The concept of mucosa-associated lymphoid tissue (MALT) has progressively developed over the past 10 years with respect to the lymphoid component observed in various organs that do not correspond to peripheral sites of the immune system ( Isaacson & Wright, 1983, 1984; Isaacson & Spencer, 1987; Isaacson, 1995a). Two types of MALT tissue can be identified ( Isaacson, 1994; Pileri et al, 1994 ), the ‘native’ type which consists of the lymphoid tissue physiologically present in the gut, and the ‘acquired’ MALT type which develops after phlogistic events in response to either infectious conditions such as Helicobacter pylori gastritis in the stomach, or autoimmune processes like Hashimoto's thyroiditis and myo-epithelial sialadenitis (MESA), usually due to Sjogren's syndrome. In the context of these prolonged lymphoid reactive proliferations a pathological clone can progressively replace the normal lymphoid population, giving rise to a malignant lymphoma of MALT type. Low-grade MALT lymphoma was initially interpreted as a form of follicular centre cell lymphoma but is now considered a separate entity originating from the marginal-zone B cells, distinct from other low-grade lymphomas ( Harris et al, 1994 ; Isaacson, 1995a).
The MALT lymphoma concept, as it has come to be known, has grown to include a number of low-grade extranodal B-cell lymphomas which share similar clinical, pathological and molecular features. Gastric MALT lymphoma is the most common and the most frequently studied and can be taken as representative of the group as a whole.
The clinicopathological features of gastric lymphomas appear to be more closely related to the structure and function of MALT than of peripheral lymph nodes. In the gastrointestinal tract MALT consists of four components, the organized mucosal lymphoid nodules (distributed throughout the small intestine, the appendix, and the colorectum, and concentrated in the terminal ileum to form the Peyer's patches), the lamina propria, the intra-epithelial lymphocytes which may be involved in antigen recognition, and the mesenteric lymph nodes ( Isaacson, 1995b).
In contrast to peripheral lymph nodes, which are adapted to dealing with antigens carried from afferent lymphatics, the lymphoid tissue of the gut and other mucosae appears to have evolved to protect mucosal tissue which is directly in contact with antigens in the external environment, thereby suggesting that the mucosae could be considered a common mucosal immune system ( McDermott & Biemenstock, 1979). Following antigenic stimulation MALT B cells may leave the mucosa via efferent lymphatics, pass through the mesenteric lymph nodes into the thoracic duct and enter the circulation. These B cells possess specific homing receptors that recognize site-specific addressins on the endothelium that mediate preferential ‘home back’ to the gut mucosa as plasma cells or memory B cells ( Berg et al, 1989 ). A preferential expression of the mucosal homing receptor integrin α4β7 was recently demonstrated in gastrointestinal lymphomas ( Drillenburg et al, 1997 ). The lymphocytes' ‘homing’ mechanism is still incompletely understood ( Butcher & Pickert, 1996) but it could be relevant for clarifying the behaviour of B-cell lymphomas of MALT type which tend to remain localized in the mucosal sites ( Isaacson & Spencer, 1995; Drillenburg et al, 1997 ).
Paradoxically, MALT lymphoma usually arises as a consequence of a pre-existing disorder in sites such as the stomach, salivary gland, and thyroid where MALT is not normally present. This is well illustrated by MALT lymphomas of the salivary gland and thyroid which arise in the context of MESA and Hashimoto's thyroiditis, respectively ( Hyjek et al, 1988 ; Hyjek & Isaacson, 1988).
The onset of MALT lymphomas in the stomach is preceded by the acquisition of MALT as a result of Helicobacter pylori infection. The micro-organism can be found in the gastric mucosa in nearly all instances, with several lines of evidence suggesting a link between H. pylori chronic gastritis and MALT lymphoma of the stomach. A close association has been reported in epidemiological studies ( Wotherspoon et al, 1991 ; Doglioni et al, 1992 ; Parsonnet et al, 1994 ). In vitro experiments have also demonstrated that the tumour cells of B-cell low-grade gastric MALT lymphoma proliferate in strain-specific response to H. pylori, and that this response is dependent on T-cell-specific activation by H. pylori ( Hussel et al, 1993 , 1996). Moreover, regression of gastric MALT lymphoma after antibiotic eradication of H. pylori has been reported in more than half of the treated patients ( Wotherspoon et al, 1993 ; Roggero et al, 1995 ; Bayerdörffer et al, 1995 ; Montalban et al, 1997 ). This close association of H. pylori with gastric MALT lymphoma has led to the hypothesis that the micro-organism may provide the antigenic stimulus for sustaining the growth of the lymphoma in the stomach ( Isaacson, 1994; Isaacson & Spencer, 1995; Zucca & Roggero, 1996; Carlson et al, 1996 ). Furthermore, chronic gastritis specimens taken several years before the development of a MALT lymphoma have recently been shown to contain the B-cell clone which will become predominant in the transformation to malignant lymphoma, demonstrating the clonal progression from H. pylori gastritis to MALT lymphoma ( Zucca et al, 1996 ). There are, however, rare cases in which H. pylori infection is not demonstrable, and it has been speculated that other infectious agents and environmental factors might have a pathogenetic role ( Fagioli et al, 1994 ; Luppi et al, 1996 ; Xu et al, 1997 ).
Histologic features and diagnostic problems
Histopathology and immunohistochemistry of low-grade MALT lesions
The histological features of low-grade B-cell lymphomas of MALT type are similar regardless of site of origin and simulate those of MALT as exemplified by Peyer's patches. They have been best characterized in gastric lymphomas, since the stomach is by far the most common site.
The pivotal feature of low-grade MALT lymphoma is the presence of a variable number of lymphoepithelial lesions that can be defined as unequivocal invasion and partial destruction of gastric glands or crypts by aggregates of tumour cells (Fig 1). Some of the epithelial cells may take on the appearance of oncocytes or of signet-ring cells ( Chan et al, 1990 ; Papadaki et al, 1992 ; Zamboni et al, 1996 ). Lymphoepithelial lesions are of striking relevance for the diagnosis of low-grade gastric MALT lymphoma; however, they can sometimes be seen in the context of florid chronic gastritis and can also be present in other sites of both native and acquired MALT ( Isaacson & Norton, 1994; Chang, 1996).
The tumour's cells are usually small-to-medium-sized lymphocytes with moderately abundant cytoplasm and irregularly shaped nuclei resembling those of follicle-centre centrocytes and have been designated centrocyte-like (CCL) cells ( Isaacson & Spencer, 1987; Isaacson & Norton, 1994). However, the morphologic pattern of the neoplastic cells covers a quite large spectrum: resemblance to centrocytes is the general rule but some are similar to the so-called monocytoid cells (abundant pale cytoplasm and well-defined cell borders) or to small lymphocytes, sometimes with lymphoplasmacytic features. Any of these cytological aspects can predominate or they can coexist to various degrees in the same case. Scattered transformed blasts can also be found. Some degree of plasma cell differentiation is often present; the plasma cells in certain cases are so conspicuous as to suggest a diagnosis of extramedullary plasmacytoma ( Isaacson & Norton, 1994). Plasma cells often present varying degrees of atypia, Dutcher bodies or other intracellular inclusions ( Isaacson & Norton, 1994).
The lymphoma cells diffusely infiltrate in lamina propria and grow around reactive follicles. The complex interaction between the CCL cells and the reactive B-cell follicles is known as follicular colonization and can show several patterns ( Chan et al, 1990 ; Isaacson et al, 1991 ). The pre-existing germinal centres are selectively overrun and sometimes extensively replaced by the neoplastic cells that often undergo striking plasma cell differentiation after follicle invasion. In other cases the CCL cells within the follicles undergo prominent blast transformation. These aspects of follicular invasion may in some cases closely correspond to those observed in follicular lymphomas. Follicular colonization also represents a common pattern of lymph node involvement by low-grade B-cell MALT lymphoma, and the differentiation from follicular lymphoma is even more difficult in the lymph nodes. In cases in which CCL cells have a prominent monocytoid appearance the histologic pattern of lymph node involvement can be indistinguishable from that of primary nodal marginal-zone (monocytoid) B-cell lymphoma ( Isaacson, 1994).
It should be reiterated that, unlike most low-grade B-cell lymphomas of peripheral lymph nodes, MALT lymphomas are usually localized. Bone marrow involvement, which is very rare, occurs only late in the course of the disease. We recently reviewed a series of 93 patients with low-grade gastric MALT lymphoma ( Pinotti et al, 1997 ). Eighty-two (88%) of them had stage I disease and four (4%) stage II1 (enlarged paragastric lymph nodes); seven patients (7.5%) had stage IV disease because of bone marrow involvement, but presented with symptoms due to the gastric localization. Analogous to gastric carcinoma, the depth of infiltration of the gastric wall has been reported to be strongly associated with spread to the regional lymph nodes, and it has therefore been recommended that the depth of infiltration be included in pathology reports concerning primary gastric lymphoma of the MALT ( Eidt et al, 1994 ). A recent report indicates that endoscopic ultrasound may be useful to evaluate the depth of infiltration and to distinguish benign lymphoid aggregate from lymphoma ( Pavlick et al, 1997 ).
Within the stomach the low-grade MALT lymphoma is often multifocal ( Wotherspoon et al, 1992a ). Small lymphoma foci can be present at sites distant from the main tumour and may sometimes consist of only single follicles surrounded by an enlarged marginal zone of CCL cells. The presence of such microscopic lymphoma foci explains the frequent report of relapses in the gastric stump after surgical excision. In the above-mentioned series ( Pinotti et al, 1997 ) multicentricity was found in one-third of the cases. In three cases MALT lymphoma first presented in the gastric stump after a Billroth II resection for peptic ulcer, and in a fourth it relapsed in the gastric remnant after partial gastrectomy performed as the sole treatment for the low-grade MALT lymphoma. Moreover, there is some evidence that MALT lymphomas sometimes involve multiple mucosal sites. Some cases have been identified with simultaneous gastric and intestinal involvement; thyroid and salivary glands MALT lymphomas may also disseminate to the gastrointestinal tract ( Du et al, 1996b ; Diss et al, 1993b ; Isaacson & Norton, 1994). It has been postulated that this finding may be due to preserved homing properties similar to those of the normal B cell of the MALT ( Isaacson & Spencer, 1995).
Although diagnosis of low-grade B-cell MALT lymphomas in gastric biopsies is usually straightforward, early or borderline cases can be confused with H. pylori related follicular gastritis. In certain cases the distinction on purely morphologic grounds can be very difficult. In the older literature the presence of reactive lymphoid follicles was often regarded as indicative of a benign process, but in fact this is not the case: reactive follicles are almost always present in low-grade MALT lymphoma. Certain parameters have been reported as very useful for the diagnosis of low-grade lymphoma on a gastric biopsy specimen: prominent lymphoepithelial lesions, moderate cytologic atypia of neoplastic lymphocytes, and plasma cells with Dutcher bodies. However, none of these criteria can be found in up to a third of cases and their absence does not exclude the diagnosis of lymphoma ( Zuckerberg et al, 1990 ). A scoring system has therefore been devised to enhance diagnostic confidence ( Wotherspoon et al, 1993 ). Score 0 defines the histologic pattern of the normal gastric mucosa; score 1 describes a chronic active gastritis with small clusters of lymphocytes in lamina propria, without lymphoid follicles and without lymphoepithelial lesion; score 2, a chronic active gastritis with prominent lymphoid follicles and surrounding mantle zone and plasma cells without lymphoepithelial lesion; score 3, a probably reactive lymphoid infiltrate, with lymphoid follicles surrounded by small lymphocytes that infiltrate diffusely in lamina propria and occasionally into epithelium; score 4, a suspicious, probably neoplastic, lymphoid infiltrate composed by lymphoid follicles surrounded by centrocyte-like cells that penetrate diffusely in lamina propria and into epithelium in small groups; score 5 represents the typical low-grade MALT lymphoma with dense diffuse infiltrate of centrocyte-like cells in lamina propria and prominent lymphoepithelial lesion. From the viewpoint of the histological diagnosis the greatest difficulty resides in evaluating those suspicious lymphoid infiltrates scored as grades 3–4. In the past, since the consequence of a lymphoma diagnosis was a gastrectomy, pathologists may have been more cautious than they are now in defining any worrisome lymphoid infiltrate as a lymphoma. The incidence of gastric lymphoma appears to have risen in recent years ( Severson & Davis, 1990) and this might be partially due to the inclusion of cases that in the past would have been misinterpreted as ‘pseudolymphomas’.
Immunohistochemistry studies can therefore be relevant in many aspects. There is an almost complete homology between the CCL cells and the marginal-zone B cell (of spleen, Peyer's patches and lymph node) phenotype, with typical positivity for surface immunoglobulins and pan-B antigens (CD19, CD20 and CD79a) and a lack of CD5, CD10 and CD23 expression; BCL-2 protein expression is also usually demonstrable ( Isaacson, 1994; Isaacson & Norton, 1994; Pileri & Sabattini, 1997).
A search of surface immunoglobulins with effective antigen retrieval techniques may be essential for distinguishing between a suspicious reactive lymphoid infiltrate and a MALT lymphoma in the cases where the morphologic pattern is equivocal. In fact, CCL cells express surface, and to a lesser extent cytoplasm monotypic immunoglobulins (IgM in most cases, IgA in a few and very rarely IgG; IgD are usually negative) which show light-chain restriction. Demonstration of a clear-cut restriction of light chains (kappa:lambda ratio >10:1, or vice versa) strongly supports the diagnosis of B-cell lymphoma. Immunostaining with the pan B-cell CD20 antibody may help distinguish CCL cells from plasma cells and to identify lymphoepithelial lesions. These latter can be highlighted with antibodies to cytokeratin (Fig 2).
The degree and distribution of the T-cell component can be evaluated with a panel of T-cell antibodies (CD3, CD4, CD8, CD45Ro). Abundant CD4 positive T cells are sometimes comprised within the neoplastic B cells in low-grade MALT lymphomas, confirming the necessity of T-cell help to sustain the initial tumour growth (Fig 3); on the contrary, this CD4 infiltrate is not evident in the high-grade lesions ( Koulis et al, 1997 ; Greiner et al, 1997 ; Hussel et al, 1996 ).
Native and acquired MALT usually present in the gut primary and secondary follicles surrounded by T-cell-rich areas containing very few B cells, and therefore any extended diffuse dense B-cell infiltrate with a few interspersed T cells is an element of aberrant immunoarchitecture which suggests the presence of a lymphoproliferative process. Immunostaining with the CD5, PRAD-1 (BCL-1, cyclin D1) and CD23 markers can be useful for differential diagnosis from mantle cell lymphoma (typically CD5+, PRAD-1+ and CD23−) and from small lymphocytic lymphoma/CCL (CD+, CD23+ and usually PRAD-1−); CCL cells of MALT lymphoma are always CD10-negative and usually PRAD-1, CD5 and CD23-negative ( Zukerberg et al, 1993 ; Pileri & Sabattini, 1997). Immunostaining for CD21 (follicular dendritic cells), for Ki-67, and for BCL-2 and BCL-6 proteins may contribute to the identification of residual reactive follicles (whose cytologic features can be sometimes erroneously interpreted as an indication of high-grade transformation) and together with CD10 also to distinguish follicular colonization from the very rare extranodal follicular lymphomas ( Pileri & Sabattini, 1997). Immunostaining with H. pylori associated antigens may be useful in the evaluation of responses to eradication treatment, since following antibiotic treatment the micro-organism can often assume a ‘coccoid’ appearance ( Price, 1996). Monoclonality studies can help determine the diagnosis ( Wotherspoon et al, 1993 ), but, as will be discussed later, this type of analysis is not always conclusive.
Identification of high-grade MALT lymphomas is often problematical; a small component of low-grade MALT lymphoma can be identified in a significant proportion of high-grade GI lymphomas ( Figs 4A and 4B), and conversely, foci or high-grade (large-cell) lymphoma can be seen in low-grade MALT lymphoma, suggesting transition from one to the other, analogous to other low-grade lymphomas ( Boot et al, 1995 ). The prevalence and time interval of histologic transformation of MALT lymphomas are unknown. Scattered blasts (large transformed cells) can usually be encountered in low-grade MALT lymphoma. There is therefore a problem with the definition of high-grade MALT lymphoma which is related to the amount of transformed (high-grade) cells; as yet there has been no agreement. The number of large cells may vary in different areas and the cut-off percentage has not been defined. Indeed, the high-grade form represents only a relatively small percentage (approximately 15–30%) of gastric lymphomas diagnosed as being clearly derived from MALT. It has been proposed that the presence of compact confluent clusters or sheets of large cells may indicate the emergence of new clones and can be used as the criterion for the transformation into a high-grade MALT lymphoma ( Chan, 1996). With transformation into a high-grade MALT lymphoma, the lymphoid follicles can be lost and the number of lymphoepithelial lesions decreases. A retrospective clinicopathological study on a series of 56 patients with primary gastric lymphoma ( Castrillo et al, 1992 ) showed that the presence of limited areas of high grade within a low-grade specimen does not change the survival pattern when patients are treated with surgery, chemotherapy or a combination of the two. However, in a very recent study ( Thieblemont et al, 1997 ) the presence of a large-cell component at diagnosis was significantly associated with a poorer outcome in patients with gastrointestinal MALT lymphoma but not in those with MALT lymphoma presenting in non-gastrointestinal sites. In another recent study ( de Jong et al, 1997 ) the presence of a diffusely distributed increased number of blasts was also found to be predictive of a less favourable long-term outcome in gastric lymphoma patients.
A second problem is posed by the diffuse large-cell lymphomas in which no low-grade MALT component can be identified. These must be considered as primary gastric high-grade lymphomas de novo. Whether this latter group should be categorized as MALT lymphoma remains controversial, it is uncertain how many are of MALT origin and how many represent the usual nodal type. At least some of these lymphomas are almost certainly transformed low-grade MALT lymphomas. It has been demonstrated that high-grade gastric lymphomas may represent — at least in certain cases — clonal evolution from a previously low-grade MALT lymphoma and that aggressive disseminated nodal disease may be a final step in the progression of a low-grade gastric MALT lymphoma ( Chan et al, 1990 ; McCormick et al, 1994 ; Montalban et al, 1995b ). The hypothesis of a clonal outgrowth from a pre-existing low-grade lesion is strongly supported by the epidemiological findings of Parsonnet et al (1994 ), who found that H. pylori infection precedes the growth of low-grade MALT lymphoma and is also strongly associated with the development of high-grade gastric lymphomas that therefore appear to represent a histologic progression from a previous low-grade MALT lymphoma. Moreover, the prognosis for large B-cell lymphomas of the stomach seems better than in nodal lymphomas with equivalent histology ( Zucca & Cavalli, 1996; Coiffier & Salles, 1997) and they might be biologically different as suggested by some (partially controversial) studies of oncogene involvement ( Peng et al, 1996b ; van Krieken et al, 1990 ; Raghoebier et al, 1991 ).
However, the histological and immunohistochemical features and the clinical behaviour of both primary (de novo) and secondary (derived from low-grade MALT lesions) high-grade gastric lymphomas appear to be similar ( Cogliatti, 1991; Isaacson & Norton, 1994). Therefore, it does not seem necessary, at least for the present, to classify them as separate groups. This distinction is almost impossible unless a low-grade component is found, but it should, in our opinion, be verified whenever possible, in reporting clinical trials, since this might contribute to a better understanding of the clinical behaviour of high-grade MALT lymphomas.
Molecular detection of clonality as a diagnostic tool
As discussed earlier, a crucial issue for the study of MALT lymphomas is the pathological evaluation of a gastric biopsy in which it is sometimes difficult to differentiate between phlogistic infiltrates and low-grade lymphomas. This problem is evident in the histological score proposed for the evaluation of gastric MALT lymphoma biopsies ( Wotherspoon et al, 1993 ), which shows that there is a continuous spectrum of lesions during transition from H. pylori associated gastritis to low-grade MALT lymphoma. Since lymphoma represents a clonal outgrowth of cells that have acquired certain genetic alterations, possible diagnostic support can come from the finding of a monoclonal B-cell population ( Diss et al, 1993a ; Pan et al, 1994 ; Wotherspoon et al, 1993 ).
Immunohistochemistry staining and genotype investigations of MALT lymphomas using the Southern blot technique or polymerase chain reaction (PCR) analysis have revealed clonal rearrangements of the immunoglobulin genes. In fact B cells rearrange their immunoglobulin genes to become functional and this rearrangement is unique in each cell ( Stewart & Schwartz, 1994). This rearrangement can be detected with Southern blotting but the methodology requires a large amount of high-quality DNA from fresh or frozen tissue and is not very sensitive. To circumvent this problem a PCR analysis that can be done directly on the paraffin-embedded material from the pathological slide has been developed ( Pan et al, 1994 ).
Many investigators now require the detection of a monoclonal B-cell population for the diagnosis of low-grade MALT lymphoma in doubtful cases or for confirming histological remission. However, there are instances in which clonal B-cell populations that do not behave like lymphomas are detected. In fact, in the normal immune response monoclonal proliferation of B or T cell, mimicking a malignant process, can be sometimes observed under antigenic influence; this may be a limiting factor when a very sensitive method such as PCR is used for clonality detection. Therefore PCR monoclonality alone does not always equal malignancy and does not necessarily have implications for treatment approaches; some genetic lesions have a clonal abnormality that remains sufficiently responsive to normal regulators of growth and differentiation ( Fend et al, 1994 ; Ludwig et al, 1993 ).
The border between chronic inflammation and certain lymphoid neoplasms is unclear but is well illustrated by a number of mucosa-associated processes. Patients with MESA and Sjögren's syndrome have an increased risk of lymphoma over that of the normal population ( Kassan et al, 1978 ) and immunoglobulin heavy or light chain monoclonal rearrangements appear to be a relatively common finding in MESA ( Hsi et al, 1996b ; Jordan et al, 1995a ). Analogous findings were reported in Hashimoto's thyroiditis ( Matsubayashi et al, 1990 ) and they might be useful for identifying a subset of patients with a higher risk for development of MALT lymphoma, but cannot always be taken as evidence of an already existing neoplastic lesion. Similar to that reported in Sjögren's syndrome and in Hashimoto's thyroiditis, it has recently been shown that PCR monoclonality was present in approximately 15% of patients with H. pylori associated chronic gastritis ( Calvert et al, 1996 ; Hsi et al, 1996a ), and none developed a lymphoma during a median follow-up period of 59 months (range 1–66) ( Hsi et al, 1996a ). Of interest is the very recent report that it seems to take longer to cure gastritis containing a monoclonal population than to cure polyclonal gastritis ( Rudolfph et al, 1997 ). Moreover, it has recently been shown that PCR-detectable monoclonality in some cases can persist for several months after antibiotic treatment and the disappearance of histologically detectable MALT lymphoma ( Savio et al, 1996 ).
Furthermore, a number of false-negative results (15–30%) may be due to the less than complete efficiency of the PCR, to the effect of fixation on the biopsy sample, or to sampling errors. Therefore, although the demonstration of monoclonality by PCR may support the diagnosis of B-cell lymphoma, its absence does not necessarily exclude the presence of a clonal neoplastic population. False-positive results are less common (<5%) and could be due to the presence of a benign monoclonal population, to a very low number of B cells in the examined sample, or to contamination. In addition, DNA sequencing analysis can sometimes demonstrate that a monoclonal-appearing band is in fact the product of a polyclonal B-cell population.
The significance of PCR-detected clonality in the absence of histological evidence of lymphoma is still uncertain, so interpretation of the molecular data must always be done in the context of the histological findings. Skilled histological examination is fundamental since the presenting signs and symptoms of low-grade gastric MALT lymphomas are generally non-specific and a benign endoscopic pattern is very often present ( Pinotti et al, 1997 ; Taal et al, 1996 ).
Molecular genetics of MALT lymphomas
Sequence analysis of the immunoglobulin VH genes expressed by tumour B cells from patients with gastric MALT lymphoma shows somatic hypermutation with a high replacement-to-silence ratio and a distribution pattern which suggests that the tumour cell was positively selected through its antigen receptor in germinal centres. In addition, ongoing mutations (i.e. intraclonal variation) of the immunoglobulin genes have been seen in some cases of low-grade lymphoma, suggesting that tumour cells' clonal expansion continues to be dependent on antigen drive ( Qin et al, 1995 ; Chapman et al, 1996 ; Du et al, 1996a ).
Further evidence for antigen-driven selection was recently shown by a study of the third complementarity determining region (CDR3) of the immunoglobulin gene in MALT lymphoma reporting that the pattern of rearrangement of the D segments of the immunoglobulin gene is often unusual, with frequent use of inverted D-J segments and/or D-D recombination ( Bertoni et al, 1997 ). In fact D-D fusions, as well as inverted D-J rearrangements, have bene postulated to be important mechanisms for the generation of antibody diversity and antigen-binding affinity ( Meek et al, 1989 ; Sanz et al, 1994 ). Remarkably, in two patients reported in the cited study, despite different nucleotide sequences at the DNA level in the CDR3, the resultant amino acid sequences matched almost completely, suggesting the presence of common selecting antigens ( Bertoni et al, 1997 ).
Karyotype studies of gastric MALT lymphoma present technical difficulties and only a few cases have been published. The most frequent abnormality appears to be the trisomy 3 which has been reported to be detectable in approximately 60% of both low-grade and high-grade MALT lymphomas ( Wotherspoon et al, 1995 ). Less frequently, trisomies of chromosome 18, 12 and 7 have also been reported. The chromosomal translocations t(1;14) in a few cases and, much more frequently, t(11;18) have been observed and also proposed as possibly significant in the genesis of MALT lymphomas ( Wotherspoon et al, 1992b ; Horsman et al, 1992 ; Leroux et al, 1993 ).
The distinctive nature of low-grade MALT lymphoma of the stomach also seems to be confirmed by the lack of bcl-1, bcl-2 gene rearrangements which characterize mantle cell and follicular lymphomas, respectively ( Wotherspoon et al, 1990 ).
Rearrangement of c-myc is usually not detected in MALT lymphoma, whereas point mutations in this gene have been found in approximately 15% of cases and could be related to the development of early MALT lymphoma lesions ( Peng et al, 1996b ). In nodal lymphomas c-myc is usually rearranged in Burkitt's lymphoma and has been associated with the transition from low- to high-grade lymphoma ( Cotter & Zucca, 1991); it might therefore be postulated that the pattern of c-myc regulation can be different in MALT lymphomas and nodal lymphomas ( Peng et al, 1996b ).
Replication error (RER+) phenotype is a recently defined manifestation of genetic instability caused by a deficiency in the DNA mismatch repair genes, which is evidenced by increased mutation rates in non-coding (microsatallites) and coding regions of the genome. In contrast to other types of non-Hodgkin's lymphoma where microsatellite instability seems absent ( Gamberi et al, 1997 ), the RER+ phenotype appears to be a common genetic feature of MALT lymphomas, detected in approximately 50% of cases ( Peng et al, 1996a ). In MALT lymphoma genetic instability occurs throughout the spectrum of the lymphoma development, since it can sometimes be found also in reactive lymphoid infiltrates adjacent to the tumour, and may be related to the accumulation of genetic aberrations such as p53 mutations. Studies of the p53 gene showed a loss of heterozygosity (LOH) in approximately 7% of low-grade and in 29% of high-grade (large cell) MALT lymphomas. Mutations in the p53 were shown in 19% of low-grade and 33% of high-grade MALT lymphoma. Moreover, only one of 11 low-grade cases showed concomitant p53 allelic loss and mutation, whereas both phenomena were present in 66% of the high-grade tumours. Therefore it appears that p53 partial inactivation may play an important role in the development of low-grade MALT lymphoma, whereas complete inactivation may be associated with high-grade transformation in at least some cases ( Du et al, 1995 ).
This genetic instability could be the reason for the increased incidence of additional solid tumours that has been reported in patients with gastric MALT lymphoma in southern Switzerland and northern Italy ( Zucca et al, 1995 ) and in other gastric lymphoma series ( Castrillo et al, 1992 ; Luppi et al, 1996 ).
Allelic imbalance at microsatellites for some tumour suppressor genes has also been reported during transition from H. pylori induced gastritis to MALT lymphomas: loss of heterozygosity of the DCC (for deleted in colon cancer) tumour suppressor gene, located on chromosome 18q, has been described in two cases of transition from chronic gastritis to low-grade MALT lymphoma and in one case of transition between low-grade to high-grade MALT lymphoma; allelic loss in the regions of the APC (for adenomatous poliposis coli) tumour suppressor gene, located on chromosome 5q was also found in two patients with high-grade MALT lymphoma ( Calvert et al, 1995 ). However, it is unclear as to whether these changes have a causal role in the lymphomagenesis.
The precise significance of most of the described abnormalities is not known, but it is possible that some genomic alterations can provide a selective growth advantage to malignant cells. A tentative explanation for the pathogenesis of gastric MALT lymphomas may be that B and T lymphocytes are recruited in the gastric mucosa as part of the immune response to H. pylori. B-cell proliferation is secondary to specific activation of reactive T cells by H. pylori and cytokines. It is not clear whether the B-cell activation requires the continuous presence of H. pylori as antigenic source or is related to an indirect autoimmune mechanism. In fact, the tumour B cells often show antibody specificity for autoantigens and, to proliferate, need contact-dependent help from the intratumour T cells, which is apparently mediated by CD40 and CD40 ligand (CD154) interaction ( Hussel et al, 1996 ; Greiner et al, 1997 ). This immunologic drive mediated by mucosal T cells may explain the tendency of low-grade MALT lymphoma to remain localized and to regress after H. pylori eradication. However, genetic alterations can continue to occur until a point is reached at which autonomous (i.e. H. pylori independent) growth is possible, and additional alterations can result in high-grade transformation. However, the exact mechanism of the transition from H. pylori infection to low-grade MALT lymphoma is still unclear. Most patients with H. pylori gastritis do not develop lymphoma. Therefore it is widely accepted that additional host-related factors must play a pivotal role in gastric lymphomagenesis. Indeed, the recent observation in some series of an increased incidence of B-cell marginal zone lymphomas in HCV-positive populations ( Luppi et al, 1996 ) suggests a possible pathogenetic role for the hepatitis C virus, but this is still a highly controversial issue ( Luppi et al, 1997 ; Silvestri et al, 1997 ).
Clinical features and therapeutic problems
Low-grade MALT lymphoma is usually a very indolent disease, often remaining localized for a prolonged period, with some cases showing no progression for several years with no treatment; systemic dissemination occurs only in a limited percentage of cases. Hence, a loco-regional therapy may conceivably be curative.
In a large multicentre series of 93 patients from southern Switzerland (Bellinzona) and northern Italy (Brescia and Varese) ( Pinotti et al, 1997 ) with low-grade gastric MALT lymphomas of the stomach, the disease was most often localized in the antrum (41%) or multifocal (33%); the lymphoma was confined to the stomach (stage I) in 88% of cases.
The patients were treated by different modalities: chemotherapy alone, surgery alone or with additional chemotherapy or radiation therapy; 49 patients, all with clinical stage I, were given antibiotics against H. pylori as their sole initial treatment, and seven patients refused any treatment. The actuarial 5-year overall survival was 82% (95% CI 67–91%) in the series as a whole. At a median follow-up of 3 years, 10/93 patients have died, all but one of a second solid tumour. There was no significant difference in overall survival and event-free survival between patients who received different treatments.
In the antibiotic-treated group eradication of H. pylori was achieved in 97% of patients (95% CI 88.2–99.9%). All but one of the patients were symptomatic at presentation. The main symptoms were those of non-specific dyspepsia, epigastric pain being the commonest manifestation. In keeping with other reports ( Taal et al, 1996 ), the endoscopic findings were usually those of non-specific gastritis and or peptic ulcer, the presence of a mass being very unusual.
Histologic regression of the MALT lymphoma was documented in 67% of patients (95% CI 51–80%) after H. pylori eradication. The median time required to achieve histologic regression was 5 months; however, in approximately a third of the cases the lymphoma regression after antibiotic therapy required a prolonged time (up to 18 months).
Long-term follow-up of the antibiotic-treated patients will be mandatory, since gastric low-grade MALT lymphoma is usually an indolent disease, and whether treatment for H. pylori will definitely cure the lymphoma and prevent its relapse is still unknown. Some cases have been reported in which tumour recurrence was documented after H. pylori re-infection, suggesting that some residual dormant tumour cells can still present despite clinical and histologic remission ( Cammarota et al, 1995 ; Horstmann et al, 1994 ).
Nevertheless, the indolent nature of the disease in most cases make a conservative approach advisable, with antibiotic therapy as the sole initial treatment. The use of antibiotics as first-line therapy may avert or at least postpone the indication for surgical resection in the majority of patients, and we recommend eradication of H. pylori before consideration of further therapeutic options. Efficacy of antibiotics may be reduced in cases with locally advanced disease, with bulky masses or deep infiltration of the gastric wall, but in our experience eradication of H. pylori is also worthwhile in these cases. Chemotherapy with single-agent chlorambucil ( Hammel et al, 1995 ) and local radiotherapy ( Schechter et al, 1997 ) have been shown capable of inducing complete remission in most cases and can therefore be used in patients who do not respond to antibiotics. The need for surgical resection should be re-defined. Surgery has been widely used in the past with excellent results for localized disease. However, follow-up endoscopy may reveal, in the remaining gastric mucosa, the re-appearance of lymphoepithelial lesions that can be responsible for local recurrence ( Wotherspoon et al, 1992a ). Indeed, the fact that MALT lymphoma is often a multifocal disease suggests that clear excision margins are not necessarily a guarantee of radical resection. Therefore, if surgery is chosen, a total gastrectomy may offer greater chances of cure. However, this operation carries a certain risk of mortality and may severely impair the patient's quality of life.
It must be reiterated, however, that lymphoma regression may require a very prolonged time, and if patients undergo regular careful endoscopic follow-up that shows stable disease a second-line treatment can be postponed for several months with no particular danger. Post-treatment assessment of gastric biopsies is sometimes problematic; enough biopsies should be taken to map the entire stomach. Histologic evaluation using the terminology applied in the scoring system proposed by Wotherspoon et al (1993 ) enables the allotment of cases in which definite conclusions cannot be reached on histologic grounds. Molecular follow-up (with monoclonality analysis of gastric biopsy) can be useful, but, as discussed previously, the histological assessment of gastric biopsies by an experienced pathologist remains essential ( Savio et al, 1996 ).
Low-grade MALT lymphomas with focal high-grade areas behave like other low-grade MALT lymphomas when treated with surgery or chemotherapy or both ( Castrillo et al, 1992 ) but their susceptibility to being cured with antibiotics has not yet been adequately studied. With the growing popularity of stomach-conserving approaches, evaluation of the prognostic significance of the variable high-grade component in low-grade gastric MALT lymphomas ( de Jong et al, 1997 ) and of the histologic and echoendoscopic degree of infiltration ( Eidt et al, 1994 ; Pavlick et al, 1997 ) is becoming a crucial issue and should be evaluated in prospective studies.
A further challenge to the ongoing research is the definition of the molecular events responsible for the escape from H. pylori mediated antigen-dependent growth of low-grade MALT lymphoma and of those that can result in the transformation into a high-grade lymphoma. This will help in the choice of the appropriate therapeutic strategy for the individual patient.
Treatment of localized high-grade gastric lymphomas remains a very controversial issue, but mounting evidence suggests that surgery may not be necessary and that most patients can be cured with a stomach-conserving treatment using combination chemotherapy with or without radiotherapy ( Zucca & Cavalli, 1996; Coiffier & Salles, 1997).
The growth of high-grade MALT lymphoma is likely to be no longer dependent on antigenic drive; however, anecdotal cases of regression of high-grade lesions after anti-H. pylori therapy have been reported ( Kolve et al, 1997 ; Seymour et al, 1997 ). These data need to be confirmed and, in our opinion, gastric lymphoma with aggressive histological features must be aggressively treated ( Zucca & Cavalli, 1996). However, also in the setting of high-grade MALT lymphoma, eradication of H. pylori should be attempted (in addition to chemotherapy) since it can eliminate a residual/relapsed low-grade component that upon antigen stimulation can be responsible for tumour recurrence ( Bayerdörffer et al, 1995 ; Boot et al, 1995 ).