Follicular lymphoma (FL) is defined as a neoplasm of germinal center B cells, usually with a follicular growth pattern. The neoplastic population typically consists predominantly of centrocytes, the resting cells present in the light zone of normal germinal centers, and relatively few centroblasts, the more immature and rapidly proliferating cells of the germinal center dark zone. FL cells usually express Bcl6, CD10 and other proteins that are also expressed in germinal center B cells. A major difference between FL cells and germinal center B cells is the former’s expression of the anti-apoptotic bcl2 protein, which is caused by the characteristic translocation t(14;18) involving the BCL2 gene or its variants t(2;18) and t(18;22) and enables the cells to survive in the absence of antigenic stimulation. This results in accumulation of non-proliferating centrocytes with a minority of proliferating centroblasts and a low proliferation index with Ki-67 staining. According to the pathologist, this configuration is “translated” into a low histological grade of the lymphoma.
Although FL is typically widespread at the time of the diagnosis and indolent in its clinical evolution, it has long been recognized that clinical aggressiveness and risk of transformation to diffuse large B-cell lymphoma (DLBCL) increases proportionally to the numbers of large cells (centroblasts) and the proliferation fraction.1–3 Pathological classification schemes developed largely in the USA (where FL is very common) suggested dividing FL into three subtypes (or grades), based on the number of large cells.4 In contrast, the European Kiel classification5 considered that cases containing both centrocytes (usually predominant) and centroblasts comprised one disease (centroblastic/centrocytic lymphoma, equivalent to the World Health Organization FL1-2, and possibly 3A), while those in which centroblasts predominated were considered a follicular variant of centroblastic lymphoma (a morphological variant of DLBCL). The WHO classifications of 2001 and 20086 recognized three grades - FL1-2, FL3A and FL3B (Table 1), based on the number of centroblasts present per high power field; the difference between FL3A and FL3B is the presence of a mixture of centrocytes and centroblasts in FL3A and the presence of follicles existing exclusively of centroblasts, immunoblasts, or both in FL3B. The continuation of a grading system is supported by the fact that several studies suggested that grades predict outcome, higher grades being associated with poor clinical outcome and more rapid progression to DLBCL. However, the reproducibility of grading is low, and it has not been easily replaced by immunohistochemistry for Ki67/MIB1 (proliferation index).
Several studies have tried to address the issue of whether there are biological differences between typical low-grade FL1-2 and FL with increased centroblasts (grade 3), and between FL3A and FL3B.7–11 Several recent studies suggest that higher grade FL, in particular FL3B, are different with regards to genetic, immunophenotypic and clinical features from the prototypic FL1-2. While FL1-2 lymphomas are relatively homogeneous (expression of CD10, BCL2 and BCL6, and carrying a t(14;18) in 90% of the cases), FL3A and, in particular, FL3B are more heterogeneous with respect to immunophenotype, genetic abnormalities, and gene expression. Individual studies have taken different approaches to this issue but encountered the same problem: some cases of FL3B have many features in common with FL1-2, whereas other cases have more in common with DLBCL (Table 2).
Any study on the biology of FL3B is complicated by relatively simple factors related to the definition itself and purely histopathological aspects. First, some cases of FL contain large centrocytes (large cleaved cells) or small centroblasts that may be interpreted differently by individual pathologists, so that the border between FL3A and FL3B may be blurred. Second, some cases of FL show considerable heterogeneity from area to area. Third, poor tissue handling, fixation or technical processing problems may introduce artifacts that interfere with grading. Fourth, in many cases of FL3B the biopsy contains diffuse areas with sheets of centroblasts, consistent with DLBCL. The presence of such diffuse areas is associated with a worse outcome,2,12 and the WHO classification requires that a separate diagnosis of DLBCL be made in such cases (Table 2). However, this interpretation is not unambiguous, and in addition, areas of DLBCL may be easily missed in small biopsies. Another problem is that in some cases of FL there are discrepancies between the cellular composition and the Ki67 proliferation index.13,14 Some cases of FL1/2 may show a majority of cells being in cycle, whereas some FL3B cases have a relatively low proliferation index. Finally, cases of FL3B with a history of FL1-2 were sometimes included in published series.
The problems regarding the histological definition of FL3B are illustrated in the paper by Horn et al. published in this issue of Haematologica.15 These authors collected a large series of 23 cases of FL3B and compared the immunophenotypic and genetic features of these cases with other cases of FL, including FL3A, FL3B with DLBCL, cases in which a distinction between FL3A and FL3B was not possible (FL3U), cases of FL consisting of large centrocytes or having a high proliferation fraction, and cases of typical FL1-2. Interestingly, even after introducing these six different categories, it appeared that while FL1-2 was very homogeneous, all other categories, including FL3A and FL3B, remained heterogeneous with regards to most features analyzed. Half of the FL3B cases expressed the germinal center marker CD10, as FL1-2 and FL3A do, while half lacked CD10 and expressed MUM1/IRF4, a post-germinal center marker not seen in FL1-3A. Only 9% of FL3B contained a BCL2 break, in contrast to 88% of FL1-2; 17% of FL3B cases had a BCL6 break and 22% a MYC break, which were not seen in FL1-2. However, FL3A was also not homogenous; only 58% had a BCL2 break and 22% had a BCL6 break, the latter being similar to the frequency in FL3B. FL3U, in which stratification as 3A or 3B was not possible by morphological criteria, was not surprisingly heterogeneous, with some cases resembling FL1-2 and others FL3B. Both FL3A and FL3B differed importantly from DLBCL with a component of FL3B, which rarely expressed CD10, often expressed MUM1/IRF4, and had a BCL6 break in 50% of cases, similar to DLBCL without a partially follicular pattern. Interestingly, in all categories, including those of FL1-2 with a high proliferation fraction, when MYC breaks occurred, they were often seen together with BCL2 or less often, BCL6 breaks, suggesting that these are events associated with progression or transformation to a higher-grade process.
What can we conclude for clinical practice from this study? Unfortunately clinical follow-up data are not available on the patients studied. The biological heterogeneity of all categories of FL3 suggests that a simple system of subclassification of this category may not be possible. No clinical differences have been found to date between cases of FL3A and FL3B without a component of DLBCL treated with current anthracycline-containing regimens, so that the clinical relevance of this distinction is not evident.15,16 In the absence of clinical follow-up data from the series reported by Horn et al., it is not possible to determine whether separating FL1-2 with a high proliferation index or large centrocytes from typical FL1-2 is clinically relevant. It is also not possible to determine from this study whether FL3A (or for that matter FL3B) with a FL1-2 biological profile (CD10+, t(14;18)+) should be folded in with FL1-2 or kept as FL3, because the presence of numerous centroblasts may still predict a more aggressive course than that associated with FL1-2.
An important question is whether the histological grading system could be replaced or supplemented by a genetically oriented classification system, for instance by separating t(14;18)-positive FL from translocation-negative ones. Such an approach is appealing and has been followed in several studies.11,17–24 However, while FL that lack a t(14;18) have clinical and pathological differences from those with the translocation, the category of t(14;18)-negative FL is not homogeneous. Nonetheless, it might be useful to identify such cases for further study or individualized treatment. Importantly, even within the category of t(14;18)-positive FL, a grading system might still be desirable to identify cases with increased centroblasts (FL3A or B) or a high proliferation index, which may still have a more aggressive clinical behavior than grade 1–2 cases. Thus, stratification of FL based solely on genetic features does not seem practical at present.
In summary, the study by Horn et al. has demonstrated the relative (and not unexpected) immunophenotypic and genetic homogeneity of typical FL1-2, including those cases with a high proliferation fraction or large centrocytes. It is also clear that FL3B with a component of DLBCL is genetically and immunophenotypically similar to DLBCL, and distinct from any of the FL categories. FL3, regardless of the exact proportion of centroblasts, is heterogeneous. While the majority of cases of FL3A have a germinal center B-cell immunophenotype, similar to FL1-2, 40–50% lack BCL2 rearrangement or expression. While only rare cases of FL3B have a BCL2 break, expression of CD10 or Bcl2 was seen in about 40%. Assessment of BCL2, BCL6, and MYC breaks as well as expression of CD10 and Mum1/IRF4 could be used to identify FL3 cases that are biologically closer to FL1-2 and those that are distinct; however, clinical data to support the need for this distinction in practice are lacking.
Nancy Lee Harris is the Austin L. Vickery professor of pathology at Massachussetts General Hospital and the Editor of the Case Records of the Massachussetts General Hospital at the New England Journal of Medicine. Philip Kluin is professor in hematopathology at the University Medical Center Groningen, The Netherlands.
Financial and other disclosures provided by the author using the ICMJE (www.icmje.org) Uniform Format for Disclosure of Competing Interests are available with the full text of this paper at www.haematologica.org.
- Copyright© Ferrata Storti Foundation