Burkitt lymphoma in adults

Authors


Professor D C Linch, Department of Haematology, Cancer Institute, University College London, 72, Huntley Street, London WC1E6BT, UK. E-mail: d.linch@ucl.ac.uk

Summary

The diagnosis of Burkitt Lymphoma (BL) and B-cell lymphomas unclassifiable with features intermediate between Diffuse Large B-cell Lymphoma and BL (BLU) in adults remains problematic even with immunophenotyping and MYC gene analysis. Gene expression profiling may improve categorization but is not routinely available. BL and its variants should be treated with specific regimens incorporating intensive courses of chemotherapy with fractionated alkylating agents and cell cycle phase-specific agents that readily cross the blood brain barrier. Subsequent courses should be given as soon as haematological recovery occurs, with the whole course completed within a few months. A number of regimens have been developed that encompass these principles but there have been no comparative randomized trials. The results from several studies suggest that the addition of rituximab is highly efficacious and this may be particularly valuable in older patients. It is usual to employ ‘risk-adapted’ strategies in the treatment of BL but these must be continually re-evaluated, and ‘response-adapted’ approaches should be explored. The role of transplantation is limited and largely confined to autologous transplants in patients who only achieve a partial response on front-line therapy or who have a chemosensitive relapse. Further advances will be greatly facilitated by randomized trials, which will require international collaboration.

Burkitt Lymphoma (BL) can be endemic, sporadic or associated with immunodeficiency and this review focuses on the latter two forms of the disease in adults. A number of broad conclusions can be drawn about the incidence, clinical and pathological features of BL as well as the results of various therapeutic regimens, but it is more difficult to provide evidence-based recommendations than in the childhood form of the disease and many other types of adult lymphoma. Several reasons account for this uncertainty, including the relative rarity of this type of lymphoma in adults, the small size of many of the published series and the variable pathological inclusion criteria used in these studies. Most importantly, there are no randomized trials in adults.

Incidence in adults

BL is estimated to account for only 1–5% of all Non Hodgkin Lymphoma (NHL) in adults, but it is not less common than in children; in fact the converse is true. As NHL overall is so much more common in adults than children, the absolute number of BL cases in adults exceeds that in children. Data from the Netherlands Cancer Registry (Boerma et al, 2004) indicates that the distribution of BL is bimodal with an early peak between 0 and 15 years and a later peak over 60 years of age. For children under the age of 16 years the incidence was just over four cases per million per year. For adults the incidence was approximately 2·5 cases per million per annum, and overall, because of the greater age-span of the adult population, this means that over half of all cases are adults. Analysis of the US Surveillance, Epidemiology and End Results database is in accord with this, with the median age of BL being 45 years and with about 30% of patients being over 60 years of age (Kelly et al, 2009). The limited size of the adult data-sets is not therefore due to fewer cases of BL in adults, but is due to the fact that there are far more centres treating adult lymphomas with consequently fewer cases of BL seen at each individual centre. Added to this is the weaker tradition of trial entry in adult practice, which is exacerbated by the fact that many adult patients are elderly, with this age-group being under-represented in most cancer trials.

Definition of Burkitt lymphoma

The robust definition of BL has undoubtedly been challenging. Initially, the diagnosis of BL was based on a clinical description of endemic forms of the disease associated with morphological features classified as ‘undifferentiated lymphoma’ in the Rappaport classification. The cells of BL are medium-sized with round nuclei, without cleaves or folds, with multiple basophilic, medium-sized nucleoli and deeply basophilic cytoplasm often containing vacuoles, which are best seen on tissue-imprints or blood or marrow smears. The growth pattern is diffuse and monotonous with a ‘starry-sky’ appearance due to numerous macrophages that have ingested the many apoptotic lymphoma cells. There are multiple mitotic forms reflecting the high proliferation rate. There are, however, cases with more pleomorphic nuclei and larger, more prominent nucleoli and these were referred to as ‘atypical BL’. In the Working Formulation BL was included in the category of ‘small non-cleaved cell lymphomas’ (SNCL) and this was subdivided into BL and non-BL types. Most reported series using the Working Formulation included all forms of SNCL. The revised European-American classification of lymphoid neoplasms (REAL) defined an entity of Burkitt-like lymphoma (BLL), recognizing the difficulty of distinguishing some BL with variant morphology from Diffuse Large B-cell Lymphoma (DLBCL) (Harris et al, 1994). In the World Health Organization (WHO) 2001 classification this entity was referred to as atypical BL (Diebold et al, 2001). The WHO 2008 classification does not include atypical BL or BLL but there is a new subcategory of ‘B-cell lymphomas unclassifiable with features intermediate between DLBCL and BL’ (BLU) (Kluin et al, 2008), and this includes many of the cases previously classified as atypical BL or BLL. Those cases with slightly deviant morphology, but in whom the immuno-phenotypic and genetic criteria for BL are fulfilled are now classified as BL, as are cases with typical BL morphology but no demonstrable MYC rearrangement (see below).

Immunophenotyping has clarified the fact that BL cells are mature B-cells without Tdt or CD34 expression, with moderate to strong levels of IgM, and expression of the B cell-specific antigens CD19 and CD20. They have a germinal centre phenotype expressing CD10 and BCL6 but BCL2 only weakly, if at all. This is not fully diagnostic, however, and there are cases of BL with atypical phenotypes, such as moderately strong BCL2 positivity. Nearly all the cells in BL are Ki67 positive, but again this is not diagnostic of BL, as the fraction of proliferating cells in some cases of DLBCL can also approach 100%. The inexactitude of morphological and immuno-phenotypic analysis was well illustrated in study of 208 childhood and adolescent lymphomas analyzed by eight pathologists (Lones et al, 2000). For the BL cases the percentage of agreement of individual pathologists with the final consensus diagnosis was only 42%. In the UK LY06 trial further evidence of the difficulty in diagnosis was illustrated by the fact that central review disagreed with the local diagnosis of BL in 20% of cases (Mead et al 2002).

The clearest hallmark of BL is the over-expression of MYC due to a balanced translocation between the MYC/8q24 locus and an immunoglobulin (Ig) gene, most commonly IGH [t(8;14)(q24;q32)]. However, such translocations are not apparent by fluorescence in situ hybridization (FISH) in up to about 10% of otherwise typical cases, perhaps due to the apparent failure of the ‘break-apart probes’ used to detect all MYC translocations, and occasionally perhaps because high level MYC expression can be achieved by other mechanisms including MYC amplification and down-regulation of MIR34B (Leucci et al, 2008). Furthermore the presence of MYC translocations is not diagnostic of BL. Not surprisingly, they occur in 30–50% of the BLU as defined in the 2008 WHO classification (Kluin et al, 2008), but also they occur in a significant proportion of DLBCL (5–15%). This means that in absolute terms, MYC translocations are more commonly found in DLBCL than in BL. The BLU tend to have karyotypic abnormalities additional to the MYC-IG rearrangements such as the so-called ‘double hits’ where there are also t(14;18) translocations or breakpoints in BCL6 (Aukema et al, 2011). These ‘double hits’ are mainly found in de novo lymphomas but a proportion is found in transformation of more indolent lymphomas (Lin & Medeiros, 2007). It is interesting that most of the ‘double hit’ cases described by Lin and Medeiros (2007) had a BL (17%) or BLL (52%) morphology, whereas in the series described by Le Gouill et al (2007) the morphology was that of a DLBCL, further illustrating the diagnostic difficulties that still exist. Only a few of the most recent studies have used such karyotypic information in the selection of the cases in their series. In the UKLY10 trial, BL was rigorously defined as a lymphoma with a proliferation fraction approaching 100%, with a germinal centre phenotype, the absence of BCL2 and abnormal TP53 expression, and a MYC rearrangement as the sole cytogenetic abnormality as determined by FISH (Mead et al 2008). A total of 128 patients were entered into the study on the basis of a near 100% proliferation fraction but only 58 were considered to have BL after the more extensive analysis. The remainder were classified as DLBCL.

More recently it has been shown that greater diagnostic precision can be achieved with gene expression profiling (GEP). Two major published studies both showed that there was a specific BL signature, which the authors suggested should be considered as the diagnostic ‘gold standard’ for this disease (Dave et al, 2006, Hummel et al 2006). There is high expression of a subset of germinal centre B cell genes and low expression of major histocompatibility complex class I genes and genes in the nuclear factor (NF) kB pathway. Dave et al (2006) identified a number of atypical cases that had a BL signature, but conversely there were some cases morphologically deemed to be BL, atypical BL or BLL that did not harbour the BL molecular signature. Hummel et al (2006) also identified cases with an intermediate signature, confirming that there is a spectrum of disease between BL and DLBCL. GEP is not yet a routine diagnostic tool, however, and no clinical studies reported to date have prospectively used GEP as an entry criterion.

It is clear that the diagnostic criteria for BL have changed considerably over time, and cases in one series from one time period will not be identical to those from a different time period, with probably more heterogeneous patients in the older studies.

Comparison of childhood and adult disease

The presentation of childhood and adult disease is similar in so far as there is often rapidly progressive extra-nodal disease with frequent bone marrow, intestinal tract and leptomeningeal disease, although extra-nodal disease is reported to be more frequent in children. Nodal involvement, by contrast, is more common in adults, as is bulky disease, defined as >10 cm in size. (Boerma et al, 2004). The GEP signature is similar in children and adults, indicating that it is the same disease in both age groups. BLL or BLU are more frequently reported in adults, but paradoxically, in children as many as 30% of cases designated as DLBCL have a BL signature (Klapper et al, 2008). Such discrepancies appear to be considerably less frequent in adults.

Staging

Staging must be completed rapidly once a diagnosis of BL, or one of its variants has been made, as treatment should commence within 48 h of diagnosis. Positron emission tomography (PET) scanning has been shown to have high sensitivity for BL with no false negative lesions detected in two small series using computerized tomography (CT) scanning as the arbiter. (Just et al, 2008; Karantanis et al, 2010). In the first study (Just et al, 2008) additional lesions were identified in four of the five patients studied at presentation, so PET scanning may be of value in low risk patients in whom upstaging would change their risk status and the treatment they would receive. Treatment should not, however, be delayed unduly to obtain such scans. The staging systems used vary, with the Murphy/St Jude system most often used in children and the Ann Arbor system in adults (Table I). The Murphy staging system takes particular account of the frequent extra-nodal disease that occurs in BL and the relatively poor prognosis of intra-thoracic and intra-abdominal disease, at least at the time this system was devised. It should be appreciated that this system was developed in an era when surgery for either diagnosis or treatment was more frequent and the entity of Stage IIR (fully resected abdominal disease) disease is now rarely encountered. The systems differ in the way that a single extra-nodal site is categorized and that the Murphy system includes upgrading of non-resected intra-thoracic and intra-abdominal disease. The Murphy system also distinguishes central nervous disease (CNS) and bone marrow disease from other types of Stage IV disease as defined in the Ann Arbor system. The Ann Arbor system takes account of significant B symptoms. In practice, the variable use of the different staging systems in adults only makes a difference to the prognostic stratification and treatment offered to a minority of patients, but it does make the comparison of different series potentially more problematic.

Table I.  Staging systems used in Burkitt lymphoma.
Staging systems used in Burkitt lymphoma
 Murphy systemAnn arbor system
  1. Italics identify differences between the two systems.

Stage ISingle nodal or extranodal siteSingle nodal or extranodal site
  excluding mediastinum or abdomen 
Stage IITwo or more nodal areas on one side of diaphragmTwo or more nodal areas on one side of diaphragm
  or
  Localized involvement of an extralymphatic site and of one or more nodal sites on the same side of the diaphragm (IIE)
Stage IIRCompletely resected intra-abdominal disease 
Stage IIITwo or more nodal areas on opposite sides of the diaphragmTwo or more nodal areas on opposite sides of the diaphragm which may include/involvement of the spleen (IIIs)
oror localized involvement of an extranodal site (IIIE)
Primary intrathoracic tumour
Paraspinal or epidural tumours
Extensive intra-abdominal disease
Stage IIIALocalized non resectable abdominal disease 
Stage IIIBWidespread multiorgan intra-abdominal disease 
Stage IVCentral nervous system or bone marrow involvementDiffuse or disseminated involvement of one or more extralymphatic sites
  Two single extranodal tumours on opposite sides of diaphragm
  Also A + B depending on symptoms
FavourableStage I or IIRStage IA and IIA
  Stage I, II and III

Treatment of adult Burkitt lymphoma

General

Prompt treatment with adequate supportive care is essential. Attempts must be made to prevent tumour lysis syndrome upon the instigation of therapy and rigorous hydration and rasburicase are now used routinely (Pui et al, 2001).

In older studies, where adult patients with BL were treated with either CHOP (cyclophosphamide, doxorubicin, vincristine and prednisolone) or with acute lymphoblastic leukaemia (ALL) regimens, the outcome was very poor. In general there was a high complete response (CR) rate but an equally high and usually rapid relapse rate. In a French series of 14 adult patients treated with chemotherapy alone between 1981 and 1987, there were only three survivors (Fenaux et al, 1989). Connors and Klimo (1988) reported that in adults with BL treated with the MACOP-B (methotrexate, doxorubicin, cyclophosphamide, vincristine, prednisone, bleomycin) regimen the CR rate was only 40% and there were no long-term survivors. In a German study of B-ALL (FABL3) treated with a standard ALL-regimen there were also no leukaemia-free survivors (Hoelzer et al, 1996).

In the mid-1980s a number of new regimens were introduced, initially for paediatric BL. These regimens addressed the rapid proliferation rate of BL cells and the propensity of BL to spread to the CNS. In 1986 the Stanford group reported their results in 18 adults with SNCL, using a CHOP-like regimen with intra-cycle intravenous high dose methotrexate and intrathecal methotrexate to provide CNS prophylaxis (Bernstein et al, 1986). Radiotherapy was also given before or during the first cycle of therapy to those with bulky intra-abdominal disease. The CR rate was 78% and the overall survival (OS) was 67% at 2 years, establishing the fact that BL could be cured in a significant proportion of adults, although the median age of these patients was only 25 years. This protocol employed a risk-adapted strategy with six cycles of chemotherapy prescribed for good risk disease and 8–9 cycles for poor risk disease. It is noteworthy that the poor risk patients still did very badly and, as many of them had been recipients of radiotherapy, radiation treatment was largely abandoned except perhaps for CNS disease. Lopez et al (1990) from the MD Anderson Cancer Center (MDACC) reported their results using three alternating NHL regimens to maximize the number of different drugs used, with six cycles of induction and up to nine further cycles of maintenance. Overall the CR rate was 80% but the 5-year OS was a disappointing 52%. There was a high rate of relapse in Stage IV patients, often involving the CNS. This illustrated that there was no benefit to prolonged therapy and emphasized the need for high doses of CNS-penetrating drugs.

Philip et al (1992) reported the successful use of the French Lymphome Malins de Burkitt (LMB) protocol in 14 adults with SNCL, and an extended French experience of 65 adults with SNCL or L3-type ALL (see Table II) was reported 3 years later (Soussain et al, 1995). In these studies a relatively non-intensive one-week treatment pre-phase (COP; cyclophosphamide, vincristine, prednisone) was introduced to minimize tumour lysis syndrome. The therapeutic cycles were more intensive than CHOP with inclusion of the cell cycle phase-specific agent cytosine arabinoside in addition to high dose methotrexate to both increase the kill against rapidly proliferating cells and to augment drug exposure in the CNS. Cyclophosphamide, the most effective single agent in this disease, was also fractionated because although cyclophosphamide is an alkylating agent, it preferentially kills cell in cycle and fractionation over a number of days increases the chance of there being high drug levels at the time the malignant cells entered into cycle. With the more intensive therapy, the total duration of treatment was reduced to 3·5 months. The CR rate was 89% and 74% of patients were alive at 3 years. The LMB 89 regimen was subsequently used by the Groupe d’ Etude des Lymphomes de l’ Adulte (GELA) and Groupe Ouest Est d’Etude des Leucémies et Autres Maladies du Sang (GOELAMS) with broadly similar results (Divinéet al 2005). A Korean group have also demonstrated that a modified LMB protocol is efficacious in Asian patients (Choi et al, 2009).

Table II.  Results of selected regimens in adult BL series with at least 50 patients.
StudyGroupNMedian ageCREFSOS
  1. CR, complete response; EFS, event-free survival; OS, overall survival; LMB, Lymphome Malins de Burkitt; CALGB, Cancer and Leukemia Group B; GELA, Groupe d’ Etude des Lymphomes de l’ Adulte; GOELAMS, Groupe Ouest Est d’Etude des Leucémies et Autres Maladies du Sang; MDACC, MD Anderson Cancer Center; GMALL, German Multicentre Study Group for Adult ALL; CODOX-M/IVAC, cyclophosphamide, cytarabine, doxorubicin, leucovorin, methotrexate, vincristine/cytarabine, etoposide, ifosfamide, methotrexate; Hyper-CVAD, hyperfractionated cyclophosphamide, vincristine, doxorubicin, dexamethasone, R, rituximab.

Soussain et al (1995)French LMB652689%64%74% (3 years)
Lee et al (2001)CALGB regimen544480%42% est.
Mead et al (2002)UK/Multinational CODOX-M/IVAC523577%65% (2 years)73% (2 years)
Rizzieri et al (2004)CALGB 9251     
Cohort I524479% 54% (3 years)
Cohort II405068% 50% (3 years)
Divinéet al (2005)GELA/GOELAMS LMB 89723372%65% (2 years)70% (2 years)
Thomas et al (2006)MDACC     
Hyper-CVAD     
Cohort I484885%52% (3 years)53% (3 years)
Cohort II (+R)314686%80% (3 years)89% (3 years)
Hoelzer et al (2007)GMALL (+R)1153690% 91% (3 years)
Mead et al (2008)UK/Multinational CODOX-M/IVAC5337 64% (2 years)67% (2 years)

Similar therapeutic strategies have been employed by most other groups. The Cancer and Leukemia Group B (CALGB) also incorporated a pre-phase into its 9251 trial (Lee et al, 2001) although such pre-phases are probably unnecessary in most patients and potentially disadvantageous in the rasburicase era. This CALGB 9251 trial was most notable for the very intensive CNS-directed therapy, which included cranio-spinal irradiation and 12 doses of triple intrathecal chemotherapy in the first cohort of 52 patients. This resulted in unacceptable neuro-toxicity with three cases of transverse myelitis, two severe neuropathies, three cases of aphasia and one case of blindness (Rizzieri et al, 2004). The use of cranio-spinal radiation as prophylaxis was subsequently abandoned except in very high-risk patients. The OS in this second CALGB cohort at first appears to be disappointing at only 50%, but it must be noted that the median age of this cohort was 50 years, considerably older than many other studies of this era.

The Berlin-Frankfürt-Münster group also developed BL-specific protocols for B-cell ALL, using six short intensive cycles based on high dose methorexate, fractionated alkylators and moderate dose cytosine arabinoside, and this resulted in an increase in the OS compared to conventional ALL therapy (Hoelzer et al, 1996).

Magrath et al (1996) reported on the use of the CODOX-M/IVAC (cyclophosphamide, cytarabine, doxorubicin, leucovorin, methotrexate, vincristine/cytarabine, etoposide, ifosfamide, methotrexate) regimen in both children and adults. Three cycles of CODOX-M were given to low-risk patients and high-risk patients were treated with a total of four cycles of CODOX-M alternating with the hopefully non-cross resistant IVAC regimen (i.e two cycles of each regimen). Each cycle was given as soon as haematological recovery had occurred from the previous cycle to minimize inter-cycle tumour re-growth, and as a result, even with the intensive CODOX-M alternating with IVAC protocol treatment could be completed in less than 4 months. The low risk regimen could be completed within 2·5 months. The results were excellent for both children and adults but the median age of the adults was only 24 years. The CODOX-M or CODOX-M/IVAC regimen (depending on risk status) was subsequently tested in a larger number of adult patients up to the age of 60 years in a multi-national study, in which the median age was 35 years. Low risk was defined in this study by a score of 0 in the age-adjusted International Prognostic Index [i.e Ann Arbor Stage I or II, normal lactate dehydrogenase (LDH) and a WHO performance status of 0 or 1] with no bulky disease >10 cm. The CODOX-M regimen was modified in so far as one of the three vincristine injections given with each cycle was omitted (day 15) because of concerns about vincristine-induced neurotoxicity in older patients. Overall, the 2-year event-free survival (EFS) was 64·6% and the 2-year OS was 72·8%. In a follow-on study with more rigorous definition of BL, but a wider definition of low risk, and a reduction of the methotrexate dose in CODOX-M from 6·7 to 3·0 g/m2, the 2-year progression-free survival was 64% and the 2-year OS was 67%. The authors therefore suggested that this lower dose of methotrexate, which is associated with less toxicity, should be used in future regimens. The Boston group similarly reduced the intravenous methotrexate dose to 3 g/m2 (Lacasce et al, 2004).

The Hyper-CVAD regimen (hyperfractionated cyclophosphamide, vincristine, doxorubicin, dexamethasone), initially developed at the MDACC for use in adult B-ALL, was also applied to the non-leukaemic forms of this disease, with a CR rate of 85% and a 3-year EFS of 52% (Thomas et al, 2006). Although the EFS and OS was less than in some series it is important to note that the median age in this series was 58 years.

The DA-EPOCH regimen (dose-adjusted etoposide, prednisone, vincristine, cyclophosphamide, doxorubicin), which uses infusional drug scheduling and pharmacodynamically-modified drug dosing based on haematological toxicity, was originally introduced for the treatment of relapsed and resistant DLBCL (Wilson et al, 1993) and was later used successfully in immunodeficient patients with BL (see below).

Rituximab was generally incorporated into the various BL regimens as its value in other forms of high-grade lymphoma became apparent. There was concern that the addition of rituximab to already highly myelotoxic and immunosuppressive regimens would result in unacceptable toxicity but this does not seem to occur (Maruyama et al, 2010; Mohamedbhai et al, 2011) The outcomes associated with the addition of rituximab appear to have improved dramatically, although some caution must of course be exercised when comparison is made with historical controls. For instance, when rituximab was added to Hyper-CVAD in 31 adult patients with BL, the 3-year OS was 89% with equally good results in the elderly patients (Thomas et al, 2006). In a German Multicentre Study Group for Adult ALL (GMALL) trial of 115 patients with BL or BLL, the addition of rituximab to their previously used strategy was associated with a major improvement in outcome with a CR rate of 90%, and in patients less than 55 years of age the OS at 3 years was 91% (Hoelzer et al, 2007). In BL patients over 55 years of age the results were again good, with an OS at 3 years of 84%. In a Japanese study, the addition of rituximab to a modified CODOX-M/IVAC regimen resulted in a CR rate of 87% and an excellent estimated 5 year survival of 87% (Maruyama et al, 2010). Addition of rituximab to DA-EPOCH in 29 patients resulted in 28 achieving a CR and 100% being alive at 2 years, although it must be noted that none of these patients had CNS or bone marrow involvement (Dunleavy et al, 2011). No randomized trials of rituximab have been performed in BL and it is most unlikely that adequately powered studies will ever be performed. Nonetheless, the data is reasonably convincing that rituximab represents a major advance in the treatment of adult BL and its use may negate the previous poor prognosis associated with older age.

There are inevitable differences in the prognostic factors identified in the various studies and this is more likely to represent statistical noise rather than real differences with the different regimens used. There is common agreement that older age is a poor prognostic factor with data from an overview of 14 series suggesting that the threshold for high risk is 40 years of age although the differences due to age do appear less marked in the more recent series (Kelly et al, 2009). Advanced stage, especially bone marrow and CNS involvement, and a raised LDH level are also poor prognostic factors. It is clear from nearly all reports that failure to achieve a CR after 4–6 weeks is an extremely poor feature as salvage is highly problematic. Most groups use prognostic factors to stratify patients into different therapeutic regimens, with the high-risk patients either receiving more cycles of therapy or alternating non-cross resistant regimens or both as in the CODOX-M/IVAC regimen. In accord with such an approach, the 2008 National Comprehensive Cancer Network (NCCN) guidelines recommend CODOX-M or Hyper-CVAD for low risk disease or CODOX-M/IVAC or Hyper-CVAD alternating with methotrexate and cytarabine either with or without rituximab for high-risk disease (http://www.nccn.org/professionals/physician-gls/pdf/nhl.pdf).

The good risk patients are spared unnecessary and potentially toxic therapy, which is fully reasonable in the younger adults where the cure rate approaches 100%. In older patients where a proportion of low risk patients will fail on the reduced regimens, one must question the appropriateness of such stratification. It may be that instead of stratification based on risk factors, it should be based on early response assessment. Information is needed on the results of PET scanning after one and two cycles of therapy and on whether this accurately predicts for outcome. With regard to the high-risk patients it is noteworthy that even when risk adapted therapy is employed as in the CODOX-M/IVAC protocol, a difference in outcome remains for the different risk groups. In the UK LY06 trial, for example, the 2-year OS for low risk patients was 82% and for high-risk patients was 70% (Mead et al 2002). Such differences may be reduced when rituximab is added to the regimens but more data is needed.

Atypical BL, BLL, BLU and DLBC lymhomas with a MYC translocation

There is general agreement that atypical BL, BLL and BLU fare relatively poorly when treated with CHOP or CHOP-like regimens. Nomura et al (2008) identified a series of patients with either BL or atypical BL who had been treated with CHOP and they fared much worse than those treated with BL-specific regimens (OS of 22% cf 75%P < 0·01). The inference is that patients with atypical BL, BLL and presumably BLU should be treated with BL-specific regimens although evidence that such patients do better with BL-specific therapy is limited. The GMALL group reported that BL and BLL had a similar outlook with a rituximab-containing regimen (Hoelzer et al, 2007) but most reports either do not give the results of these histological subtypes separately, or have too few patients for a meaningful comparison. Many BLU cases have a ‘double hit’ and they appear to have a particularly poor prognosis (Le Gouill et al, 2007; Lin & Medeiros, 2007). Patients with DLBCL with a MYC rearrangement also fare less well than other cases of DLBCL when treated with R-CHOP (Barrans et al, 2010). Savage et al (2009) examined a series of 135 cases of DLBCL by FISH and detected a MYC rearrangement in 12 cases. The OS with R-CHOP was only 33% compared to 72% in those who did not have a MYC rearrangement.

Another group identified 54 patients with both MYC and BCL2 rearrangements; the OS at 3 years was <20% in those treated with CHOP and was only about 35% in those treated with R-CHOP (Johnson et al, 2009). It seems reasonable, therefore, to also treat cases of DLBCL with a MYC re-arrangement with BL-specific regimens, but once again, there is no solid evidence that this will result in an improvement in outcome. In one series all 4 ‘double hit’ lymphomas treated with CODOX-M/IVAC died within 6 months of diagnosis, which is certainly not encouraging (Mead et al 2008).

Adolescent patients

There is long-standing debate as to whether adolescent patients with leukaemia and lymphoma should be treated with paediatric or adult protocols. This is less of an issue in BL, however as similar regimens are used for all patients but the elderly. It is worth noting however that adolescents generally fare only marginally worse than children treated on the same protocols (Burkhardt et al, 2011) and in the French-American-British (FAB)/LMB 96 trial the excess failure in adolescents compared to children could be accounted for by the distribution of stage and sites of disease (Hochberg et al, 2009). This trial clearly demonstrated that reductions in therapy were possible without detriment to good risk patients and this can be reasonably extrapolated to the adolescent population. In the very small group of patients with resected Stage I disease or fully resected abdominal Stage II disease it was possible to give just two cycles of COPAD (cyclophosphamide, vincristine, prednisone, doxorubicin) with 99% overall survival (Gerrard et al 2008) and COPAD × 2 would presumably have less toxicity than three cycles of CODOX-M or some other regimens used for good risk disease. In the FAB/LMB 96 trial it was also shown that therapy could be reduced in intermediate risk early-responding patients with a total dose of cyclophosphamide of 3·3 g/m2 and doxorubicin of 120 mg/m2 (Patte et al, 2007). This compares to 4·8 g/m2 and 120 mg/m2 of cyclophoshamide and doxorubicin, respectively, with three cycles of CODOX-M. Finally it should be noted there is some reticence about the use of rituximab in paediatric patients, but it is the view of this author that rituximab should be given to adolescents.

Elderly patients

Adult BL is relatively frequent in elderly patients who are often not included in published series and they represent a special challenge. If such patients are to be treated with the most intensive regimens that they can tolerate, then this requires very careful assessment at presentation and continual re-evaluation of co-morbidities, toxicities and functional capacities during treatment. As discussed above, rituximab appears to be particularly valuable in older patients and its use should be mandatory. It is notable that good results have been reported with the DA-EPOCH-R in elderly patients, this is important as this regimen is less toxic than many other regimens and is potentially an out-patient treatment for most of the time. The excellent results reported in the elderly in the GMALL study (Hoelzer et al, 2007) are also of particular interest because in the older patients the high dose cytosine arabinoside was omitted and the dose of high dose methotrexate was reduced to 500 mg/m2. Some elderly patients cannot tolerate even these regimens, however, and standard R-CHOP is a reasonable alternative in those fit enough to receive anthracyclines but not more intensive regimens.

Immunodeficiency-associated BL

A key feature of immunodeficiency-associated BL is the higher incidence of Epstein-Barr virus positivity compared to sporadic adult disease (30–40% cf approximately 20%) (Perkins & Friedberg, 2008). In human immunodeficiency virus (HIV)-associated BL there is also a subgroup with marked plasmacytic differentiation which is very rare in sporadic disease. An intriguing observation has been made, that very low levels of CD4+ cells (<0·05 × 109 cells/l) are unusual in acquired immunodeficiency syndrome (AIDS) patients with BL (cf DLBCL), raising the possibility that the CD4 cells provide a survival signal in cells with markedly upregulated MYC (Guech-Ongey et al, 2010). Understanding of this mechanism may point to novel therapeutic approaches. A corollary of the association of BL with relative high CD4 counts is that the introduction of effective anti-retroviral therapy has not reduced the incidence of BL as it has for other types of AIDS-related lymphomas (Biggar et al, 2007).

Until recently it was thought that the severely immuno-compromised patient could not tolerate the intensive chemotherapy regimens used in sporadic BL, this view was supported by the finding of a worse outcome associated with rituximab (+CHOP) in a randomized trial for patients with AIDS-related lymphomas (Kaplan et al, 2005). This view no longer predominates, and almost all centres advocate aggressive short duration regimens although some still withhold rituximab (Perkins & Friedberg, 2008). The change in attitude is partly due to the efficacy of modern anti-retroviral therapy in AIDS, with these drugs generally being given concurrently with intensive chemotherapy. In addition, there has been the realization discussed above that BL patients are not the most severely immuno-suppressed, although a low CD4 count remains a poor prognostic factor (Galicier et al, 2007). One study compared the toxicity of an intensive immuno-chemotherapy regimen incorporating rituximab in HIV-infected and non-infected patients, and although there was more mucositis and more severe infections in the HIV positive patients, there was no significant difference in survival (Oriol et al, 2008). In the series reported by Galicier et al (2007), using the LMB86 regimen, 70% of 63 HIV-infected patients with Murphy Stage IV disease achieved a CR and the estimated OS at 2 years was 47%. A comparison of BL patients treated in California suggested that the relative mortality rate at 2 years was 1·3 (1·0–1·7) in HIV-infected compared to non-infected individuals (Chao et al, 2010). The successful use of DA-EPOCH in 39 newly diagnosed patients with AIDS-related lymphomas has been reported (Little et al, 2003); seven had BL and three survived. Rituximab was added to this regimen, which is of particular interest as the high response rate reported in sporadic disease is maintained in HIV-positive patients (Sparano et al, 2010).

Transplantation

In 1978 the Seattle transplant team reported their experience of high dose chemotherapy followed by autologous bone marrow transplantation in 14 adult patients with BL resistant to standard therapy (Appelbaum et al, 1978). The procedure-related mortality was high, with four early deaths. Of the surviving patients, seven progressed and died but three remained in complete remission. Philip et al (1988) subsequently reported on the results of high dose therapy and autologous bone marrow transplantation in a mixture of 17 children and adults who had only achieved a partial response (PR) after induction therapy and the OS at 2 years was an excellent 75%. Sweetenham et al (1996) reviewed the European Bone Marrow Transplant (EBMT) experience of autologous transplantation in 117 patients with BL or BLL. Thirty-two of these patients had chemosensitive relapsed disease (including PR to front-line therapy) and the OS at 3 years in this group was a more modest 37%. Although these results are not as impressive as the earlier study (Philip et al, 1988), the high dose therapy and autologous stem cell transplantation still remains as the standard of care for such chemosensitive patients who have failed to achieve CR or who have relapsed. For those patients with chemoresistant disease in the EBMT registry the OS was only 7%, indicating that high dose therapy and autologous transplantation is generally an unsuccessful strategy in all forms of chemoresistant lymphoma. The generally positive results in chemosensitive disease encouraged the use of high dose therapy and ASCT as a component of first line therapy. Jost et al (1995) reported on their experience of treating nine patients with a standard weekly regimen for 12 weeks followed by CBV [cyclophosphamide, BCNU (carmustine), etoposide] autografting. The results were disappointing, with 42% of patients failing to reach transplantation, indicating that late intensification cannot substitute for initial intensive therapy. van Imhoff et al (2005) reported on a similar strategy where high dose sequential therapy was followed by BEAM (BCNU, etoposide, cytarabine, melphalan) autograft providing at least a PR had been attained. Of the 27 adults (median age 33 years) with BL or BLL, the EFS was 73% and the OS was 81%. Nademanee et al (1992) restricted transplantation to 10 SNCL patients with high-risk features who had already achieved CR; the disease-free survival and OS at 3 years was 60%. The EBMT Registry contained 70 similar patients with BL or BLL, deemed to be high risk, who were transplanted in first CR and the OS at 3 years was 72% (Sweetenham et al, 1996). These results were initially thought to be a significant step forward, but with the realization of the greatly improved results that were being achieved with short duration intensive therapy without autologous transplantation, this modality was soon abandoned for those already in CR.

Troussard et al (1990) reported the results of allogeneic transplantation in first CR in nine adult patients with BL, deemed to be at high risk of relapse by virtue of having had CNS disease at presentation and with bone marrow involvement in seven of them. Conditioning was with cyclophosphamide and total body irradiation and seven patients remained disease-free beyond 18 months. It must be noted however that the median age was only 21 years with a range of 15–36 years. Peniket et al (2003) subsequently analyzed the EBMT registry experience of allografting in a mixture of 71 children and adults with BL and there was no obvious difference from the results with autografting. Although a patient who relapsed after an allograft and then went into remission when immuno-suppression was removed, in association with the development of graft-versus-host disease, has been reported (Grigg & Seymour, 2002), in the EBMT series there was no obvious graft-versus-BL effect. In a matched analysis the relapse rate was not lower in the allografted patients than in those who had received an autograft, in contrast to the situation in other types of lymphoma. This may relate to the rapid growth rate of any residual disease after the high dose procedure, swamping any graft-versus-lymphoma effect before it can have any impact, and if this is the case, it does not bode well for the use of reduced intensity allogeneic transplantation.

The future

The rigorous diagnosis of BL and BLU in adults can still be difficult, and although GEP may improve categorization, is not yet routinely available. If, however, BLU (and atypical BL and BLL) as well as DLBCL with MYC re-arrangements are, in future, all treated in the same way as BL then there may be less concern over precise classification in individual cases. The lack of randomized trials means that it is impossible to pick a ‘best regimen’ although rituximab should be a component of all regimens. Significant issues remain about the individual drugs used in most regimens, including the optimal fractionation and dose of the alkylating drugs used, the optimal dose of both methotrexate and cytarabine, from the perspective of both controlling systemic disease and from preventing CNS progression, and whether there is any benefit to intrathecal therapy when high doses of systemic methotrexate and cytarabine are given. It is usual to employ risk-adapted strategies in the treatment of BL but this must be continually re-evaluated as new treatments develop. Reduced regimens in good risk children and young adults are fully appropriate, but may not be so in older patients, where some patients deemed to be low risk are not cured by such an approach. On the other hand, the impact of rituximab may allow more general dose reductions of other drugs as shown in older patients in the GMALL trial (Hoelzer et al, 2007). One way forward may be to utilize response-adapted rather than risk-adapted therapy and more information is urgently required on the predictive value of CT/PET scans after 1 and 2 cycles of therapy. There is scope for novel approaches and the MYC gene is an obvious target, although down regulation of MYC might not only reduce tumour proliferation but could also reduce the tumour’s propensity to chemotherapy-induced apoptosis. To resolve existing therapeutic uncertainties and to facilitate future advances, randomized trials are required and it is unacceptable to say the disease is too rare in adults for such trials to be delivered. Consideration should be given to combined paediatric and adult trials and improved international collaboration.

Ancillary