Portal de Periódicos da CAPES

CHRONIC NEUTROPHILIC LEUKAEMIA: A DISTINCT CLINICAL ENTITY?

2002; Wiley; Volume: 116; Issue: 1 Linguagem: Inglês

10.1046/j.1365-2141.2002.03234.x

1365-2141

John T. Reilly,

Chronic Lymphocytic Leukemia Research

Chronic neutrophilic leukaemia (CNL) has been reported as a rare disorder characterized by persistent mature neutrophilia in association with toxic granulation and Döhle bodies, hepato-splenomegaly, elevated serum vitamin B12 and transcobalamins, hyperuricaemia and a raised neutrophil alkaline phosphatase. These features, together with the lack of a basophilia or monocytosis, and the absence of bcr-abl transcripts, distinguishes CNL from chronic myeloid leukaemia (CML), atypical chronic myeloid leukaemia (aCML) and chronic myelomonocytic leukaemia (CMML), as defined by the French–American–British (FAB) Cooperative Group (Bennet et al, 1994). However, despite CNL being first reported 80 years ago (Tuohy, 1920) and the fact that nearly 100 alleged cases have been documented, a clear understanding of the disease's pathogenesis and natural history is awaited. The literature, which consists essentially of isolated case reports published in a wide range of journals, is frequently confusing and often incomplete, with the result that CNL's natural history and prognosis remain obscure. Despite these limitations, the Steering Committee for the World Health Organization Classification of Neoplastic Diseases have recently acknowledged CNL to be a distinct myeloproliferative disorder (Harris et al, 1999). The present author has critically reviewed the available literature and concludes that the term 'chronic neutrophilic leukaemia' has been used to encompass a number of distinct pathological entities. Indeed, the problems of nomenclature have been further compounded by recent reports of rare p230 bcr-abl CML patients with phenotypes that appear to closely mimic chronic neutrophilic leukaemia. The aim of the present article is to highlight the diagnostic difficulties, to discuss the differential diagnoses and to emphasize the preliminary conclusions of Ben-Tal et al (1997) that many reported cases of CNL would be better classified as (i) neutrophilic-chronic myeloid leukaemia (N-CML), (ii) plasma cell dyscrasia-associated neutrophilia, or (iii) chronic neutrophilic leukaemia with dysplasia. It is only by the careful exclusion of these entities, as well as the more common leukaemoid reaction, that the clinical features, natural history and pathogenesis of 'true' CNL can be established. The cause of a neutrophilia is usually obvious from a careful history and examination, and is typically secondary to either chronic infection or an underlying malignancy (Coates & Baehner, 1995). Neutrophilia is frequently seen in large cell lung cancer (Ascensao et al, 1987), although it has been reported in association with a variety of tumours, particularly in the presence of marrow involvement (Eichenhorn & Van Slyck, 1982). The leucocytosis isusually modest, in the range of 12–30 × 109/l, although levels as high as 100 × 109/l have been reported (Eichenhorn & Van Slyck, 1982). It is likely that such leukaemoid reactions aremediated by tumour-related cytokines, including granulocyte–monocyte colony-stimulating factor (GM-CSF) (Wetzler et al, 1993; Watanabe et al, 1998), interleukin 1α (Chen et al, 1995), interleukin 6 (Hisaoka et al, 1997) and granulocyte colony-stimulating factor (G-CSF) (Sato et al, 1994; Chen et al, 1995; Hisaoka et al, 1997; Ohbayashi et al, 1999). Nevertheless, the fact that no detectable cytokines can be found in other cases suggests that carcinoma-related neutrophil leucocytosis may be induced by mechanisms other than those described above (Saussez et al, 1997). Typically, tumour-associated leukaemoid reactions reflect an aggressive underlying clinical course and both diagnoses are often established simultaneously. An occult malignancy, however, should be sought in all cases of apparent CNL, as leukaemoid reactions have been reported to predate the diagnosis of carcinoma by many years. Ferrer et al (1999), for example, described a patient in which a persistent neutrophilia was detected 4 years before the diagnosis of a colorectal carcinoma and which rapidly normalized followingtumour resection. A diagnosis of CNL can only be made with confidence once bcr-abl-positive disease has been excluded, a fact even more germane following the reports of rare patients presenting with a neutrophilia and an associated p230 bcr-abl transcript. The leukaemia-associated bcr-abl oncogenes vary in the amount of the bcr gene that is included within the chimaera (Melo, 1996) and recent evidence suggests that such structural variation may influence the disease phenotype (Quackenbush et al, 2000). p190 bcr-abl, for example, which derives from an e1a2 transcript and contains the dimerization-binding SH2 and serine/threonine kinase domains of bcr, is associated with an aggressive form ofacute lymphoblastic leukaemia (Ph′-positive ALL). Incontrast, p210, resulting from either an e13a2 or an e14a2 transcript that contains additional bcr sequences coding for the Pleckstrin homology (PH) and Dbl-like domains, is associated with approximately 95% of 'classic' CML (Shepherd et al, 1995). A greater degree of bcr-abl 'genotype–phenotype' subtlety may exist, however, as indicated by recent reports of rare patients exhibiting a p230-type bcr-abl gene. Pane et al (1996) reported three cases of Ph1-positive chronic myeloid leukaemia that fulfilled the criteria of chronic neutrophilic leukaemia and which exhibited a bcr-abl rearrangement with a breakpoint between exons e19 and e20 of the bcr gene (designated the µ-bcr region). This entity was given the term neutrophilic-CML (N-CML). The same group first described such a breakpoint in 1990 (Saglio et al, 1990) and the authors now believe that their initial two cases would be better reclassified as N-CML, rather than 'classic' CML (Pane et al, 1996). These five cases appear to have a much more benign course than classic CML, with a lower white cell count, a lower proportion of circulating immature granulocytes, milder anaemia, less prominentsplenomegaly, a normal leucocyte alkaline phosphatase (LAP) score and a low propensity to acute transformation. However, although five of the 10 published cases of p230 CML fit the classification of chronic neutrophilic leukaemia (Melo, 1996), it should be stressed that at least two cases appear identical to 'classic' CML (Briz et al, 1997; Wilson et al, 1997), and that several exhibited high platelet counts and would be classified better as Ph′-positive essential thrombocythaemia (ET) (Yamagata et al, 1996; Mittre et al,1997). The p230 bcr-abl translation product carries an additional 180 amino acids, encoded by 540 bp of 'extra' bcr sequence, compared with the p210 protein. Specifically, p230 bcr-abl contains the calcium-phospholipid binding (CalB) domain and the first third of the domain associated with GTPase activating activity for p21rac (GAPrac). Like the p190 and p210 proteins, p230 bcr-abl has tyrosine kinase activity (Wada et al, 1995) and, although only the first third of the bcr GAPrac domain is included, it may function cooperatively with the CalB domain as both domains are frequently found together (Ponting & Parker, 1996). It is possible therefore that the additional bcr sequences within p230 bcr-abl, especially the GAPrac domain, may function to partially abrogate the properties of activated p21 Rac and result in less disruption of granulocytic differentiation (CNL-like disease) and/or more potent megakaryocytic expansion (ET-like disease)(Melo, 1997; Quackenbush et al, 2000). Interestingly, in vitro evidence in support of p230 bcr-abl producing a more benign phenotype has been provided by Quackenbush et al (2000) who reported that primary mouse bone marrow cells expressing p230 require exogenous haematopoietic growth factors for optimal growth, in contrast to either p190 or p210 expressing cells. Clearly, the identification and detailed clinical study of additional patients with p230 bcr-abl disease is required before the true biological significance of p230 expression can be determined. Two further cases of CNL have been documented in association with the Philadelphia-chromosome, but in neither case was the breakpoint determined (Sanada et al, 1985; Christopoulos et al, 1996). It is important therefore to screen patients with apparent CNL for bcr-abl transcripts and for p230 bcr-abl inparticular, as, if present, the cases should be classified as neutrophilic-chronic myeloid leukaemia (N-CML) rather than as chronic neutrophilic leukaemia. This distinction is therapeutically relevant, especially following the recent development of specific bcr-abl inhibitors (Druker et al, 1999). Chronic neutrophilic leukaemia is a rare disease and yet, rather surprisingly, more than 20 cases have been reported in association with plasma cell dyscrasias, either monoclonal gammopathy of undetermined significance (MGUS) or multiple myeloma (MM) (Tursz et al, 1974; Carcassonne et al, 1977; Vorobiof et al, 1978; Naparstek et al, 1980; Franchi et al, 1984; Watanabe et al, 1984; Mehrotra et al, 1985; Lewis et al, 1986; Saigo et al, 1986; Kubo et al, 1989; Zoumbos et al, 1989; Rovira et al, 1990; Dieguez et al, 1992; Masini et al, 1992; Troussard et al, 1992; Florensa et al, 1993; Cehreli et al, 1994; Mori et al, 1995; Ito et al, 1996). Attention has been drawn to the disproportionate excess of lambda light chain restriction in this subset of patients (Standen et al, 1990) and the fact that the neutrophilia may precede myeloma by many years (Rovira et al, 1990). Clearly, the high frequency of this unusual involvement of myeloid and lymphoid lineages raises the question of whether the neutrophilia represents the existence of a genuine myeloproliferative disorder or is merely a secondary, or reactive, phenomenon. A number of observations suggest that, at least in some cases, the neutrophilia might be a reactive phenomenon. Ito et al (1996) emphasized the fact that the survival of patients with CNL and MGUS appears to be much better than those cases associated with multiple myeloma. In addition, they noted a case of spontaneous remission of CNL during a 12-year follow-up of MGUS, while Standen et al (1990) have stressed the rarity of chromosomal abnormalities in this group of patients. Nagai et al (1996) were the first to provide compelling evidence of an associated leukaemoid reaction, by demonstrating a relationship between neutrophil count and plasma granulocyte colony-stimulating factor (G-CSF) levels in a patient with myeloma. Pulsed steroid therapy was shown to rapidly reduce the high G-CSF concentrations and that this coincided with a decline in the patient's neutrophil count. Interestingly, G-CSF mRNA is expressed in most bone marrow cells of patients with multiple myeloma (Portier et al, 1993), although the phenotype of the cells producing G-CSF in vivo, i.e. myeloma cells and/or stromal cells, remains to be elucidated. Nevertheless, it is noteworthy that high levels of G-CSF mRNA were observed in peripheral blood cells from patients with plasma cell leukaemia, suggesting that G-CSF could be produced from the tumour cells themselves (Portier et al, 1993). Physiological G-CSF production is mainly by monocytes, macrophages, fibroblasts and endothelial cells (Demetri & Griffin, 1991), and its production can be induced by cytokines, including interleukin 1(IL-1) and tumour necrosis factor (TNF) (Koeffler et al, 1987; Fibbe et al, 1988). IL-1 therefore could be one of the inducers of G-CSF in patients' cells as Portier et al (1993) reported a strong correlation between IL-1 and G-CSF gene expression. Further support for the 'reactive' hypothesis has been provided by X-linked polymorphism analysis. Standen et al (1993), for example, confirmed the polyclonal nature of myeloid cells in a single patient, while Stevenson et al (1998) demonstrated polyclonal granulopoiesis in a case that progressed to diffuse myelofibrosis. Clearly, many cases of 'apparent' CNL with an associated paraproteinaemia are probably leukaemoid reactions and it is therefore important that an underlying monoclonal gammopathy is sought in all suspected cases of neutrophilic leukaemia. It is well recognized that, as a result of clonal evolution, some myelodyplastic syndromes may develop myeloproliferative features. Merlat et al (2000) described a case of 5q-refractory anaemia that progressed after 3 years to a CNL-like condition, while Pascucci et al (1997) documented a CNL transformation that was associated with del(11)(q14). Similarly, typical myeloproliferative disorders may develop dysplastic features, either as part of the disease or as a result of exposure to mutagenic agents (Bain, 1999). In addition, there are an important group of 'overlapsyndromes', for example chronic myelomonocytic leukaemia, atypical chronic myeloid leukaemia, juvenile chronic myeloid leukaemia and childhood monosomy 7 syndrome, which possess both dysplastic and proliferative features. Zoumbos et al (1989) first drew attention to the fact that some patients with apparent CNL possess features of myelodysplasia, especially of the granulocytic series. They proposed the term chronic neutrophilic leukaemia with dysplasia (CNL-D) and suggested that such cases were distinct from both CNL and the other 'overlap syndromes'. Further cases of CNL with marked myelodysplasia were reported by Cervantes et al (1988) and Ota et al (2000), while at least one of the patients reported by Zittoun et al (1994) would be more appropriately re-classified as CNL-D. The nosology of neutrophil proliferations is further complicated by the close relationship of CNL to other chronic myeloproliferative disorders, including polycythaemia vera (PV) and idiopathic myelofibrosis (IMF). At least six cases of PV, for example, have been reported to evolve features indistinguishable from CNL (Lugassy & Farhi, 1989; Iurlo et al, 1990; Fujisawa et al, 1992; Harada et al, 1993; Higuchi et al, 1999), suggesting that derangement at the haematopoietic stem cell level can lead to a CNL-like disorder (Higuchi et al, 1999). The interval from diagnosis of PV to transition ranged from 5 to 17 years and was accompanied by a decrease in haemoglobin in all cases. The number of reports is too small to assess the role of therapy administered during the PV phase, but all cases received cyto-reductive agents. Further diagnostic difficulties may occur when cases of IMF present with a polymorphonuclear leucocytosis. Indeed, Thiele et al (1996) have reported the existence of a hypercellular, or prefibrotic, stage of myelofibrosis that is characterized by abnormal megakaryocyte morphology. It is possible therefore that the few cases of CNL reported to have significant fibrosis would be better classified as proliferative-phase myelofibrosis rather than 'true' CNL (Hirayama et al, 1994; Zittoun et al, 1994). The first case of CNL was reported by Tuohy (1920), although Emil-Weil & Sée (1932) cited two probable cases from earlier French and German literature. The term 'chronic neutrophilic leukaemia' was first used by Tanzer et al (1964), while Jackson & Clarke (1965) reported a case of 'neutrophilic leukaemia' a year later. These early cases, however, as well as many subsequent reports, often lack sufficient data to support a definite diagnosis. A critical review of the literature by the author has resulted in only 33 cases in which the above differential diagnoses can be excluded and which would appear to fulfil the criteria for 'true' CNL (see Table I). An analysis of these reports, together with data from an unpublished case of the author's, reveals a 2:1 male:female ratio, with a mean age at diagnosis of 62·5 years (range 15–86). The actuarial survival curve of the series (Fig 1) demonstrates an overall median survival of 30 months and a 5-year survival of only 28%. Transformation to acute myeloid leukaemia appears common, occurring in 21·2% (7/33) of the reported cases. Kaplan–Meier survival plot for 33 patients with chronic neutrophilic leukaemia demonstrating an overall median survival of 30 months and a 28% 5-year survival. Most patients presented with a normal haemoglobin and platelet counts below 100 × 109/l were rare. All cases demonstrated a peripheral leucocytosis, with a mean diagnostic leucocyte count of 54·3 × 109/l (range 10–172 × 109/l). The increase in white cell count was almost exclusively at the segmented and band stage of development (mean 46·3 × 109/l; range 7·5–151·0 × 109/l).Metamyelocytes, myelocytes and nucleated red cells were infrequent and myeloblasts were rarely present. Vitamin B12 levels were increased in all patients in whom it was tested, while low neutrophil alkaline phosphatase scores (NAP) were the exception (Hasle et al, 1996). Uric acid concentrations were frequently increased and episodes of gout were documented in several patients (Tanzer et al, 1964; You & Weisbrot, 1979). The bone marrow biopsies were uniformly hypercellular, with all cases showing an essentially packed marrow. The biopsies were consistent with the bone marrow aspirate findings, showing a marked granulocytic proliferation with no unusual distribution pattern. Low serum G-CSF levels have been documented by a number of groups (Ohtsuki et al, 1992; Hirayama et al, 1994; Saitoh & Shibata, 1996; Higuchi et al, 1999), suggesting that the neoplastic granulopoiesis can exert a suppressor effect on G-CSF synthesis. Interestingly, CML patients, in contrast to those with CNL, invariably have low LAP scores and yet both disorders have significantly low G-CSF levels (Saitoh & Shibata, 1993, 1996). It would seem therefore that, while serum G-CSF may be an important cause, it is unlikely to be the sole reason for the induction and/or stabilization of LAP activity (Rambaldi et al, 1990). Neutrophil function can be elevated, as defined by increased phagocytosis, superoxide production and nitroblue tetrazolium (NBT) reduction activity, although paradoxically intracellular bactericidal activity may be reduced (Mehrotra et al, 1985; Oogushi et al, 1989; Ohtsuki et al, 1992). These data and the frequent presence of toxic granulation and Döhle bodies suggests that the neutrophils in CNL are in an activated stage (Ohtsuki et al, 1992). The clonal nature of CNL has been confirmed in individual cases on the basis of karyotypic abnormalities, X-inactivation patterns and progenitor cell assays, and supports the inclusion of CNL as a true myeloproliferative disorder. Cytogenetic abnormalities occurred in 37% of cases (12/32) and include trisomy 8 (Orazi et al, 1989), trisomy 21 (Hasle et al, 1996), del(20q) (Orazi et al, 1989; Harada et al, 1993; Matano et al, 1997) and t(1;20) (Mehrotra et al, 1985). The most frequent abnormality appears to be involvement of chromosome 20, reported in four cases, of which three had loss of the long arm (DiDonato et al, 1986; Matano et al, 1997; Frank et al, 2000). Importantly, three cases exhibited cytogenetic abnormalities before the commencement of therapy. 20q deletions are not restricted to CNL and have been associated with other chronic myeloproliferative disorders, including polycythaemia vera (Diez Martin et al, 1991) and idiopathic myelofibrosis (Demory et al, 1988; Mertens et al, 1991; Reilly et al, 1994). These findings suggest that putative tumour suppressor gene(s), associated with the pathogenesis of chronic myeloproliferative disorders, including CNL, are located on the long arm of chromosome 20. The critically deleted region has been narrowed to a 18 cM interval between SRC and D20S17, an area that contains three possible genes, namely, PLC1, hepatocyte nuclear factor-4 and topoisomerase 1 (see review Asimakopoulos & Green, 1996). In contrast, a normal karyotype has been recorded in the majority of well-documented cases (see Table I), suggesting that the primary genetic event is submicroscopic and is not detected by conventional cytogenetic analysis. Indeed, Elliot et al (2001) have suggested that the chromosomal abnormalities reported to date may be related to cytogenetic evolution and represent secondary events in the pathogenesis of CNL. The establishment of clonality in patients lacking cytogenetic abnormalities has not been convincinglydocumented. Kwong & Cheng (1993) purported to have demonstrated monoclonal haematopoiesis in a patient using the methylation status of the X-linked hypoxanthine phosphoribosyl transferase gene, following restriction endonuclease digestion. However, because skewed X-chromosome inactivation patterns may occur, Gale et al (1994) have stressed the importance of using T-lymphocytes as a germ-line control. Unfortunately, the data by Kwong & Cheng (1993) failed to include the T-lymphocyte pattern and, as a result, the clonality must remain in doubt. Nevertheless, the author has demonstrated myeloid clonality and T cell polyclonality in a 48-year-old female patient (case 27), using the related human androgen receptor gene-based assay (HUMARA) (unpublished observations). Clearly, the determination of X-linked gene methylation patterns can provide valuable diagnostic information, although the assays are limited to informative female patients. The question of the stage at which myeloid derangement occurs in CNL is not yet clear. Rare transformations to myeloid metaplasia (Zittoun et al, 1994), the frequent development of acute leukaemia (Katsuki et al, 2000), and the fact that a number of patients with polycythaemia have been reported to transform to CNL (Shirakura et al, 1979; Lugassy & Farhi, 1989; Iurlo et al, 1990; Foa et al, 1991; Fujisawa et al, 1992; Harada et al, 1993; Higuchi et al, 1999) suggests, at least in some cases, that the abnormality may be at the haematopoietic stem cell level. Yanagisawa et al (1998), however, reported a patient whose haematopoietic progenitors spontaneously formed colonies consisting of large numbers of mature granulocytes that possessed the same cytogenetic abnormality as that found in the bone marrow. In contrast, the karyotype of the granulocytic-macrophage and macrophage colony-forming units, as well as the erythroid burst-forming unit, were normal. Froberg et al (1998), using fluorescence in situ hybridization (FISH) analysis, also localized the cytogenetic defect to the granulocyte population in a patient with an 11q14 deletion. The level of neoplastic involvement in CNL is therefore likely to be heterogeneous, although the mechanism(s) that lead to neutrophil accumulation remain unknown. Preliminary data suggests that malignant neutrophils may exhibit a reduced susceptibility to spontaneous apoptosis, with a resultant survival advantage (Hara et al, 2001). The optimal therapy of CNL remains unclear. Splenic irradiation and splenectomy have been used to reduce tumour bulk and relieve abdominal discomfort (You & Weisbrot, 1979), although the latter may aggravate the degree of neutrophilia (Hasle et al, 1996). Oral chemotherapeutic agents, including busulphan and hydroxyurea, while not curative, have been demonstrated to control the excessive leucocytosis and maintain a chronic stable phase (Lorente et al, 1988; Kwong & Cheng, 1993; Hirayama et al, 1994; Zittoun et al, 1994; Hasle et al, 1996). Alpha interferon can rapidly reduce tumour mass, although whether it can restore normal haematopoiesis, as documented for a proportion of CML cases, is unknown. Interestingly, alpha interferon may enhance the impaired natural killer (NK) and phagocytic functions and, as a result, could theoretically prevent the onset of fatal infections that are commonly observed in CNL (Meyer et al, 1993). Given the potential for leukaemic transformation and progressive refractory neutrophilia, a more aggressive approach may be appropriate for younger patients. Hasle et al (1996), for example, reported a 15-year-old girl withtrisomy 21 who remains in complete remission six-and-a-half years after allogeneic bone marrow transplantation. CNL is a rare, but distinct, entity within the chronic myeloproliferative disorders, characterized by a sustained mature neutrophilia, hepatosplenomagely, a high LAP score, lack of detectable bcr-abl transcripts, elevated serum B12 and uric acid levels, as well as a low G-CSF concentration. However, the diagnosis remains difficult (Fig 2) and a leukaemoid reaction due to an underlying infection or malignancy, including plasma cell dyscrasias, needs to be rigorously excluded. Careful documentation of future cases is required to enable a better understanding of the disease's natural history and to determine the optimal therapeutic approach. The author thanks Dr D. A. Winfield and Dr K. Muta for providing updated survival data for cases 1 and 31, respectively, and to Mr F. Abu-Duhier for performing the survival analysis.