Professional Documents
Culture Documents
1136/jclinpath-2016-204294
Original article
This retrospective review of tumours with plasmablastic available morphological and clinical features were not sufficient
morphology aims to evaluate the usefulness of various para- for definitive classification as either PBL or PBM (classified as
meters that differentiate PBL and PBM. In order to do so, we ‘indeterminate’). Cases classified as PBL received lymphoma
classified cases with plasmablastic features as lymphoma or regimens, and the cases classified as PBM received therapy for
myeloma based on review of all pertinent histopathological and myeloma, indicating 100% concordance between the study clas-
clinical materials. Based on the lessons learned from this review, sification and the original clinical diagnosis. The clinical and
we propose a diagnostic algorithm for the classification of neo- morphological features of these 11 patients are described below
plasms with plasmablastic features. within the Results section, and summarised in tables 1 and 2.
Lessons learned from this review were used to design a diagnos-
MATERIALS AND METHODS tic algorithm depicted in figure 2.
We retrospectively searched the electronic medical records for
all pathology reports containing the term ‘plasmablastic’ from Plasmablastic lymphoma
January 2000 through January 2015. Only cases with available Of the eight PBL-classified patients, three were HIV-positive
glass slides were included. The pre-existing H&E-stained tissue (patients #2, #4 and #8). Of these, two ( patients #2 and #4)
sections were reviewed by three haematopathologists (MS, JAV had positive EBER in tissue from the oral cavity and oesopha-
and FGR) for morphological categorisation into plasmablastic gus, respectively. Although patient #8 had high EBV viraemia
(>60% plasmablastic cells), plasmacytic (<40% plasmablastic by PCR (10 314 copies/mL), EBER was negative. Of the remain-
cells) or mixed (approximate equal numbers of plasmablastic/ ing patients, four were HIV-negative ( patients #1, #3, #5 and
plasmacytic cells 40%–60%) groups (figure 1). We then #6) and one had an unknown HIV status ( patient #7). Of the
reviewed the available ancillary studies including flow cytometry, HIV-negative cases, one was EBER-positive and three were
immunohistochemistry, in situ hybridisation and cytogenetics. EBER-negative. The single HIV-negative EBER-positive case
Clinical data were extracted from electronic medical records by (patient #1) was of a male aged 42 years with end-stage liver
independent study personnel. It comprised demographic infor- disease presenting with ascites and plasmablastic cells in the
mation, results of physical examination, pertinent radiological peritoneal fluid. A negative HHV-8 excluded primary effusion
findings and laboratory values. lymphoma in this patient. The three EBER-negative
Using the current knowledge of plasmablastic neoplasms pre- HIV-negative cases ( patients #3, #5 and #6) were from a
viously reported,2 7 10 12 14 we classified cases into ‘PBL’, ‘PBM’ female aged 61 years with isolated posterior nasopharyngeal
or ‘indeterminate’ categories. An unbiased and independent soft tissue mass, and from a female aged 58 years and a male
classification was assured by having different study members aged 74 years, both presenting with diffuse lymphadenopathy.
extract data from medical records and classify the cases. Cases Negative HHV-8 and/or ALK immunostains excluded the diag-
categorised as ‘indeterminate’ were those in which clinical nosis of an HHV-8-associated neoplasm and/or ALK-positive
and/or histopathological data were insufficient for confident large B-cell lymphoma in patients #3 and #6, respectively.
classification. We then compared the classification results with Concomitant bone marrow involvement was seen in four
the documented clinical diagnosis. This study has been approved cases ( patients #3, #6, #7 and #8). In these patients, the pres-
by the institutional review board. ence of diffuse hypermetabolic lymphadenopathy and/or extra-
medullary presentation led to the classification of these cases as
RESULTS PBL, despite the presence of at least one myeloma-defining sign
A total of 11 cases were included. Of these, eight (73%) were in four cases (table 1). For patient #7, whose HIV status,
classified as PBL and two (18%) as PBM. In one case (9%), the history of immunosuppression and presence/absence of
Table 1 Plasma cell neoplasms: clinical characteristics and Epstein-Barr virus (EBV) status
Age BM Free Bone Outcome
Case# Sex HIV EBER Site (%) LN LC M-spike CRAB Hb lesions Cr Ca Therapy (months)
Plasmablastic lymphoma
1 42 − + Ascites NA NA NA NA ± 8.4 − 1.15 7.9 Supportive Dead (<1)
M
2 35 + + Oral NA + − NA − 14.8 − 1.1 9.6 DA EPOCH Rx Alive (117)
M
3 74 − − Lymph node 40% + 17.2 + ± 10.6 − 0.9 9.5 EPOCH Alive (23)
M
4 49 F + + Oesophagus − + NA NA ± 8.9 − 0.71 8.6 DA-EPOCH Alive (15)
5 61 F − − Nasopharynx − − 0.8 − − 13.7 − 0.85 9.3 DA-EPOCH Alive (15)
Rx
6 58 F − − Lymph node >90% + 2.8 − ± 8.6 + 0.91 9.9 CHOP Dead (10)
7 54 NA + Testicle 30% + NA NA NA 15.4 NA NA NA Hyper-CVAD Dead (13)
M
8 55 + − Bone 80% + 3.45 + ± 8.5 + 0.7 8.1 EPOCH Dead (4)
M marrow
Plasmablastic myeloma
9 71 NA NA Bone >90% − 5.7 + + 10 + 5.8 12 DVD Dead (17)
M marrow
10 72 F NA NA Bone 80% − NA + ± 7.0 + 0.8 9.1 Radiation, Thal Dead (3)
marrow DA
Indeterminate
11 79 NA − Mandible NA NA NA NA NA NA NA NA NA NA NA
M
At time of presentation, the patients had incomplete CRAB signs, with normal creatinine levels (0.8 mg/dL).
BM (%), percentage of involvement at time of presentation; Ca, serum calcium, 8.9–10.1 mg/dL; CHOP, cyclophosphamide, doxorubicin, vincristine, prednisone; Cr, creatinine, reference
0.4–1.2 mg/dL; CRAB, myeloma-defining signs (+: complete; −: absent; ±: incomplete); DA, dexamethasone; DVD,doxorubicin, vincristine, dexamethasone; EBER, Epstein-Barr
virus-encoded RNA; EPOCH, etoposide, prednisone, vincristine, cyclophosphamide, doxorubicin; F, female; free LC, free light chain ratio, 0.26–1.65 mg/dL; Hb, haemoglobin, reference
12.5–16.3 (males) and 11.2–15.2 g/dL (females); hyper-CVAD, alternating doses of cyclophosphamide, vincristine, doxorubicin and dexamethasone with high-dose methotrexate and
cytarabine; LN, significant lymphadenopathy; M, male; M-spike, monoclonal immunoglobulin spike by serum and/or urine protein electrophoresis, or immunofixation electrophoresis; NA,
not available or performed; outcome, months of available follow-up data; Rx, radiation; Thal, thalidomide.
Plasmablastic lymphoma
1 PC − − − + + − − − − − NA Kappa NA NA >95%
2 PC − − NA NA + NA NA − NA Weak NA Lambda NA NL >95%
3 Mixed + − − + Var − − − − Var+ NA Kappa NA NL 80%
4 PB NA − NA NA + NA NA NA NA NA NA Lambda NA NL >95%
5 Mixed NA − − NA + NA NA NA NA NA NA Kappa NA NL 80%
6 PB Var+ − − + +† NA −* + − − − Kappa − NL 75%
7 PB − − NA NA + NA NA NA Weak + − NA Lambda −* NA >95%
8 Mixed NA − NA NA + NA NA NA + NA NA Kappa +* C >90%
Plasmablastic plasma cell myeloma
9 PB − − NA NA + NA NA NA NA − NA Lambda NA NL‡ NA
10 PC − − NA NA + NA NA NA NA NA NA Kappa NA NA 80%
Indeterminate
11 PC Weak+ − NA NA + NA − − NA + NA Lambda NA NA NA
All were immunohistochemical studies.
*Fluorescence in situ hybridisation, HHV-8, Var.
†Weakly positive.
‡One of 13 cells cultured had trisomy 11 and additional structural defects.
ALK, anaplastic lymphoma Kinase; C, complex cytogenetics; gene, genetics; HHV-8, human herpes virus-8; morph, morphology based on predominant cells type; NL, normal; PB,
plasmablastic; PC, plasmacytic; var, variable.
lymphadenopathy or tissue-based disease were unknown, the with approximately equal numbers of plasmablastic and plasma-
positive EBER favoured the histological classification as PBL. cytic cells (patients #3, #5 and #8). All cases were positive for
Morphologically, the eight tumours classified as PBL had vari- CD138 (weak in patient #6) and negative for CD20, indicating
able numbers of plasmablasts: three cases had a plasmablastic a plasma cell differentiation. Two of four cases (50%) were posi-
morphology ( patients #4, #6 and #7), two were predomin- tive for CD45. All cases demonstrated light chain restriction.
antly plasmacytic ( patients #1 and #2) and three were mixed, PAX-5, HHV-8 and ALK were negative in all evaluable cases
Ahn JS, et al. J Clin Pathol 2017;0:1–6. doi:10.1136/jclinpath-2016-204294 3
Original article
(table 2). With exception of patient #7, all evaluable cases were presented with an EBER-negative mandibular mass, without any
negative for CD56 (table 2). Ki-67 demonstrated a high prolifer- other information available in the medical records.
ation rate in all evaluable cases, ranging from 75% to >95%. Histologically, plasmablastic cells were admixed with plasmacy-
The single CD10-positive case ( patient #6) coexpressed tic cells as well as those resembling centroblasts. These cells
MUM-1 and lacked PAX-5 and CD79a, with weak CD138 were positive for CD138, CD79a and CD45 (dim), while nega-
expression, negative CD79a and negative ALK and MYC altera- tive for CD20 and PAX-5. By in situ hybridisation, they were
tions by fluorescence in situ hybridisation (FISH). Of note, the lambda-restricted and EBER-negative. ALK immunohistochemis-
single MYC translocation-positive case ( patient #8) had a try was negative. Studies to assess for MYC rearrangement and
complex karyotype by FISH. HHV-8 infection were not available. The clinical diagnosis,
treatment, follow-up data and imaging records describing
Plasmablastic myeloma whether the lesion was soft tissue-based or bone-based were
The patients classified as myeloma ( patients #9 and #10) were also not available.
a male aged 71 years and a female aged 72 years, who presented
with isolated bone marrow disease (30% and 40% bone DISCUSSION
marrow clonal plasma cells, respectively). Imaging excluded Using a combination of morphological, phenotypical and
soft-tissue masses and lymphadenopathy. Patient #9 presented selected clinical findings, we were able to classify 10 of 11 plas-
with complete myeloma-defining signs for the diagnosis of mul- mablastic tumours into 8 PBLs and 2 PBMs, independent from
tiple myeloma. Patient #10, however, had incomplete myeloma- the previously established clinical diagnosis. Of note, the term
defining signs at presentation with anaemia and bone lesions but ‘PBL’ in this study refers to the general category of lymphomas
normal serum calcium and creatinine (table 1). Neither had a with plasmablastic features, which may include
history of immunosuppressive therapy or transplantation, HHV-8-associated tumours and ALK-positive large B-cell
lymphadenopathy, extramedullary masses, splenomegaly or lymphoma. As a result of this study, we propose a diagnostic
body cavity effusions. HIV and EBV testing were not performed algorithm to aid in the prospective classification of these
on either patient, likely due to a low clinical suspicion for neoplasms.
lymphoma, that is, extramedullary disease or lymphadenopathy. The single most useful histopathological feature in this differ-
The morphological features of the PBM case were similar to entiation was the detection of EBV-positivity by EBER in situ
those classified as PBL (table 2). The immunohistochemical find- hybridisation. Although the overall rate of EBER-positivity has
ings of PBM also mirrored those of PBL; however, most studies been reported to range from 60% to 73% of PBL cases,1–3
5 7 16 20
performed on PBL cases had not been performed on myeloma only a minority of our cases (4 of 8 evaluable cases,
cases. In both groups, the neoplastic cells were uniformly posi- 50%; patients #1, #2, #4 and #7) were EBER-positive. This
tive for CD138 and negative for CD20 and CD45. CD79a was difference may be related to specific characteristics of the popu-
negative and proliferation rate by Ki-67 was 80% in the single lation included in this study, such as prevalence of
case evaluated for these markers. EBV-ISH, HHV-8, ALK and HIV-positivity.21 Although not frequently seen in this series, a
MYC testing had not been performed. positive EBER determined the classification of a case as lymph-
oma, regardless of all other parameters, including HIV status,
Indeterminate presence of CRAB signs, concomitant bone marrow disease,
The single case that could not be categorised into lymphoma or lytic lesions or lymphadenopathy. The high positive predictive
myeloma ( patient #11) was that of a male aged 79 years who value of EBV-positivity for lymphoma is due to the infrequent
4 Ahn JS, et al. J Clin Pathol 2017;0:1–6. doi:10.1136/jclinpath-2016-204294
Original article
association of myeloma with EBV infection.6 While a positive case into either PBL or PBM. In the literature, a similar case was
EBER in a plasmablastic neoplasm allows for the classification diagnosed as a ‘neoplasm indiscriminate between lymphoma
of lymphoma, it does not discriminate PBL from primary effu- and myeloma’.33
sion lymphoma (PEL), which requires demonstration of Lymphadenopathy and/or an isolated head and neck mass
HHV-8-positivity.20–23 24 were used to favour the diagnosis of PBL in four of four
The importance of obtaining HHV-8 immunohistochemistry in EBER-negative/unknown cases, as plasmacytoma cannot be
the setting of EBER-positive plasmablastic tumours has been excluded at this anatomical site (figure 2). In addition to lymph-
emphasised in previous diagnostic algorithms.7 Nevertheless, of all adenopathy, three of four EBER-negative cases also had incom-
four EBER-positive tumours described in this series, only the one in plete myeloma signs, which were considered non-specific for
the peritoneal fluid had an HHV-8 immunostain performed myeloma, regardless of bone marrow status. In such cases, the
(patient #1). In the instances of PEL that are positive for HHV-8, diagnosis of PBL was favoured. Conversely, in the absence of
but negative for EBV,24 25 26 the differential diagnosis may also lymphadenopathy, plasma cell neoplasm/myeloma was favoured
include large B-cell lymphoma arising in HHV-8-associated multi- in the presence of bone marrow disease and at least one
centric Castleman disease (Castleman-LBCL). Therefore, in the myeloma-defining sign (figure 2).
algorithm proposed in this study, we recommend that an HHV-8
immunostain be performed to exclude PEL/Castleman-LBCL in all CONCLUSION
plasmablastic tumours regardless of HIVand EBER status (figure 2). This review illustrates the difficulties associated with the clin-
In an EBER-negative plasmablastic neoplasm, a positive ALK ical and histopathological diagnosis of neoplasms with plasma-
determines the diagnosis of ALK-LBCL,7 20 which otherwise blastic features. A specific histopathological diagnosis may only
overlaps with PBL. In our series, ALK had been performed in be possible in those cases in which association with EBV,
only three of eight plasmablastic cases ( patients #1, #3 and HHV-8 and the rare expression of ALK can be demonstrated.
#6), two of which were EBER-negative, suggesting that the As such, EBER-negative cases are likely to be reported non-
infrequent use of this study is a possible explanation for the specifically as ‘hematolymphoid neoplasm with plasmablastic
under-recognition of ALK-LBCL.27 Immunophenotypical fea- features’.
tures may be helpful to guide the request for ALK studies. We show that in EBER-negative cases, distinction between
ALK-LBCLs are usually negative for CD79a and CD30,20 unlike lymphoma and myeloma relies heavily on an extensive and
other plasmablastic neoplasms, in which expression of CD79a detailed knowledge of the clinical history. Our data suggest that
and/or CD30 is variable.2 8 22 the diagnosis of lymphoma may be favoured in the presence of
MYC alterations (including MYC/IgH) are found in approxi- significant lymphadenopathy or oropharyngeal soft tissue mass
mately 50% of PBL cases.19 28–30 However, the identification of at presentation, while the diagnosis of myeloma may be
MYC rearrangement in 15% of plasma cell myeloma31 and up to favoured on demonstration of isolated bone marrow disease and
50% of MYC alterations PBM,14 limits its value in distinguishing at least one myeloma-defining sign.
PBL from myeloma. In this study, only one of three cases classi- The diagnosis of lymphoma may be more easily established
fied as PBL had positive MYC rearrangement by FISH. on retrospective review of clinical and laboratorial data, as in
PBL and myeloma show significant morphological and pheno- this study. However, prospectively, when complete medical
typical overlap, limiting the ability of a pathologist to render a information is often unavailable or unknown, specific classifica-
specific diagnosis in the majority of EBER-negative cases. tion of a plasmablastic neoplasm may be impossible. Future
Although traditionally described as a tumour with a predomin- studies are warranted in order to further characterise PBL and
ance of plasmablasts, PBL may be predominantly plasmacytic in PBM, as well as to determine histopathological markers to
up to 41% of cases.2 6 16 In fact, two of eight cases in this study better differentiate these two entities.
(25%, patients #1 and #2) were plasmacytic. We demonstrate
that the number of plasmablasts is variable in both plasmablastic
and plasmacytic morphological variants, and mixed cases with Take home messages
approximate equal number of both cell types also exist.
Interestingly, two cases in this series also showed cells resem-
bling centroblasts (figure 1), further contributing to morpho- ▸ Hematolymphoid neoplasms with plasmablastic morphology
logical heterogeneity of plasmablastic neoplasms. ( plasmablastic lymphoma (PBL), plasmablastic myeloma
Phenotypically, as both PBL and PBM are of terminally differ- (PBM), primary effusion lymphoma, large B-cell lymphoma
entiated B-cells/plasma cells,3 6 7 16 32 both tumours are positive arising in human herpesvirus 8 (HHV-8)-associated
for CD138 and MUM-1, while negative for CD20 and PAX-5, multicentric Castleman disease and anaplastic lymphoma
as shown in our study (table 2). CD56, while originally thought Kinase (ALK)-positive large B-cell lymphoma) may show
to favour myeloma,2 8 16 has since been demonstrated in up to significant morphological and immmunophenotypical overlap.
56% of PBL cases,6 thereby reducing its utility. In our study, ▸ The differential expression of Epstein-Barr virus, HHV-8 and/
two of five (40%) cases were CD56 positive, one of which or ALK is useful in distinguishing neoplasms with
stained weakly. CD10 expression has been reported in 29% of plasmablastic morphology in the absence of clinical history.
PBM6 and in 41%–67% PBL cases3 6 16 32 (in this study one of ▸ When myeloma-defining signs are incomplete or absent, the
four PBL, 25%), and is therefore of little utility. A high Ki-67 is presence of significant lymphadenopathy and/or
unlikely to distinguish between PBL and PBM as both entities oropharyngeal soft tissue mass favours the diagnosis of
usually demonstrate high rates of proliferation.6 32 lymphoma, regardless of bone marrow status.
Clinical correlation is essential to classify neoplasms with ▸ Future studies are warranted in order to determine
plasmablastic morphology, as illustrated by the case classified as histopathological markers to differentiate PBL and PBM.
‘indeterminate’ in this study ( patient #11). In this case,
although histological data were available, the lack of sufficient
clinical data and EBER-positivity precluded classification of this Handling editor Mary Frances McMullin