- © 2002 by American Society of Clinical Oncology
Molecular Biology of Anaplastic Lymphoma Kinase–Positive Anaplastic Large-Cell Lymphoma
- Address reprint requests to Jon C. Aster, MD, PhD, Department of Pathology, Brigham and Women’s Hospital, 75 Francis St, Boston, MA 02115; email: jaster{at}rics.bwh.harvard.edu
Abstract
ABSTRACT: Anaplastic large-cell lymphoma (ALCL) provides an excellent example of how molecular insights into tumor pathogenesis are influencing and improving tumor classification. ALCL was described initially as a subtype of T-cell/null-cell lymphoma characterized by unusual tumor cell morphology and the expression of CD30. However, it was soon recognized that a subset of ALCLs contained chromosomal translocations involving anaplastic lymphoma kinase (ALK), a novel receptor tyrosine kinase gene. These rearrangements create chimeric genes encoding self-associating, constitutively active ALK fusion proteins that activate a number of downstream effectors, including phospholipase C-gamma, phosphoinositol 3'-kinase, RAS, and signal transducer and activator of transcription proteins, all of which seem potentially important in cellular transformation. Not all tumors classified as ALCLs have ALK rearrangements and, conversely, ALK rearrangements occur in lymphomas of widely varying morphology. Hence, only molecular markers can reliably identify ALK+ ALCL. The importance of doing so is reflected by clinical studies suggesting that ALK+ ALCLs have a significantly better prognosis than other aggressive peripheral T-cell or B-cell lymphomas, including ALK− ALCLs. The unique molecular pathogenesis of ALK+ ALCL is likely to lead to novel therapeutic approaches directed at specific inhibition of ALK or downstream effectors.
THE IDENTIFICATION AND characterization of genes involved by recurrent chromosomal translocations has played a central role in understanding the pathogenesis of hematolymphoid neoplasia. Because these genetic events typically create oncogenes that initiate cellular transformation and, in collaboration with other genetic or epigenetic changes, dictate tumor biology, the presence of particular translocations often correlates well with clinical behavior and outcome. As a result, tests for specific chromosomal translocations (or their molecular consequences) are playing an increasingly important role in the diagnosis and classification of hematolymphoid neoplasms. This role will undoubtedly increase further with the advent of therapies directed at targets created by tumor-specific genetic aberrations.
One current example demonstrating how molecular pathology is altering and improving lymphoma classification lies within the diagnostic category known as anaplastic large-cell lymphoma (ALCL). This entity was initially recognized by the presence of the Ki-1 antigen, later designated CD30, on a subset of diffuse large-cell lymphomas characterized by bizarre, highly pleomorphic cytologic features and an unusual tendency to grow within lymph node sinuses, thus simulating metastatic solid tumors, such as carcinoma, melanoma, and even seminoma.1 Some ALCLs were thought originally to be derived from cells of monocyte/macrophage lineage, but on subsequent immunophenotyping2-4 and gene rearrangement studies5,6 most were definitively shown to be of T-cell origin. On the basis of these features, the Kiel classification introduced the diagnostic entity Ki-1+ anaplastic large-cell lymphoma in 1988.7
It was evident early on, however, that CD30 was expressed in a diverse group of neoplasms, some of which overlapped morphologically, and immunophenotypically, with ALCL. The Ki-1 antigen was initially detected on a Hodgkin’s disease cell line8 and is uniformly expressed by Reed-Sternberg cells in classic forms of Hodgkin’s disease. ALCLs were observed to often contain tumor cells resembling Reed-Sternberg cells, leading to speculation about the possible relationship of ALCL and Hodgkin’s disease.1 It was also noted that other neoplasms, including cutaneous T-cell proliferations (eg, lymphomatoid papulosis),1,9,10 enteropathy-associated T-cell lymphoma,11 and large T-cell lymphoma arising in the setting of mycosis fungoides,12,13 frequently express CD30 and sometimes bear a morphologic resemblance to ALCL.
The relationship of these diverse entities has been clarified in part by the identification of chromosomal translocations involving the anaplastic lymphoma kinase (ALK) gene in a subset of tumors with ALCL morphology. Most ALK-associated lymphomas present in children or young adults as systemic disease with nodal involvement. ALK rearrangements do not seem to be present in Hodgkin’s disease,14-25 anaplastic cutaneous CD30+ T-cell proliferations, or ALCL-like CD30+ tumors arising secondary to other T-cell malignancies such as mycosis fungoides.17,18,22,26-30 Furthermore, most primary systemic CD30+ ALCLs of older adults also lack ALK rearrangements. Of clinical and diagnostic importance, lymphomas with ALK rearrangements appear to have a significantly better prognosis than other systemic large-cell lymphomas (including systemic ALCLs lacking ALK rearrangements),20,31-35 and a much more varied morphologic appearance than originally appreciated.20,33-37 On the basis of these findings, three groups of neoplasms with ALCL-like morphology are currently recognized, each with a different natural history: primary systemic ALK+ ALCL, primary systemic ALK− ALCL, and primary cutaneous ALCL. This review focuses on the molecular pathogenesis of the most distinctive and best characterized of these entities, ALK+ ALCL.
ALK REARRANGEMENTS IN ALCL
In the late 1980s, several groups noted that some CD30+ ALCLs were associated with a balanced (2;5)(p23;q35) chromosomal translocation.38-42 Subsequent cloning of the chromosomal breakpoints showed that the affected genes were ALK, located at chromosome 2p23, and NPM (encoding nucleophosmin), located at 5q35.43 The t(2;5) is the most common karyotypic aberration involving ALK, being present in approximately 75% of the ALCLs with ALK rearrangements.44 In the remaining approximate 25% of cases, a number of variant rearrangements involving 2p23 are seen, including t(1;2)(q21;p23), inversion 2(p23;q35), t(2;3)(p23;q21), t(2;17)(p23;q23), and t(X;2)(q11-12;p23) (Table 1). These uncommon rearrangements result in the juxtaposition of ALK with one of a diverse collection of partner genes, which include TPM3, encoding a nonmuscle tropomyosin; TFG (TRCK fusion gene), encoding a polypeptide with a predicted coiled:coiled domain; ATIC, which encodes an enzyme, 5-aminoimidazole-4-carboximide-1-β-d-ribonucleotide transformylase/inosine monophosphate cyclohydrolase, that participates in purine metabolism; CLTC, encoding the clathrin heavy-chain gene; and MSN, encoding moesin, a member of the protein 4.1 family of membrane-associated polypeptides.
MOLECULAR CONSEQUENCES OF ALK REARRANGEMENTS
Work from many laboratories has shown that oncogenic rearrangements of receptor tyrosine kinase genes are commonly found in a range of human neoplasms. In the vast majority of such rearrangements, one DNA breakpoint falls within intronic sequences of the involved receptor tyrosine kinase gene and splits the gene in two, dissociating most or all of the 5' extracellular coding sequences and promoter elements from 3' coding sequences for the intracellular signaling domain. The second DNA breakpoint typically falls within the 3' portion of genes that encode structurally diverse polypeptides that appear to share only one feature: the ability to self-associate. Rejoining of DNA in trans thus results in the creation of a chimeric gene with heterologous 5' promoter elements and coding sequences for a self-association domain fused to 3' receptor tyrosine kinase coding sequences for the intracellular signaling domain. Normal activation of receptor tyrosine kinases is dependent on ligand-induced oligomerization.45 In contrast, oncogenic receptor tyrosine kinase fusion proteins self-associate in a ligand-independent fashion, and are therefore constitutively active. In addition, strong chimeric 5' promoter elements often drive high levels of receptor tyrosine kinase fusion gene expression. The net effect of these alterations is to exaggerate and dysregulate otherwise normal downstream signals, which promote cellular transformation. Each of these themes also holds for the ALK rearrangements seen in ALCL, which produce fusion genes encoding self-associating constitutively active ALK tyrosine kinases.
The genomic consequences of the most frequent chromosomal aberration in ALK+ ALCL, the t(2;5), are outlined in Fig 1. Intact ALK spans approximately 315 kb and has 26 exons. Much of the gene consists of two large introns (introns 1 and 2), which together measure approximately 170 kb in size. ALK normally gives rise to an mRNA transcript of about 6.5 kb encoding a 205-kDa type I transmembrane glycoprotein, ALK, which is a member of the insulin receptor superfamily of receptor tyrosine kinases.46,47 ALK is normally expressed only within the developing and mature nervous system17,46,47; it is not expressed in normal lymphoid cells, a point that has important diagnostic implications, as will be described. Alk knockout mice are not noticeably developmentally abnormal,48 and the function of ALK is currently unknown.
Normal NPM spans approximately 25 kb and contains 12 exons,49 which encode a ubiquitous, highly expressed 37-kDa phosphoprotein. NPM exists in cells as a homohexamer that shuttles ribonuclear complexes between the nucleolus and the cytoplasm, thereby playing a role in ribosome assembly.50-52 As might be anticipated given the importance of protein synthesis in growth, the expression of NPM is increased when cell division and growth are stimulated.53-59
In the t(2;5), the chromosomal breakpoints consistently fall within intron 4 of NPM, which spans 911 bp,49 and intron 16 of ALK, which spans 2,094 bp. The reciprocal recombination event produces a NPM/ALK gene on the chromosome 5 derivative and a ALK/NPM gene on the chromosome 2 derivative, both of which have the potential to produce chimeric mRNAs. However, only the NPM promoter is active in lymphoid cells, and thus only the NPM/ALK fusion transcript is detected in ALCLs at appreciable levels.43,60,61 The NPM/ALK fusion transcript encodes a chimeric polypeptide of 80 kDa (p80) consisting of the N-terminal portion (amino acids 1 to 117) of NPM and the C-terminal 563 amino acids of ALK, which includes virtually the entire ALK intracellular domain.43 The portion of NPM in the fusion polypeptide includes the NPM self-association domain,62 which is responsible for the formation of higher order NPM/ALK oligomers in the nucleus, and cytoplasm of NPM/ALK-expressing cells.62 These NPM/ALK fusion proteins are also heavily phosphorylated on tyrosine residues, consistent with constitutive activation of the ALK tyrosine kinase.62,63
Other ALK fusion genes and proteins are less well-characterized biochemically, but each share the capacity for self-association (Table 1). TFG and TPM3 were initially identified as being involved in chromosomal translocations in papillary thyroid carcinomas that juxtapose their coding sequences with those of another transmembrane tyrosine kinase, NTRK1.64 The breakpoints in TPM3, located on chromosome 1q21, occur within intron 7 (as in papillary thyroid carcinomas). DNA translocation leads to the formation of a chimeric TPM3/ALK gene encoding a fusion polypeptide consisting of the N-terminal 221 amino acids of TPM3 and the C-terminal 563 amino acids of ALK.65 The TFG gene is located on chromosome 3q21. Because of alternative splicing, the TFG/ALK fusion gene gives rise to two fusion proteins, TFG/ALKL and TFG/ALKS, which contain the N-terminal 193 or 138 amino acids of TFG, respectively, and the C-terminal 563 amino acids of ALK.66 The involved portions of TPM3 and TFG both contain predicted coiled-coil domains that self-associate, and each produce constitutive tyrosine kinase activity when fused to NTRK1.67,68 TFG/ALK also shows constitutive tyrosine kinase activity in vitro,66 and it is hypothesized, on the basis of its ability to dimerize, that the TPM3 fusion partner has similar effects on ALK.65
Like ALK, ATIC is also located on chromosome 2; as a result, rearrangements usually take the form of chromosome 2 inversions. The chimeric gene encodes a 96-kDa fusion protein that includes an N-terminal 229-amino-acid ATIC dimerization domain. When fused to ALK, this ATIC sequence induces constitutive tyrosine kinase and transforming activities.69,70 Another uncommon fusion gene, CLTC/ALK, encodes a 248-kDa polypeptide consisting of most of the clathrin heavy chain (amino acids 1 to 1,634 of 1,675 total) fused to the C-terminal 563 amino acids of ALK.71 The ALK fusion partner in this rearrangement was misidentified initially as a locus on chromosome 22q11.2, CLTCL, which is highly homologous to CLTC; however, the identification of another rearrangement72 and sequencing of the human genome has unambiguously identified CLTC on chromosome 17q23 as the involved gene. Clathrin exists normally as a trimer, suggesting that it also imparts the ability to self-associate when fused to an ALK. Less is known about the recently identified MSN/ALK fusion gene, which encodes a chimeric protein consisting of the N-terminal portion of moesin fused to intracellular ALK.73 However, other members of the protein 4.1 family, such as ezrin, have the capacity for heterodimerization and homodimerization,74,75 and this feature presumably extends to moesin.
Finally, Mason et al76 has studied the effect of fusion of the intracellular portion of ALK to a leucine zipper dimerization domain derived from translocated promoter region (TPR) protein. TPR participates in a chromosomal translocation that produces a TPR/MET kinase fusion protein in thyroid carcinomas, but is not known to participate in ALK rearrangements seen in sporadic human cancers. Nonetheless, this artificial TPR-ALK fusion protein has transforming activity that is comparable to that of NPM ALK. In total, these data and the analyses of naturally occurring ALK fusion proteins suggest that covalent coupling of any reasonably strong self-association domain to the intracellular portion of ALK will suffice to create an oncoprotein.
Of interest, the NPM self-association domain also participates in the formation of fusion oncoproteins that do not have tyrosine kinase activity. A t(5;17) seen in a small subset of acute promyelocytic leukemias fuses NPM to the retinoic acid receptor alpha gene (RARα), which encodes a DNA-binding transcription factor. The NPM/RARα fusion protein appears to act as a dominant negative inhibitor of normal RARα and its binding partners,77 thereby inhibiting the terminal differentiation of myeloid cells.78 A t(3;5) found in a small subset of myelodysplastic syndromes and acute myelogenous leukemias fuses NPM to MLF1,79 which normally encodes a cytoplasmic protein of uncertain function.80 Hence, the simple ability of NPM to self-associate may contribute to oncoprotein function in several different ways.
TRANSFORMING ACTIVITIES OF ALK FUSION PROTEINS
The effects of NPM/ALK on cultured cells resemble (at least superficially) those of other oncogenic tyrosine kinases such as BCR/ABL and TEL/PDGFβR, which are expressed in typical and atypical forms of chronic myelogenous leukemia, respectively. NPM/ALK confers factor-independent growth on immortalized rodent cell lines, such as NIH-3T3 fibroblasts, 32D myeloid cells, and the pro-B cell line Ba/F3.62,76,81,82 Transforming activity requires both the NPM self-association domain and the tyrosine kinase activity of ALK, as deletions removing the former and mutations abrogating the latter cause loss of function.62,81,82
As might be anticipated from its rather broad transforming activity in vitro, the transforming activity of NPM/ALK in vivo is not limited to T lymphocytes. NPM/ALK induces aggressive B-cell lymphomas when expressed in murine bone marrow progenitor cells.83 More recently, it has been shown that ALK rearrangements may be present in up to 50% of inflammatory myofibroblastic tumors,84-86 rare mesenchymal spindle-cell neoplasms that are typically composed of bland-appearing myofibroblasts mixed with a prominent population of reactive inflammatory cells. Four types of rearrangements that produce ALK fusion proteins have been identified in inflammatory myofibroblastic tumors, two of which involve fusion partners, TPM3 and CLTC,72,87 that are also fused to ALK in some ALCLs (Table 1). The remaining two ALK fusion partners, RANBP2 and TPM4, have only been identified in rearrangements occurring in inflammatory myofibroblastic tumors. Given its apparent capacity to transform a spectrum of cell types, the explanation for the strong association of ALK rearrangements with a specific, limited set of human neoplasms is an intriguing question relevant to many other oncogenic rearrangements as well. Perhaps the ALK gene is particularly prone to DNA breakage in activated T cells, but factors that could produce this type of limited sensitivity to DNA damage are unknown.
The NPM/ALK fusion protein localizes partially to the nucleus and nucleolus,62 as the N-terminal NPM domain retains its normal function as a nucleolar transporter,50 raising the possibility that nuclear or nucleolar localization might play a role in cellular transformation. This appears not to be the case, however, as other ALK fusion proteins show predominantly cytoplasmic localization in either diffuse (TPM3/ALK, TFG/ALK, ATIC/ALK) or finely granular (CLTC/ALK) patterns (Table 1).
Although activated ALK undoubtedly plays an important role in inducing ALK+ ALCL, as with other oncoproteins, it is probably not sufficient to transform otherwise normal lymphoid cells, as judged by several observations. There is a relatively long lag time between bone marrow reconstitution of mice with activated ALK transgenes and the appearance of ALK+ tumors, which only occur in a subset of mice,83 suggesting that secondary events are needed for lymphomagenesis. In addition, several studies using highly sensitive polymerase chain reaction (PCR) tests have detected evidence of ALK fusion genes in reactive lymphoid tissues and peripheral blood,88,89 which may largely explain some reports suggesting that ALK rearrangements occur in Hodgkin’s disease, and primary cutaneous ALCLs and related lesions. At present, the secondary events that collaborate with ALK in cellular transformation are unknown.
SIGNALS DOWNSTREAM OF ACTIVATED ALK
The common feature of diverse ALK fusion proteins is the constitutive activation of ALK signaling (Fig 2). Except for a single variant rearrangement lacking 10 amino acids,90 all fusion proteins include the C-terminal 563 amino acids of ALK. The roles of specific ALK residues in mediating transformation have been evaluated thus far only in vitro. Conclusions drawn from these data must be regarded as preliminary, as in vitro and in vivo requirements for transformation by various oncoproteins, including tyrosine kinases, sometimes differ.91
NPM/ALK tyrosine kinase activity is necessary for cellular transformation.82 A major substrate is NPM/ALK itself, which undergoes transphosphorylation on self-association. There are 21 putative autophosphorylation sites, each serving on phosphorylation as potential docking sites for downstream factors with SH2 or PTB domains, such as phospholipase C-gamma (PLC-γ), SHC, the p85 subunit of phosphoinositol-3'-kinase (PI3K), the adaptor protein GRB2, and other proteins such as insulin receptor substrate 1. Mutation of tyrosine residues 156 and 567 abolishes the binding to insulin receptor substrate 1 and SHC to NPM/ALK, respectively, but has no effect on the transformation of cell lines,81 indicating that these downstream factors are not critical in vitro. These mutations do not affect the binding of the adaptor protein GRB2, which may therefore still contribute to transformation by activating the RAS pathway.
NPM/ALK also activates several members of the signal transducer and activator of transcription (STAT) family, including STAT3 and STAT5,48,92 through mechanisms that are not yet defined. Of note, STAT activation is important in certain other forms of hematolymphoid malignancy driven by tyrosine kinase fusion proteins.93,94 Nieborowska-Skorska et al92 have shown that expression of NPM/ALK in cell lines leads to the phosphorylation and nuclear translocation of STAT5. Conversely, a dominant-negative STAT5B mutant inhibits the antiapoptotic and pro-proliferative effects of NPM/ALK in cell lines, and diminishes the growth of NPM/ALK-expressing cell lines after injection into nude mice.
Other data point to the importance of downstream pathways involving PI3K and PLC-γ. Mutation of a tyrosine residue 664 in NPM/ALK prevents binding of PLC-γ and leads to a failure to transform rodent fibroblasts and Ba/F3 cells. This loss of function is partially rescued by overexpression of PLC-γ,82 suggesting that PLC-γ contributes directly to NPM/ALK’s transforming activities. More recently, investigators have implicated the PI3K/akt pathway in transformation by NPM/ALK.95,96 PI3K is activated by NPM/ALK, and in turn activates the antiapoptotic PKB/akt kinase pathway. The transforming effects of NPM/ALK on murine bone marrow cells and Ba/F3 cells are reversed by treatment with PI3K inhibitors, such as wortmannin. Furthermore, retroviral infection of NPM/ALK-expressing Ba/F3 cells with a dominant-negative PI3K mutation or a dominant-negative akt mutation inhibits the growth and clonogenicity of the infected cells. The same akt mutation also suppresses the tumorigenicity of NPM/ALK-transfected Ba/F3 cells in syngeneic mice.96 Finally, akt phosphorylates and inhibits the proapoptotic protein bad, and apoptosis induced by bad is partially blocked in NPM-ALK transformed cells.95
Other poorly understood signals may also play a part in NPM/ALK-mediated cellular transformation. Greenland et al97 recently observed that expression of NPM/ALK in a human T-cell line, Jurkat, protected these cells from drug-induced, but not FAS-induced, apoptosis. NPM/ALK tyrosine kinase activity was required for this protective effect, but PI3K activation was not, indicating the existence of NPM/ALK survival signals that appear to be independent of the PKB/akt pathway.
The role of CD30, a member of the tumor necrosis factor receptor family that is uniformly expressed by ALK+ ALCL, has also been the subject of considerable investigation, but its contribution to the transformed phenotype remains unclear. Physical association between CD30 and NPM/ALK has been demonstrated63 and is mediated by the intracellular domain of CD30 and the ALK portion of the fusion protein.98 Although this interaction may result in altered subcellular localization of the NPM/ALK fusion protein, NPM/ALK activity does not appear to be affected appreciably. Furthermore, activation of CD30 enhances neither NPM-ALK autophosphorylation nor phosphorylation of PLC-γ.98 In fact, stimulation of the CD30 signaling by cross-linking with immobilized anti-CD30 antibody inhibits the growth of Karpas 299, an NPM-ALK+ ALCL cell line.99 Hence, it is possible that CD30 is merely a marker of ALK+ ALCL, contributing nothing to the transformed phenotype.
DIAGNOSIS OF ALK+ ALCL
Because the ALK promoter is not active in lymphoid cells, the immunodetection of ALK protein in a lymphoid tumor correlates in nearly 100% of cases with the presence of chromosomal translocations involving ALK that place its expression under the control of heterologous promoter elements. As a result, immunohistochemical stains for the intracellular portion of ALK constitute a highly sensitive test for identification of lymphomas with ALK rearrangements (Fig 3A). In tumors expressing the common NPM/ALK fusion protein, staining is usually nuclear and cytoplasmic, whereas variant fusion proteins produce diffuse or punctate cytoplasmic immunoreactivity, making it possible to distinguish NPM from non-NPM rearrangements on the basis of staining patterns (Table 1). Other immunophenotypic markers are also diagnostically helpful. In addition to CD30 (Fig 3B), ALK+ ALCLs often express epithelial membrane antigen,100,101 and are usually positive for one or more markers of cytotoxic T cells,102,103 the presumed cell of origin.
The availability of specific immunohistochemical tests for ALK protein has made it apparent that the histologic spectrum of ALK+ ALCL is broader than initially appreciated. Several morphologies have been described, including common (Fig 3C), lymphohistiocytic, and small-cell variants that differ in cell size and the composition of the accompanying reactive component.101,104,105 Tumor cells may be inconspicuous when they are small in size or present as only a minor cell population, and the small-cell variant may be only focally and weakly positive for CD30, leading to diagnostic difficulties.101,106 In most cases of ALK+ ALCL, characteristic hallmark cells with irregular, eccentric, embryoid, or horseshoe shaped nuclei, variably distinct nucleoli, and large amounts of cytoplasm are present (Fig 3C and 3D).107 However, such cells are not specific for ALK+ ALCL, as they may be seen in systemic ALK− ALCL, cutaneous ALK− ALCL, and even an unusual subset of diffuse large B-cell lymphomas, which as a group appear to have ALK rearrangements rarely, if ever. In addition, ALCLs showing the greatest anaplasia are most often ALK−.20 It is estimated that only 30% to 50% of classic pleomorphic ALCLs are ALK+, whereas ALK positivity is seen in more than 80% of monomorphous ALCLs, more than 75% of small-cell ALCLs, and more than 60% of lymphohistiocytic ALCLs.44 Because of this variability in morphology of ALK+ ALCL, the only reliable diagnostic tests are those that detect ALK rearrangements or their consequences. Because of this, it has been suggested that ALKoma, or ALK+ lymphoma, are more appropriate designations than ALK+ ALCL,34,101,108,109 but for now the name stands.
Although highly sensitive,16,17,20,24,29,35,101,105,110-112 immunohistochemical staining for ALK is not entirely specific for tumors with ALK rearrangements. Rarely, diffuse large B-cell lymphomas express full-length ALK.113 The basis for ALK expression is not known, but the behavior of these unusual B-cell tumors appears to be distinct from that of ALK+ ALCLs,35 and they are considered a variant of diffuse large B-cell lymphoma. Interestingly, ALK is also expressed in neural cell lines, neuroblastomas,114 and rhabdomyosarcomas,47 and is detected by immunohistochemistry in inflammatory myofibroblastic tumors associated with ALK rearrangements.72,84,87 These other types of ALK+ neoplasms are usually easily distinguished by other criteria from ALK+ ALCLs and do not present diagnostic difficulties.
ALK rearrangements can also be identified with reverse transcriptase PCR, DNA-PCR, fluorescence in situ hybridization (FISH), and Southern blotting. As all reported breakpoints in NPM and ALK occur within the same pair of small introns,19,30 it is possible to detect t(2;5)-NPM/ALK rearrangements by either RNA- or DNA-based PCR methods using single sets of ALK and NPM primers.19,30,115 However, current PCR assays will not detect the approximately 25% of tumors with variant ALK rearrangements, and are potentially confounded (if not carefully controlled) by ALK rearrangements occurring at low frequency in nontransformed cells.88,89 FISH performed with break-apart probes that span the ALK locus will identify all tumors with rearrangements (Fig 4), and can be coupled to morphologic inspection to ensure that ALK rearrangements are occurring within the tumor cell population, giving it distinct advantages over PCR-based assays. However, the sensitivity and specificity of ALK immunohistochemistry using the ALK-1 antibody is comparable to FISH.37 Because immunohistochemistry can be applied to routinely fixed and processed tissues, and paraffin-embedded archival tissues, it is the test of choice for identifying tumors with ALK rearrangements.
CLINICOPATHOLOGIC CORRELATIONS
A number of clinical studies have supported the view that ALK+ ALCL is a clinicopathologic entity distinct from ALK− systemic ALCL or ALK− cutaneous ALCL. ALK+ ALCL is most common in the first three decades of life and demonstrates a striking male predominance (male/female ratio, 6.5 in the second and third decades).17,20,34 In contrast, ALK− ALCL is a disease of older individuals and demonstrates no sex predilection (male/female ratio, 0.9).34 Most patients with ALK+ ALCL (70%) present with advanced stage disease (stages III to IV) and have B symptoms.116 It frequently involves both nodal and extranodal sites, including (in decreasing order of frequency) skin, bone, soft tissue, lung, and liver. Unlike Hodgkin’s disease, mediastinal involvement by ALK+ ALCL is uncommon.
Although involvement of soft tissues is not uncommon in ALK+ ALCL, primary presentations limited to the skin are rarely if ever seen, as judged by the inability to detect evidence of ALK rearrangements in the vast majority of cutaneous CD30+ T-cell lymphoproliferative disorders.18,27 For unclear reasons, primary cutaneous CD30+ ALK− lymphoproliferative disorders follow a variable but generally indolent clinical course,117 despite histologic features that may closely resemble those of aggressive systemic ALK− and ALK+ ALCLs.
In most studies, the outcome for patients with ALK+ ALCL has been substantially better than for patients with ALK− ALCL (overall 5-year survival rates of 79% to 88% compared with 28% to 40%, respectively).20,33-35,37 Indeed, it has been suggested that ALK+ ALCL has the best prognosis of any aggressive peripheral B-cell or T-cell neoplasm.118 The relatively young age at presentation may not appear to account for the favorable outcome of ALK+ ALCL, as age was not an independent prognostic indicator in two separate multivariate analyses.34,35 It remains to be demonstrated that this survival advantage extends to the pediatric population, in which one study failed to detect a significant association between ALK expression in ALCL and 2-year event-free survival.16 It does not appear that the type of ALK fusion protein (NPM/ALK or variant) influences outcome.119
FUTURE PROSPECTS
ALK+ ALCL presents exciting therapeutic opportunities. On the basis of the treatment experience with the tyrosine kinase inhibitor imatinib mesylate (STI571, Gleevec; Novartis Pharmaceuticals, East Hannover, NJ) in BCR/ABL-induced hematolymphoid malignancies120,121 and gastrointestinal stromal tumors associated with activating mutations in c-KIT,122 ALK represents an excellent therapeutic target in ALK+ ALCL. Lessons learned in the treatment of other neoplasms with tyrosine kinase inhibitors123 are likely to be applied to the treatment of ALK+ ALCL in the future. Indeed, one recent report using cell lines derived from ALK+ ALCLs has shown that inhibition of NPM/ALK tyrosine kinase activity induces programmed cell death.124 It also seems, at least superficially, that many activated tyrosine kinases engage similar downstream signaling pathways, including RAS (via GRB2), PLC-γ, PI3K/AKT, and STATs, that drive tumor cell growth and survival. Each of these downstream factors is a potential target for therapies that are likely to act synergistically with ALK inhibitors. Thus, although the current outcome in this entity is relatively favorable with standard anthracycline-containing chemotherapy regimens, our improved understanding of the molecular pathogenesis of ALK+ ALCL promises to lead to more specific, effective, and less toxic therapies.
Acknowledgments
Supported in part by grants nos. CA82308 (to J.C.A.) and P30 CA6516 (to J.L.K) from the National Institutes of Health, Bethesda, MD.
ACKNOWLEDGMENT
We thank Paola dal Cin, PhD, and Geraldine Pinkus, MD, for providing molecular cytogenetic and immunohistochemical materials.
- Received December 7, 2001.
- Accepted February 14, 2002.