Mast cell density and its clinical relevance in Waldenström's macroglobulinemia

Abstract The presence of numerous mast cells (MCs) mixed with tumor cells in the bone marrow (BM) is a hallmark of the diagnosis of Waldenström's macroglobulinemia (WM). MCs have been shown to support lymphoplasmacytic cell growth, but there is thus far no demonstration of the prognostic impact of BM MC density in WM. We investigated BM MC density by sensitive and specific digital quantification, allowing the analysis of a large area infiltrated by BM tumor cells. A total of 65 WM patients were investigated, including 54 at diagnosis and 11 at relapse. Tryptase and CD20 immunohistochemisty staining was performed on contiguous sections of deparaffinized BM trephine biopsies. After numerization of each section, the BM surface area was manually marked out, excluding the bone framework and adipocytes to limit the analyses to only hematopoietic tissue. MCs were assessed using a digital tool previously used to quantify immune‐cell infiltrates on tumor‐tissue sections. Deep next‐generation sequencing and allele‐specific PCR were used to explore the MYD88 and CXCR4 mutational status. MC density was heterogeneous among the WM patients. An optimal MC density threshold (> 56 MC.mm–2) was defined according to ROC curve analysis of overall survival. A higher MC density (> 56 MC.mm–2) was associated with greater BM involvement by WM lymphoplasmacytic cells and less hepatosplenic involvement (p = 0.023). Furthermore, MC density significantly correlated with a higher ISSWM score (p = 0.0003) in symptomatic patients. Patients with a higher MC density showed shorter median OS (56.5 months vs. nonreached, p = 0.0004), even in multivariate analysis after controlling for other predictive variables, such as age, ISSWM score, and CXCR4 mutational status. In conclusion, MC density can be accurately measured in WM patients using a specific digital tool on well‐outlined hematopoietic tissue surfaces. High MC density is associated with aggressive features and a poor clinical outcome, emphasizing the need for further investigation of the involvement of MCs in the pathophysiology of WM.

excluding the bone framework and adipocytes to limit the analyses to only hematopoietic tissue. MCs were assessed using a digital tool previously used to quantify immunecell infiltrates on tumor-tissue sections. Deep next-generation sequencing and allelespecific PCR were used to explore the MYD88 and CXCR4 mutational status. MC density was heterogeneous among the WM patients. An optimal MC density threshold (> 56 MC.mm -2 ) was defined according to ROC curve analysis of overall survival.
A higher MC density (> 56 MC.mm -2 ) was associated with greater BM involvement by WM lymphoplasmacytic cells and less hepatosplenic involvement (p = 0.023). Furthermore, MC density significantly correlated with a higher ISSWM score (p = 0.0003) in symptomatic patients. Patients with a higher MC density showed shorter median OS (56.5 months vs. nonreached, p = 0.0004), even in multivariate analysis after controlling for other predictive variables, such as age, ISSWM score, and CXCR4 mutational status. In conclusion, MC density can be accurately measured in WM patients using a specific digital tool on well-outlined hematopoietic tissue surfaces. High MC density is associated with aggressive features and a poor clinical outcome, emphasizing the need for further investigation of the involvement of MCs in the pathophysiology of WM.

K E Y W O R D S
mast cells, tumor biology, Waldenström's macroglobulinemia

INTRODUCTION
Waldenström's macroglobulinemia (WM) is characterized by lymphoplasmacytic infiltration of the bone marrow (BM), along with the presence of a serum monoclonal IgM. MYD88 L265P and CXCR4 mutations have been reported in > 90% and ≈ 25% of WM cases, respectively [1][2][3][4][5][6]. The MYD88 L265P mutation results in the constitutive activation of NFκB by Bruton's tyrosine kinase and is thought to be a clonal driver mutation [1]. The CXCR4 mutations are mostly nonsense or frameshift mutations that occur most often in the S338 position.
They result in truncation of the cytoplasmic portion of the receptor for the chemokine CXCL12, similarly to CXCR4 mutations seen in WHIM (warts, hypogammaglobulinemia, infections, and myelokathexis) syndrome. Such truncation leads to impaired internalization of the receptor after ligation and results in prolounged activation [5]. CXCR4 WHIM mutations are subclonal and associated with greater BM involvement, more symptomatic disease, and higher genomic complexity [6][7][8].
One historical hallmark for WM histological diagnosis is the presence of numerous mast cells (MCs) mixed with the tumor cells in the BM [9]. MCs are myeloid-derived cells that are widely disseminated throughout all tissues and act as sentinels of the surrounding environment. They store and can release a wide spectrum of biologically active mediators after activation, leading to their central role in allergic diseases. The presence of numerous MCs in human tumors is well described and their pleiotropic molecule production is thought to explain their pro-oncogenic roles (recently reviewed in [10,11]).
In hematological malignancies, MCs have been shown to be associated with progression and poor (or even good) prognoses in various Hodgkin's and non-Hodgkin's lymphomas [12][13][14][15][16][17][18][19]. In WM, MCs have been shown to support lymphoplasmacytic cell growth through CD154/CD40 signaling [20]. MC density in WM has already been in heatlhy subjects [21]. In this series, MC density was shown to significantly increase after treatment for nonresponders, to remain stable for minor responders, and to significantly decrease for major responders.
There has thus far been no clear demonstration of the clinical relevance nor prognostic impact of BM MC density in WM. We thus investigated BM MC density by sensitive and specific digital quantification, allowing the analysis of a large area of the BM infiltrated by tumor cells, to assess its clinical relevance in WM.

Evaluation of MC density
Immunostaining was centrally performed at the Clermont-Ferrand pathology laboratory using a monoclonal mouse anti-human CD20 antibody (clone L26; Dako, Carpinteria, CA, USA, 1/100 dilution for 1 h) to stain WM tumor cells and a monoclonal mouse anti-human tryptase antibody (clone AA1; Dako, 1/1000 dilution for 24 min) to stain MCs.
Anti-CD20 and antitryptase immunostaining was performed on two contiguous sections of BM trephine biopsies to assess the tumor infiltration characteristics and compare them to the MC infiltration pattern.
All computer-based analyses were conducted using the "Immunoscore module" from Definiens Developer XT, which has already been used to quantify immunological infiltrates within tumors [22]. After digitization, each antitryptase immunostained section was For four patients, MC density was evaluated on two to four consecutive available BM trephine biopsies using the same protocol.
Fifty-three samples were assessed by deep next-generation sequencing for MYD88 and CXCR4 mutations, as previously described [23], allowing better assessment of the percentage of allele variants with 1% sensitivity. Two samples were assessed by allele-specific PCR for the MYD88 L265P mutation only.

Statistical analyses
The therapeutic requirements and time to reach the endpoints were defined according to published recommendations [4,24,25]. Statistical analyses were performed using SPSS Statistics v22 (IBM), PRISM v8.0 (Graphpad), and/or R software [26]. The Pearson Chi-square, twosided Fisher, Mann-Whitney, and Kaplan-Meier tests with Log-rank and Cox multivariate models were applied to the data as appropriate.
MC density was first treated as a quantitative parameter. Then, a sensitivity analysis was conducted to categorize MC density according to its statistical distribution (i.e., median and interquartile range) and then ROC curve analysis, applying Youden's index, to determine the optimal threshold based on overall survival (OS).

The distribution of MC density and its correlation with deep tumoral infiltration
The MC density ranged from 6.7 to 487 MC.mm -2 (mean 106.1 MC.mm -2 , median 79.9 MC.mm -2 ). The optimal threshold was defined to be 56 MC.mm -2 by ROC curve analysis based on OS, pinpointing two populations with a high versus low MC density (Figure 2A).

The association of MC density with clinical and biological features
The clinical and biological features of the patients are shown in Table 1B according to their MC density. There was a statistical association between MC density and the ISSWM prognosis score (p < 0.001) ( Figure 2B). Patients with a high MC density had significantly less hepatosplenic involvement (p = 0.023) but higher anemia (< 115 g.L -1 ; p = 0.031) and more thrombocytopenia (< 100 g.L -1 ; p = 0.037). High MC density was also statistically associ-

3.3
The association of MC density with the MYD88 nor CXCR4 mutational status

Evaluation of MC density at successive timepoints
We evaluated MC density at various timepoints for four patients by examining one to four BM trephine biopsies during the clinical course of the disease (Figure 4). MC density increased upon relapse for all four patients.

DISCUSSION
Interactions between lymphoplasmacytic cells and the microenvironment, including MCs, are known to support their survival and proliferation. This study is the first to evaluate MC density using a sensitive and specific semi-automated tool on large fields of well-outlined hematopoietic tissue of the BM of Waldenström patients.

CONFLICTS OF INTEREST
The authors have no conflicts of interest to declare.