In-depth molecular analysis of lymphomas with lymphoplasmacytic differentiation may provide more precise diagnosis and rational treatment allocation

We performed a molecular analysis of formalin-fixed paraffin embedded and decalcified bone marrow trephine biopsies of 41 patients with a B-cell disorder with lymphoplasmacytic differentiation to enable a more precise diagnosis and to describe potentially prognostic and therapeutic relevant mutations. Analysis was performed with a commercially available next-generation sequencing (NGS) lymphoma panel (Lymphoma Solution, SophiaGenetics). Results were correlated with clinical and pathological parameters. Our group covered a spectrum of B-cell disorders with plasmacytic differentiation ranging from Waldenstroem’s macroglobulinemia (WM), to small-B-cell lymphomas with plasmacytic differentiation (SBCL-PC) to IgM myeloma (MM). The most helpful diagnostic criteria included morphology and immuno-phenotype as a prerequisite for the interpretation of molecular analysis. MYD88 mutation was present in nearly all WM, but also in 50% of SBCL-PCs, while MM were consistently negative. Driver mutations, such as TP53, were already detectable early in the course of the respective diseases indicating a higher risk of progression, transformation, and reduced progression-free survival. In addition, we report on a novel BIRC3 frameshift mutation in one case of a progressive WM. Our data indicate that patients with LPL/WM might benefit from thorough pathological work-up and detailed molecular analysis in terms of precise diagnosis and targeted treatment allocation. Supplementary Information The online version contains supplementary material available at 10.1007/s00277-023-05531-9.


Introduction
Lymphoplasmacytic lymphoma (LPL) is a B-cell neoplasia defined by a variable mixture of lymphocytes, lymphocytes with plasmacytic differentiation and plasma cells.It is known as Waldenstroem's macroglobulinemia (WM), when presenting with IgM paraproteinemia and bone marrow involvement [1,2].However, (lympho)plasmacytic differentiation can also be found in other low-grade B-cell lymphomas, mainly in marginal zone lymphoma (MZL) or chronic lymphatic leukemia (CLL), but also in mantle cell lymphoma (MCL) [1,[3][4][5][6].Beside morphological characteristics, the immune phenotype may overlap between lowgrade lymphomas, resulting in clear diagnostic complexities [3,4].In fact, LPL/WM has no distinctive immune phenotype, differentiating it from MZL making a clear diagnosis nearly impossible [3].This in turn leads to a diagnosis of a small B-cell lymphoma with plasmacytic differentiation (SBCL-PC) [4].In addition, elevated IgM levels, lymphoid differentiation, and overlapping immune phenotype can also be seen in MM [7][8][9].The detection of MYD88 mutations, most frequently the MYD88 L265P mutation, which is found in more than 90% of WM, was believed to reduce this diagnostic dilemma [4,10].However, while LPL/WM can be negative for MYD88 mutations, rare cases of MZL, CLL, and even MCL can harbor a MYD88 mutations, while only MMs are consistently negative [1,3,4].A variety of clinical studies has dealt with the impact of molecular genetics in LPL/WM [5,[11][12][13][14][15][16].The vast majority of WM demonstrates MYD88 mutations associated in up to 40% of cases with mutations on CXCR4 [5].Lack of MYD88 mutation in LPL/WM, however, predicts worse outcome with higher risk of transformation, therapy-related myelodysplastic syndrome, as well as shorter time to transformation compared to MYD88 mutated patients [5,11,12].Responses in MYD88wild-type patients seem to be even worse with respect to BTK-inhibitors with the possible exception of zanubrutinib [5,11,12,17].While MYD88 is nearly always mutated in L265P, up to 40 different CXCR4 mutations to date occur in LPL/WM [5].CXCR4 mutated cases have high serum IgM levels resulting more frequently in hyperviscosity of blood [5,14].In addition, the depth of response and progressionfree survival partly depends on CXCR4 mutations when receiving BTK inhibitor-based treatment or Venetoclax, the first representative BCL-2 inhibitor [5,[13][14][15]18].No such impact is observed using upfront treatment with proteasome inhibitors [16].Similarly, TP53 mutations, though only rarely found in WM are associated with poor outcome, while its role in resistance to BTK-inhibitor therapy is still under investigation [14,[19][20][21][22].Further alterations found in LPL/ WM include mutations on KMT2D, ARID1A, CD79A and B, NOTCH2, PRDM1, and TRAF3, but their impact on prognosis and treatment remains to be determined [14,23,24].
We have previously dealt with the presence of MYD88 L265P mutation in LPL/WM [10,25].We showed that digital PCR is a sensitive tool to detect this mutation in formalin-fixed paraffin-embedded (FFPE) and decalcified bone marrow trephine biopsies of patients with LPL/WM [25].
In the present study, we attempted to integrate morphology and molecular pathology for a more precise diagnosis by performing next-generation sequencing (NGS) in primary and relapsed LPL/WM.Our results indicate that patients might benefit from a broader molecular analysis already upon diagnosis to allocate the appropriate treatment.

Patients
Forty-one patients were identified either from the database of the University Clinics of Internal Medicine V, Hematology and Oncology or from the Institute of Pathology, Neuropathology and Molecular Pathology.The patients have been diagnosed with a neoplastic B-cell disorder presenting with a plasmacytic or lymphoplasmacytic differentiation between 1993 and 2016.Bone marrow trephine biopsies were collected and re-evaluated, including follow-up biopsies, if available.Clinical data such as IgM level at diagnosis, presence or absence of osteolysis, cytogenetics and information about disease progression or transformation were assessed from the patient chart.
The study was conducted according to the ICH-GCP guidelines and the declaration of Helsinki.Ethical approval was obtained from the ethical committee of the Medical University of Innsbruck (EK-Nr.1362/2020).

Re-evaluation and immunohistochemistry
A careful review was performed of all bone marrow biopsies including routinely performed immunohistochemistry to assess morphological parameters such as infiltration pattern, volume of infiltration and number of mast cells and immune-phenotype.
Pattern of infiltration was defined according to Garcia-Reyero J et al. [26] as: Paratrabecular and interstitial, either patchy or nodular and/or diffuse Only interstitial, either patchy or nodular or diffuse or diffuse-solid A mast cell count on routinely performed Giemsa stains was done in five representative high-power fields (HPFs) in each biopsy.
Re-evaluation of immunohistochemistry included presence or absence of aberrant marker expression such as CD23 or CD5, positivity for plasma cell markers and evaluation of clonality using either immunohistochemistry or chromogen in situ hybridization (Ventana, Tucson, USA) on an automated immunostainer (Benchmarks Ultra, Ventana, Tucson, USA).

DNA isolation
DNA from FFPE tissue was extracted using a QIAamp DNA FFPE Tissue Kit (Qiagen, Hilden, Germany).DNA isolation was conducted according to the manufacturer's instructions.Briefly, 3 × 5 µm slices or 6 × 5 µm slices (depending on the size of the tissue) were prepared in DNAse-and RNAse-free Eppendorf tubes.Paraffin was removed by a Xylol wash.The samples were lysed by use of proteinase K and heat incubated at 90 °C to reposition formalin crosslinking.Then the samples were transferred into QIAamp MinElute columns possessing special DNA binding properties.Residual impurities were washed away by use of the provided buffers.Finally, the respective DNA was eluted from the column membrane.The pure and concentrated DNA samples were stored at − 20 °C until further use.

Next-generation sequencing
NGS was performed using a commercially available lymphoma panel (Lymphoma Solution, SophiaGenetics, Geneva, Switzerland).This panel includes 54 genes relevant for lymphomagenesis, including those found in lymphomas with plasmacytic differentiation, such as MYD88, CXCR4, ARID1A, KMT2D, and those associated with disease progression or transformation such as TP53.Data were evaluated using the platform provided by SOPHIA GENETICS (SOPHIA DDM version 5.10.28.2).A cutoff of 5% was determined for the variant fraction in order to be regarded as a reliable mutation.

Statistics
For statistical analysis, SPSS 26 software was used.To test for correlations with clinical-pathological parameters the chi-square test was used.To determine the optimal cutoffs for continuous variables (such as number of mast cells) considering diagnostic ability, progression, transformation, and overall survival ROC analysis was performed and AUC was interpreted as follows: 0.5-0.6 = no discrimination, 0.6-0.7 = poor, 0.7 to 0.8 = acceptable, 0.8 to 0.9 = excellent, > 0.9 = outstanding.The optimal cutoffs were calculated using Youden´s index if AUC was > 0.7 [27,28] (see suppl. Figure 1 and suppl.Table 1).Survival analysis was performed using Kaplan-Meier survival analysis compared with the log rank test.Multivariate analysis was done using Cox regression.A p value < 0.05 was considered significant.

Patient's characteristics and morphology
Eight patients had to be excluded because the FFPE material was insufficient for DNA isolation (< 20% neoplastic cells in the bone marrow) or NGS did not provide reliable results due to reduced DNA quality.Thus, finally, our study group included 30 patients diagnosed with LPL, including 24 cases with WM and 6 with SBCL-PC.Three cases of (IgM) MM were included as a proof of principle.Table 1 summarizes the clinical and Table 2 the pathological characteristics of the final study group.
Most lymphomas had a classical WM morphology.Some of them had a more plasmacytoid differentiation and Dutcher bodies were prominent.Immune phenotype was also typical for WM with negativity for CD5 as well as CD23 and variable expression of plasma cell markers (see Fig. 1A-D) in most of the cases.
The only wild-type WM showed progressive fibrosis and osteosclerosis and had a synchronous diagnosis of diffuse large B-cell-lymphoma (DLBCL) in lung and central nervous system (CNS) (see Fig. 2A-D).
Two of the six SBCL-PC showed a more variable immune phenotype with additional expression of CD5 and CD23, but also weak or focal expression of plasma cell markers.While one case had polyclonal plasma cells in the background and did not show a MYD88 mutation, the other case was positive for a MYD88 mutation and had monoclonal plasma cells.Both were finally classified as CLL (see suppl.Table 2).Nevertheless, the diagnosis of MYD88 mutations together with clonal plasma cells suggests the existence of a spectrum between B-CLL and classical WM (see Fig. 3A-D).
The other four SBCL-PCs had a more classical WM morphology with paratrabecular and interstitial nodular infiltration (two cases), a striking plasmacytic differentiation (one case) or an interstitial pattern of infiltration being suggestive of a MZL (one case).MMs showed clonal plasma cell proliferations with CD20 and cyclin D1 expression in two of three cases.The best distinguishing features as highlighted in Table 1 for the diagnosis of LPL/WM were morphological parameters, such as infiltration pattern (p = 0.029) and mast cell count (p = 0.001), immune phenotype (p < 0.001) and presence of MYD88 mutation (p < 0.001) (see Table 1).
To compare molecular analysis with other clinical parameters, we carefully reviewed the literature on known driver mutations in LPL/WM and CLL and we defined molecular groups as follows: Group 1: All patients with MYD88 and/or CXCR4 and/ or ARIDA1 mutation without other additional mutations, which are known as driver mutations in lymphomas.Group 2: All patients with MYD88 and/or CXCR4 and/ or ARIDA1 mutations plus additional mutations, which are known driver genes in lymphomas (TP53, TNFAIP3, BIRC3, FBXW7, POT1 XPO1).
The three cases of IgM MM were not grouped in either of the categories and thus were not included in further statistics.
When applying the cutoff for mast cell counts as determined by ROC in molecular group 1 significantly more often a mast cell count above the cutoff was observed than in molecular group 2 (17/21 (81%) above the cutoff versus 2/9 (22%) above the cutoff; p = 0.002).The same was observed for immune phenotype (21/21 (100%) versus 7/9 (78%); p = 0.025).However, these results most likely merely depict the fact that group 2 included more cases which were difficult to interpret diagnostically reflected by the immunophenotype and mast cell count.
Fig. 3 SBCL-PC finally classified as CLL with features reminiscent of LPL/WM including plasmacytic differentiation with paratrabecular and interstitial infiltrate of lymphocytes (A GIEMSA) admixed with plasma cells (B immunohistochemistry for vs38c), while the majority of the neoplastic cells is CD20 positive (see inlet in B).Additionally, neoplastic cells express CD5 and CD23 (C immunohistochemistry for CD5 and inlet in C immunohistochemistry for CD23).Chromogen in situ hybridization for light chains kappa and lambda showed a clear predominance of kappa positive light chains (D kappa and inlet in D lambda).(all magnification 20 ×) Table 3 Most frequent mutations detected in the study group including primary diagnoses as well as biopsies at relapse (N = number of biopsies (total 43) evaluated) 1 Includes two mutations other than MYD88 L265P 2 Includes frameshift insertions in ARID1A and POT1 and a frameshift deletion in KMT2D In seven patients, follow-up biopsies were available for analysis, resulting in 43 finally analyzed bone marrow biopsies.Five of them had progressive disease and MYD88 mutation was the most frequently observed mutation except for one case which had a BIRC3 frameshift mutation in all examined specimens, so far not described in the literature (see Table 5).In four patients with subsequent DLBCL, additional oncogenic mutations were detected in the WM component before the appearance of the aggressive lymphoma component (see Table 4).

Association with disease progression and survival
For analysis of progression-free survival (PFS) and overall survival (OS), we excluded the three MM cases, since these are clinically different and more aggressive entities.In 25 of the remaining 30 patients, information on follow-up concerning disease progression and/or transformation was available.
Table 5 depicts the association of clinical and pathological characteristics with the development of progressive disease, transformation, and death.
Using Kaplan-Meier survival analysis, a significantly better PFS was observed in patients with a paratrabecular and interstitial infiltration pattern compared to those with only interstitial infiltration pattern (8/18 patients, median PFS 112 months versus 5/7patients, median PFS 63 months; p = 0.037, see Fig. 4).None of the other parameters, such as molecular group, age or gender significantly influenced PFS, though a tendency was observed that additional driver mutations lead to a worse PFS (6/17 with progressive disease, median PFS 117.00 months versus 7/8 with progressive disease, median PFS 60.62 months; p = 0.052).
None of the evaluated parameters significantly influenced OS, though interstitial infiltration pattern showed a slight tendency for worse OS (7/18, median OS 140.90 months versus 4/7, median OS 85.67 months; p = 0.097).

Discussion
In our study, we showed that assessment of classical morphological features in the bone marrow, including infiltration pattern and increased numbers of mast cells still are a prerequisite in the diagnosis of WM [26].In fact, most WMs had higher mast cell counts and infiltration pattern was often (nodular) paratrabecular and interstitial, less likely only interstitial, irrespective if nodular or diffuse-solid [26,29].Though presence of MYD88 L265P mutation is helpful in the diagnosis, this mutation is not unique to LPL/WM [1,30].Thus, combining careful morphological evaluation with immune phenotyping and molecular analysis in debatable cases is the best way to achieve a proper diagnosis [26,29].
As already reported in various studies, MYD88 L265P was the most frequently found mutation in LPL/WM also in our study followed by mutations on CXCR4 [1,15,23].In some cases, additional mutations turned up, which are infrequent in LPL/WM.This includes TP53 mutations associated with MYD88 and CXCR4, which were present in 4% (2 cases) of LPL/WM, a finding that is well in line with the literature [19,31].Both patients experienced disease progression and finally developed DLBCL with TP53 mutation preceding transformation for 4 and 8 years, respectively (see Table 4).In fact, TP53 mutations predict a worse outcome and rapidly progressive disease [19,21,31,32].Other mutations include KMT2D missense and frameshift mutations, which appear in a variety of lymphomas.KMT2D are frequently seen in LPLs and mutations have a possible connection to hypogammaglobinemia, autoimmune phenomena and the Kabucki syndrome, but its role in pathogenesis of LPL is still largely unknown [24,26,33].In two cases, TNFAIP3 missense mutations were found, which are associated with a diagnosis of B-cell lymphoma including DLBCL [34,35].Functionally TNFAIP3 is a negative regulator in the NFκB pathway and believed to be a tumor suppressor gene [36,37].In fact, Compagno et al. [34] who closely investigated DLBCL for the presence and type of TNFAIP3 mutations reported inactivation by a two-hit mechanism such as a combination of inactivating mutations and/or deletions, which is one possible mechanism that promotes lymphomagenesis.Interestingly loss of chromosome 6q, where TNFAIP3 gene is located, is not only a frequent event in DLBCL but is also the main cytogenetic event in WM [32,34,37].Indeed, in one of our two patients with a TNFAIP3 mutation cytogenetics revealed a 6q deletion indicating a similar mechanism.In addition, a TP53 mutation was present and in one patient, who later on developed transformation.Thus, one might speculate that these findings already indicate a more aggressive disease despite morphologic characteristics of WM.Beside one patient diagnosed as LPL/WM was quite interesting due to morphology, molecular genetics, and clinical outcome.Morphologically, a diffuse lymphoplasmacytic infiltrate in the bone marrow with increased mast cells was seen in the bone marrow and immune phenotype was typically for WM.Molecular analysis revealed a MYD88 and a CXCR4 mutation and in addition a frameshift mutation on BIRC3 R434Gfs*12, which was also present in all follow-up biopsies available.Overall tumor burden in the bone marrow remained high (> 40-50%) during the course of disease despite treatment as was allelic burden of BIRC3 and the patient finally succumbed to the disease.BIRC3 mutations appear in up to 4% of newly diagnosed CLL, and rarely in other low-grade lymphomas such as MZL and WM [38].BIRC3 mutations are mostly nonsense or in-frame variants leading to a loss of function [38,39].In brief, degradation of MAPK3K14, which is a major driver of the non-canonical NFκB pathway is induced by BIRC3, thus loss of BIRC3 function constitutively activates the non-canonical pathway resulting in cell proliferation and survival [39].In CLL, BIRC3 mutations probably indicate a more aggressive course especially in patients treated with fludarabine, cyclophosphamide and rituximab (FCR) [40].Our case further adds evidence that mutations in BIRC3 might influence treatment response and outcome.Our group included only one MYD88 wild-type LPL/WM, which, in accordance with the literature, had a different molecular profile than those with MYD88 mutations [33].Furthermore, morphology in the bone marrow showed osteosclerosis and prominent fibrosis features most likely associated with an ongoing transformation to DLBCL, which synchronously occurred in the lung and central nervous system.
In the SBCL-PC group, in addition POT1, FBXW7, XPO1, and MEF2B had mutations.POT1, FBXW7, and XPO1 mutations belong to subgroups of CLL and are known driver mutations [41][42][43][44] but did not occur LPL, so far.However, shared mutations further support the fact that an alliance might exist between these diseases and that mutations are neither specific nor defining for diagnosis.MEF2B mutations present in subgroups of DLBCL and follicular lymphoma most likely act as an oncogene, but LPL or CLL did not show these mutations so far [45,46].Functionally, MEF2B is a transcriptional activator leading to enhanced transcription of BCL6 and thus promoting lymphomagenesis [45,46].
Our study has some limitations.First, the number of cases is small since LPL/WM is a relatively rare disease.Second, the number of progressive and transformed cases is even smaller, since these are even rarer events.Boiza-Sanchez et al. [47] lately reported on the molecular analysis of eight cases of WM with development into a DLBCL, which revealed shared mutations including MYD88 L265P mutation in the majority of cases.We could only analyze the DLBCL component in one patient with bone marrow infiltration, while in all other patients, high-grade lymphoma occurred extra-medullary and FFPE material was not available in sufficient amounts for further analysis.In this patient, the DLBCL component shared mutations with the preceding LPL/WM with an additional EP300 mutation (see Table 4).
To sum up, assessment of classical morphological and immunohistochemical features of WM in the bone marrow is still a prerequisite in the diagnosis of lymphomas with plasmacytic differentiation.The presence of driver mutations in LPL/WM hints at a higher risk of progressive disease and transformation and a shorter PFS, further supporting the usefulness of an in-depth molecular analysis already upon diagnosis or at least in progressive disease to allocate the appropriate treatment.For reasons of costefficiency in an increasingly troubled healthcare system stratifying patients for further molecular analysis could be done using conventional parameters, including for example atypical immune phenotype, diffuse infiltration pattern, and reduced mast cell component, and should at least be done in treatment-resistant cases.

Fig. 1 Fig. 2
Fig. 1 Classical WM with an admixture of lymphocytes, plasmacytoid lymphocytes and plasma cells (A H&E; inlet highlights plasma cell differentiation; B GIEMSA; plasma cells see arrows); CD20 highlights lymphoma cells within nodular aggregates (C), while plasma cells are located at the edge of the infiltrate (see inlet: immunohistochemistry for vs38c).Plasma cells show clonal restriction by chromogen in situ hybridization (INFORM ISH kappa and lambda, Ventana Medical Systems, Tucson, USA) for the light chains Kappa (D) and Lambda (see inlet) (A, C-D magnification 20 × ; B magnification 40 ×)

Table 4
Molecular characteristics of the seven patients in the study, where more than one biopsy was available for NGS 1