|Home | About | Journals | Submit | Contact Us | Français|
Hodgkin lymphoma (HL) was shown to be a B cell malignancy using PCR-clonality studies of microdissected Reed-Sternberg cells. While methods for the detection of B cell clonality could aid in the diagnosis of HL, microdissection is not practical in most clinical settings. We assessed the standardized BIOMED-2 IGH and IGK PCR primers for the detection of clonality using 50 consecutively diagnosed formalin-fixed paraffin-embedded (FFPE) classic Hodgkin lymphoma specimens. Without microdissection, clonality was detected in 23/47 assessable cases. The IGK assay was significantly more sensitive than the IGH assay (18 vs. 10 positive results). These data, and two representative cases, demonstrate that PCR based B-cell clonality assays have utility when the histologic differential diagnosis of an FFPE specimen includes classic Hodgkin lymphoma.
The diagnosis of Hodgkin lymphoma (HL) may become difficult due to factors such as extranodal presentation, small size of the biopsy, or histologic similarity to T cell lymphoma. When the differential diagnosis includes non-hematopoietic and T cell malignancies, robust methods to detect B cell clonality may be useful as the neoplastic cells in HL have been demonstrated to be clonal B-lymphocytes.1 A range of immunohistochemical markers, especially PAX5 and CD20, can help to distinguish HL from T cell and non-hematopoietic processes. However, some cases remain challenging particularly when the site is atypical, the clinical course is complex, or the specimen is in some way suboptimal. HL often presents in deep mediastinal nodes, which are frequently difficult to excise, and specimens obtained by mediastinoscopy are small and fragmented, rendering evaluation of architecture challenging. Atypical presentations of HL, in which lymphoma was not in the differential diagnosis at the time of biopsy, present the additional challenge that fresh tissue frequently is not available, and all additional studies must be performed on formalin-fixed paraffin embedded (FFPE) specimens.
The BIOMED-2 consortium effort has produced a set of PCR based methods for evaluating B cell clonality and validated these extensively on fresh specimens of non-Hodgkin lymphomas.2 The application of these assays to Hodgkin lymphoma has been less extensively characterized.3-5 Furthermore, validation studies for these assays applied to formalin fixed paraffin embedded (FFPE) specimens are limited. In current clinical practice, the majority of cases for which this type of ancillary study is required are likely to be FFPE specimens.
Here we assess the utility of the BIOMED-2 IGH and IGK assays applied to FFPE Hodgkin lymphoma specimens and present two cases as illustrative of the role that molecular diagnostic studies can play in problematic HL cases.
Fifty consecutive cases, diagnosed as a classic type Hodgkin lymphoma, were retrieved from the archive of surgical pathology at Strong Memorial Hospital under a protocol approved by the Research Subject Review Board. The specimens were from 2005-2007. Only cases with a definitive diagnosis of a Classic Hodgkin lymphoma (CHL) were selected for review; cases diagnosed as “suspicious”, “consistent”, “suggestive”, or any other qualifier were not retrieved. None of the cases had a reported positive IGH clonality assay that may have contributed to the diagnosis. Flow cytometry was reported on 44 cases; review of these data did not detect a clonal B cell population in any case. Diagnoses were confirmed by review of histology and immunohistochemistry; additional immunohistochemical stains were performed as needed. Three specimens were rejected from further analysis because there was too little material in the paraffin blocks for DNA extraction and/or for additional immunohistochemical studies (CD20/79a). 6/47 cases had detectable expression of CD20 on the R-S cells, ranging from about 5% to 100% of the R-S cells; the fraction of CD20 positive cases (12%, with 95% CI 6-25%) is comparable to that generally reported. The two index cases were from 2008 and 2009 and were not included in the fifty consecutive cases.
The estimated fraction of benign B cell/plasma cells in the specimen was based on CD20, PAX5 and/or CD79a stains. A CD79a stain was performed on selected cases in which the plasma cell number appeared substantial. However, in practice the number of CD79a positive cells was not in any case substantially different from the number of CD20 positive cells. For B/plasma cell and CD30+ cell quantification, 10 high-power fields (HPF) fields were selected randomly, except that fields with >50% fibrosis were excluded for the CD30 enumeration. The percentage of B/plasma cells were rounded to closest increment of 5. In the case of core biopsies, the fields were selected linearly along the cores.
DNA was isolated from four 10-μm sections of formalin fixed paraffin embedded (FFPE) tissue using the QIAamp system (Qiagen Inc., Valencia, CA). The concentration of DNA was estimated by specrophotometry using the Nanodrop ND-1000 spectrophotometer (Wilmington, DE). DNA quality was assessed using a multiplexed control PCR ladder with amplicons of 100, 200, 300, 400 and 600 bps.6 44/47 specimens showed amplification of the 400 bp control product. No specimen was excluded based on inadequacy of total DNA or the results of the control amplification.
PCR of the IGH and IGK genes to assess clonality used the BIOMED-2 system. Master mixes were purchased from Invivoscribe Technologies (San Diego, CA) and the PCR was per manufacturer instructions and used HotStar Taq DNA polymerase (Qiagen). In an initial testing phase, we assessed reactions using the IGH FR2 and FR3 primer sets, and both IGK primer sets. In a second phase, we assessed reactions using the IGH FR1 and D-J assays. The reaction was cycled 35 times between 94°C for 30 seconds, 61°C for 30 seconds and 72°C for 90 seconds, preceded by 10 minutes at 94°C, and followed by 5 minutes at 72°C. After PCR the products were heteroduplexed by raising the temperature to 95°C for four minutes followed by 2°C for 2 to 20 hours. Immediately after the heteroduplex step the amplicons were resolved by electrophoresis on a preformed 10% polyacrylamide mini-gel in TBE buffer (Invitrogen, Carlsbad, CA) together with molecular weight markers (MspI digest of pUC18 plasmid DNA, Sigma, St. Louis, MO), stained with SYBR Gold (Invitrogen) and photographed. Polyclonal (IVS-0000) and clonal (IVS-0007) control DNAs were purchased from Invivoscribe. Sensitivity of the assays was assessed using mixtures of clonal and polyclonal DNA. Products from 12 of the 18 positive IGK assays were purified and sequenced using the J-kappa consensus primer. In each case, the sequence reflected rearrangement of the IGK locus (data not shown).
23/47 FFPE HL specimens had a clonal IGH and/or IGK result (Table 1). IGK assays contributed significantly more positive results than the IGH assays (18 vs 10 positive results; p=0.03 Fisher Exact, two tailed). Of the 10 cases with detectable IGH rearrangements, 9 were detected with assays of FR2 and/or FR3 regions. The FR1 assay did not detect clonality in any specimen. The IGH D-J assay detected clonality in 2 cases; these cases had demonstrable clonality with the IGH FR2, IGH FR3, or IGK assays. Therefore, IGH FR1 and D-J assays did not independently contribute to the detection of clonality in any case.
The IGH and IGK methods differed in their sensitivities to the number of Reed-Sternberg (RS) cells and mononuclear variants. We estimated the number of RS cells and mononuclear variants by counting the average number of large CD30-positive cells per HPF (Table 2). Cases were divided into three groups with increasing numbers of CD30-positive cells. There appears to be an association between the number of CD30-positive cells and clonality detection using the IGH assay (p=0.05, chi squared). No evidence of a similar trend was seen for the IGK assay (p=0.5). These findings suggest that clonality detection using the IGK assay is less dependent than the IGH assay on the number of CD30-positive cells.
The effect of non-malignant B cells and plasma cells was assessed by estimation of the fraction of cells staining for CD20 and/or CD79a (Table 3). Cases were divided into a group with sparse (28 cases) and those with more abundant B/plasma cells (19 cases). No relationship between clonality detection and the density of non-malignant B/plasma cells was observed. Several different cutoffs for the B/plasma cell number were assessed; none yielded any evidence of a relationship between the density of B/plasma cells and clonality detection using either the IGH or the IGK assays.
While there appeared to be a relationship between expression of CD20 on the R-S cells and the detection of clonality with any PCR primer set, the association did not reach significance (p=0.08; Fisher exact, two-tailed). 12% of cases (6/47) had detectable expression of CD20 on RS cells: 5/6 CD20-positive cases had detectable clonality; 17/41 CD20-negative cases had detectable clonality.
A 44 year old male with no significant pre-existing conditions presented with a several month history of right thigh pain, and then a week history of a readily apparent mass. MRI was interpreted to show a heterogeneous soft tissue mass lying mostly in the vasti muscles and extending at least 25 centimeters distally from the femoral head. Involvement of the femur was also noted, suggesting a likely diagnosis of a soft tissue or bone sarcoma. Core biopsies were obtained using a 16 gauge needle.
The lesion was comprised predominantly of histiocytes, with numerous eosinophils, lymphocytes and plasma cells (Image 1). In addition to malignancy, the differential diagnosis for this small core biopsy included peri-tumoral inflammation. Scattered through the lesion were some highly atypical cells with large nuclei with prominent nucleoli, some of which were eosinophilic. Occasionally these cells were clustered. The atypical cells were positive for CD30, and negative for other markers, including B cell markers CD20, CD79a, and Pax5; T cell markers CD2, CD3, CD4, and CD8; macrophage/histiocyte markers CD68 and CD1a. CD15 and S100 were also negative. BIOMED-2 clonality assays for the TCRgamma (data not shown) and IGH loci (Image 2) were negative. Gene rearrangement studies for the IGK locus were positive (Image 3). A diagnosis of classical Hodgkin lymphoma was made.
Radiographic studies indicated involvement of inguinal nodes, the patient was staged as IIEA, and begun on therapy of ABVD and local irradiation. The mass and pain resolved with therapy. Therapy was complicated by the need to place an intrafemoral rod. Six months after completion of therapy, the patient is free of symptoms and a PET/CT scan shows no evidence of recurrent disease.
A 68 year old female presented with shortness of breath and cough. Radiology showed a pleural effusion and soft tissue mass. The radiologic impression was bronchogenic carcinoma. A transbronchial core needle biopsy was performed. The biopsy showed histology suspicious for Hodgkin lymphoma, but the specimen was inadequate for a definitive diagnosis. An open procedure was performed which showed sheets of CD30+ large cells (Image 4). These cells lacked expression of CD15, CD20, CD79a, and CD45. PAX5 reactivity was equivocal while in some areas there was a suggestion of CD2 expression. ALK-1 and other T cell markers were negative. BIOMED-2 clonality assays for the TCRgamma (data not shown) and IGH (Image 2) loci were negative. A clonal B-cell population was demonstrated using PCR of the IGK locus (Image 3). A diagnosis of Hodgkin lymphoma was made.
The patient was initially responsive to ABVD with regression of the mediastinal mass. However, a lesion in the right upper lobe expanded and biopsy showed Hodgkin lymphoma.
Detection of clonality using PCR of the antigen receptor genes has a relatively infrequent but occasionally critical niche in the diagnostic workup of non-Hodgkin lymphomas (NHL). The application of PCR-based clonality detection for Hodgkin lymphomas is much less developed but its role could be more critical than it is for NHL because, in contrast to the workup of NHL, flow cytometry does not detect the clonality in HL. Here we show that about half of all classic HL FFPE specimens can be expected to show B cell clonality using a combination of the BIOMED-2 IGK and IGH assays.
The two cases presented here indicate scenarios in which detection of clonality using DNA purified from FFPE tissue contributed substantially to the diagnosis. In the first case the diagnosis was unsuspected due to the unusual presentation and strong clinical suspicion of a sarcoma. The infradiaphragmatic presentation, and direct bone involvement are quite unusual sites of presentation of de novo HL7. In the second case, the diagnosis was made difficult by the small size of the specimen, the syncytial malignant cells which mimicked a NHL, and the lack of classic immunophenotype.
Only a handful of publications have described PCR-based clonality detection for HL using non-microdissected specimens.3, 5, 8-10, 4 Just 3 of these have assessed the use of the standardized BIOMED-2 primer system for FFPE HL specimens (Table 4). The sensitivity for detecting clonality in HL in all of these studies is similar, and clearly lower than that reported for most other histologic subtypes of lymphoma.4 There are several reasons that the PCR-based methods are less sensitive for HL compared to NHL. First, the malignant cells in HL are often relatively sparse. Second, there may be an abundance of non-malignant B cells which produce a polyclonal background signal that lowers the analytical sensitivity for detection of the clonal signal. Third, the RS cells appear to subject their immunoglobulin loci to a high degree of somatic hypermutation increasing the probability that a primer binding site(s) will be altered causing the PCR to fail. Nevertheless, several groups have shown the successful application of the PCR-based BIOMED-2 clonality assays to HL (Table 4).
The sensitivity of the BIOMED-2 assays depends on the histology, the specimen type, and the specific assay.6 For example, the BIOMED-2 IGH assays have very high sensitivities when applied to small lymphocytic lymphoma (SLL) specimens, whether these specimens are fresh/frozen or FFPE. In contrast, the same assays fail for about 50% of FFPE follicular lymphoma (FL) specimens while showing >90% sensitivity for fresh FL specimens. The dependence on the choice of immunoglobulin locus is also exemplified by FFPE FL specimens: while the sensitivity is about 50% with IGH assays, the IGK assays have a sensitivity of >90%.11
Although several recent publications on antigen receptor PCR in Hodgkin lymphoma employed the BIOMED-2 primer sets, our study differed in using heteroduplex analysis on polyacrylamide gels rather than Genescan analysis which involves the use of a fluorescent primer and separation by capillary electrophoresis.12 Both methods were described in the initial BIOMED-2 publication and were reported to have similar sensitivities for the IGK and IGH assays.6 In the heteroduplex procedure, the PCR products are denatured and then allowed to re-anneal. The likelihood of any specific DNA product re-associating with its exact complement in a polyclonal pool is extremely small. Therefore, the products derived from polyclonal B cells form a heteroduplex in which there are a small number of mismatches. The mismatches cause the annealed products to have a lower mobility in polyacrylamide gel electrophoresis. The specific amplicon from a clonal B-cell population is much more abundant and reassociates to form homoduplexes without mismatches. The heteroduplex signal derived from the polyclonal B cells appears at a higher apparent molecular weight easing the visualization of the signal derived from the clonal population. This is shown in Image 5, which compares the results with and without the heteroduplex procedure for IGK PCR. Without the use of the heteroduplex procedure a small clone (10% +, 5% +) could not be seen in the strong polyclonal signal. Using the heteroduplex procedure the clone is easily appreciated. Laboratories that have compared the Genescan and heteroduplex methods have noted a preference for heteroduplexing in analysis of the IGK locus due to its restricted junctional diversity, in contrast to a slight advantage in sensitivity for the Genescan method for the other antigen receptor genes.2, 6 For HL we found a similar overall clinical sensitivity compared to the other studies (Table 4) and conclude that the heteroduplex analysis provides an adequate and cost effective alternative to approaches based on capillary electrophoresis.
Chute et al3 studied clonality in FFPE HL specimens using the BIOMED-2 IGH assays and found about 25% of HL cases to be clonal. These investigators found that the likelihood of clonality detection with the IGH assay increased with greater numbers of CD30+ cells. We also found a relationship between the density of CD30+ forms and clonality detection using the IGH assay but there was not any suggestion of a similar effect of CD30+ cells on the sensitivity of the IGK assay (Table 4). This result suggests that the IGK assay is less dependent on the number of malignant cells, perhaps in part explaining its relatively increased sensitivity in detecting clonality in HL where the malignant cells are often sparse.
Chute et al3 also found that the rate of positive results with the IGH assay was inversely proportional to the density of non-malignant B cells, a finding that we did not reproduce. While we included plasma cells as a component of the polyclonal background and Chute et al did not, we do not feel that this difference is large enough to explain the disparity. We suspect that this difference may be explained by the heteroduplex step in our procedure which reduces interference from the background signal due to polyclonal B cells. As shown in Image 5, the heteroduplex method allows clonal populations as small as 5% to be clearly detected in a polyclonal background.
Our finding that PCR can demonstrate clonality in about 50% of CHL without microdissection is unlikely to be explained by “pseudoclonality”. First, sequencing of the IGK PCR products showed that all represent bona fide rearrangements of the IGK locus rather than spurious non-specific reaction products. Second, all reactions were in duplicate and gave identical clonality results, precluding false positives due to random mispriming in an early PCR cycle. Third, flow cytometry was performed on 20/23 CHL specimens positive for B cell clonality; no suggestion of B cell clonality was detected in any of these 20 specimens indicating that a “composite” lymphoma is unlikely in any of these cases.
The BIOMED-2 IGK assays may be generally more robust for clonality detection in FFPE samples from malignancies in which the IGH loci are more heavily affected by somatic hypermutation, such as follicular lymphoma and HL.13, 14 Our data showing the relatively greater robustness of the IGK compared to the IGH assay for HL are compatible with those recently suggested in a smaller study of 8 specimens (Table 4).5 We previously showed that the IGH assays failed in about 50% of FFPE follicular lymphoma specimens in which the IGK assays succeeded. Although the IGK assays are appreciably more sensitive, the IGH FR2 and FR3 assays contribute substantially to the overall sensitivity for CHL FFPE specimens. Thus, a clinical lab will need the IGK and IGH FR2/3 assays to achieve the best sensitivity in the evaluation of the occasional specimen of Classic Hodgkin lymphoma which does not show typical clinical or histologic features.
Supported by NCI Lymphoma SPORE (CA130805) and the Department of Pathology and Laboratory Medicine, University of Rochester School of Medicine, Rochester NY