic alterations and biologic features, which contained a high incidence of p53 positivity and MYC/BCL2 double-expressor and a high frequency of
TP53 mutations and concurrent
MYC and
BCL2 and/or
BCL6 abnormalities [
6]. In the 3 patients with primary CNS A-DLBCL in this study, the concurrent of
MYC and
BCL2 and/or
BCL6 abnormalities and MYC/BCL2 double-expressor occurred in all 3 patients, in which two patients had
MYC/
BCL2/
BCL6 triple extra copies, and one patient had
MYC extra copy and
BCL6 translocation. The pattern of genetic abnormalities of these genes was overwhelmingly having an extra copy except
BCL6 translocation in case 3, which is consistent with translocations recurrently affecting
BCL6 and
IG genes, whereas
MYC translocations are rare and
BCL2 translocations are absent in primary CNS DLBCL [
15‐
17]. The prognostic value of gain/amplification of
MYC is still unclear in DLBCL, and some studies reported that gain/amplification of
MYC is not associated with a poor prognosis [
18‐
20]. Moreover,
BCL2 gain/amplification is associated with cell-of-origin–specific (activated B-cell-like subtype) clinical effect in R-CHOP-treated DLBCL [
21]. In our previous study about A-DLBCL [
6],
MYC/BCL2/BCL6 triple extra copies were found in 3 cases, and the overall survival is 5, 9, and 11 months, similar to the very aggressive disease course and poor prognosis of primary CNS A-DLBCL cases in this study, suggesting that in this specific morphology (anaplastic variant),
MYC/BCL2/BCL6 copy number alterations may be an adverse prognostic factor, but more cases are needed to confirm this finding.
We detected one patient with diffuse positive p53 immunostaining and
TP53 gene mutation (R273C). The
TP53 mutation rate is approximately 20% in DLBCL [
22] and approximately 26.7–37.2% [
23,
24] in primary CNS DLBCL. However, few studies that report
TP53 mutation status in PCNSL reveal incidences < 10% [
25,
26]. Moreover, we also found one patient had a
TP53 null phenotype (completely negative), similar to that seen in high grade serous carcinomas [
27]. Tumors that were completely negative for p53 IHC expression had a mutation of
TP53 in 65% of cases and wild-type
TP53 in 11%. Therefore, complete absence of p53 immunostaining is commonly associated with a
TP53 mutation status. However, the
TP53 exon 5–10 of this patient was wild type by sanger sequencing, suggesting that there may be a wild-type
TP53 or variations in other positions besides exon 5–10
. TP53 mutations alter the wild-type p53 protein structure and disrupt its function, which is implicated in lymphomagenesis and disease progression [
28]. The genomic instability driven by the
TP53 mutation and
MYC,
BCL2, and
BCL6 abnormalities might partially explain the poor clinical course and may be good prognostic factors in patients with primary CNS A-DLBCL. It seems that these genetic alterations probably occurred more frequently in primary CNS A-DLBCL patients than in patients with ordinary primary CNS DLBCL without anaplastic features, suggesting that the morphology is closely related to genetic alterations and disease prognosis. However, as there are rare reports of primary CNS A-DLBCL, it is necessary to confirm our observations with a large number of cases in the future.
In primary CNS DLBCL, mutations of
MYD88 are believed to promote lymphomagenesis via constitutive activation of the NF-κB pathway. MYD88 is an adaptor molecule in the Toll-like receptor pathway that mediates interleukin-1 receptor signaling [
29]. A somatic mutation of
MYD88 is found in 67–86% [
26] [
30,
31] of primary CNS DLBCL, and the most common mutation site is L265P. In this study, all 3 primary CNS A-DLBCL patients displayed mutations in
MYD88 L265P, which is consistent with the high frequency of
MYD88 alterations reported in the literature for primary CNS DLBCL. The high frequency of
MYD88 mutations in primary CNS A-DLBCL are probably not unique to the anaplastic variant, but mainly due to the site of CNS. Because the frequency of
MYD88 mutation appeared to be higher in this small series than that of A-DLBCL in our previous study (100% vs 20%) [
6]. The activation of the NF-κB pathway leads to the nuclear translocation of NF-κB dimers and subsequently initiates activation of other target genes [
32]. In all 3 cases in this study, tumor cells expressed at least 2 nuclear NF-κB subunits, suggesting that NF-κB signaling activation is an almost universal feature of primary CNS A-DLBCL. Given the high prevalence of
MYD88 mutations in primary CNS A-DLBCL, treatment with ibrutinib, which inhibits Bruton’s Tyrosine Kinase (BTK) and further suppresses NF-κB and STAT3 activation and tumor growth, or treatment with bortezomib, which is effective for
CD79B/
MYD88L265P double-mutant DLBCLs [
33], could be considered in patients with primary CNS A-DLBCL [
34].
Current treatment for primary CNS DLBCL includes surgery followed by chemotherapy with methotrexate-based regimens, with or without adjuvant radiation therapy [
3,
13]. Unlike other brain tumors, primary CNS DLBCL regularly has a good response to chemotherapy and radiation therapy, but in comparison with lymphomas outside the CNS, survival is not as prominent, partly due to inadequate penetration of the blood-brain barrier [
35]. Most protocols report a median progression-free survival of approximately 12 months and an overall survival (OS) of approximately 3 years. However, the prognosis for primary CNS DLBCL that has failed first-line therapy remains poor, with a low median OS of 12 months [
13]. In the 3 patients with primary CNS A-DLBCL in this study, the median OS was only 5 months, which is much lower than that of primary CNS DLBCL and A-DLBCL (16 months) [
6]. The International Prognostic Index (IPI) score used for determining the prognosis of patients with aggressive NHL seems not fit into primary CNS DLBCL. The IELSG score, including five parameters (age, ECOG performance status, LDH level, CSF protein concentration, deep brain involvement), is widely used to predict outcomes and to better stratify patients. The presence of 0 to 1, 2 to 3, or 4 to 5 adverse risk factors correlates with 2-year survival rates of 80, 48%, or 15%, respectively [
9]. In our 3 cases, the scores of patients were all 4, suggesting an aggressive disease course and poor survival, in accordance with intricate genetic alterations and adverse prognosis factors.