Diagnosis and Treatment of Childhood Non-Hodgkin Lymphoma

Non-Hodgkin lymphoma (NHL) comprises a heterogeneous group of lymphoid neoplasms. The distribution of subtypes according to the World Health Organization Classification of Tumors of Haematopoetic and Lymphoid Tissues (WHO-Classification)¹ is significantly different in childhood and adulthood. In children, lymphoblastic lymphoma (LBL) of the precursor B- or T-cell type, Burkitt lymphoma (BL)/ leukemia (B-ALL), and anaplastic large-cell lymphoma (ALCL) predominate, while the proportion of diffuse large B-cell lymphoma (DLBCL) increases with increasing age. The typical NHL subtypes of childhood exhibit significant differences in terms of their molecular and cellular biology and their clinical features, which may be crucial for determining therapeutic strategies. The most important disparities in NHL subtypes are significantly different cell-cycle kinetics and different dispositions for invasion of the bone marrow (BM) and central nervous system (CNS), which is much higher in LBL and BL than other subtypes.² 

The Children’s Cancer Group randomized trial CCG-551,³ which compared the LSA₂-L₂ protocol (cyclophosphamide, vincristine, methotrexate, daunorubicin, prednisone, cytarabine, thioguanine, asparaginase, carmustine, hydroxyurea) with COMP (cyclophosphamide, vincristine, methotrexate, prednisone) was pivotal for stratifying treatment modalities according to biological subgroups and revealed three main findings: (1) different chemotherapy regimens exert different effects in different NHL subtypes; (2) differences in treatment efficacy are seen mainly in advanced-stage disease; and (3) in advanced-stage disease, the differences in treatment efficacy are more pronounced in patients with LBL (i.e., patients receiving LSA₂-L₂ had fewer relapses) and BL (i.e., patients receiving COMP did better), while event-free survival (EFS) rates were not significantly different between treatment regimens in patients with large cell lymphoma.

Until recently, two different methods were pursued to stratify therapy for childhood NHL: stratification of treatment modality according to histologic subgroups, and adaptation of treatment intensity according to stage and additional criteria. In the second pattern, primary stratification was for localized versus advanced-stage disease, with uniform treatment for localized disease of any histology, while subgroup-directed treatment was used in patients with advanced-stage disease. However, in a Pediatric Oncology Group trial, it was shown that even in patients with localized disease, different strategies had different effects in histological subgroups.⁴ A 24-week maintenance therapy in addition to a 9-week induction had a beneficial impact on outcome for patients with LBL but not for those with BL and large-cell lymphoma.

There are currently three major subgroups of childhood NHL that are distinguished with respect to treatment strategy (Figure 1 ): LBL of the precursor B- (pB-LBL) and T-cell types (T-LBL), B-NHL, and ALCL. Therapeutic protocols used for ALL, which are based on the principle of continual exposure to cytostatics over a long period of time, are efficacious for treating children with LBL.^3,5,6 In contrast, a strategy of rapidly repeated short, dose-intense chemotherapy courses have been shown to be more successful for treating patients with BL/B-ALL and proved also highly efficacious for treating patients with DLBCL.^7,–10 Although children with ALCL had comparable outcomes with either of those treatment strategies, the ALCL subtype emerged as a separate treatment group.^11,–14 The main reasons for this were different prognostic parameters with relevance to stratification of treatment intensity and a higher chance of survival after relapse compared to the other main NHL subtypes.¹⁵ For rarer and currently less accurately defined NHL subtypes in children, including the small but heterogeneous group of peripheral T-cell/natural killer cell lymphoma (PTCL/NK), optimal treatment is not yet established. That and other rare NHL subtypes, such as primary mediastinal (thymic) large B-cell lymphoma (PMLBL) and (juvenile) follicular lymphoma, are candidates for new subtype-specific treatments.

Comprehensive diagnostic evaluation and classification of cases is essential not only for correct allocation of patients to currently established subgroup treatment regimen but also for further identification of biologically distinct subtypes requiring different future treatment. Figure 1  depicts a rational diagnostic work-up program for children with NHL. Regarding the role of initial surgery, see below. Caution! Any invasive diagnostics may be dangerous and should, therefore, be postponed in patients with upper vena cava syndrome and respiratory distress due to a mediastinal tumor. In such cases, immediate treatment with corticosteroids, potentially combined with cyclophosphamide, for up to 48 hours is beneficial and unlikely to obscure subsequent pathologic diagnosis. Critical pleural or/and pericardial effusions require drainage and may enable comprehensive diagnostics.

Cytomorphology, histomorphology, and immunophenotyping are basic diagnostic methods. In most cases, they enable correct classification and allocation of patients to appropriate treatment subgroups. However, according to the WHO classification, cytogenetics is also required for diagnosis in certain cases, such as variant atypical BL/BL-like.¹ 

Many children present with advanced-stage disease, including advanced BM invasion or/and malignant effusions. In most such cases, correct diagnosis can be made by cytology, immunophenotyping by flow cytometry, and cytogenetics (Table 1 ). If this approach is not possible, diagnosis is based on biopsy, and most cases are correctly classified by cytology of tumor touch imprints, histomorphology, and immunohistochemistry with the available wide range of paraffin-resistant antibodies.

Regarding appropriate allocation of patients to therapy groups, there are some crucial differential diagnoses that require additional selected diagnostics. Typical interfaces are distinguishing LBL of precursor B-cell type from BL/BL variants or DLBCL; T-cell rich DLBCL from nodular lymphocyte predominant Hodgkin lymphoma; PMLBL from nodular sclerosis type Hodgkin lymphoma; and ALCL from other peripheral T-cell/NK-cell lymphomas, anaplastic variants of DLBCL, or Hodgkin lymphoma. In these cases, immunohistochemical staining for expression of genes associated with differentiation compartments such as terminal deoxynucleotidyl transferase (TdT; positive only in precursor B- and T-cell neoplasms) and the germinal center cell-associated marker bcl-6 can be helpful.¹⁶ Finally, the identification of subtype-specific chromosomal translocations may be decisive. However, appropriate material for cytogenetic evaluation is not always available in NHL patients. In such cases, fluorescence in situ hybridization (FISH), which can be performed on tumor touch preparations, or paraffin sections, is a valid method for visualizing most of the currently known subtype-specific chromosomal translocations.¹⁷ In many cases, breakpoint-spanning DNA fragments or subtype-specific fusion gene transcripts of specific chromosomal translocations can be detected by polymerase chain reaction (PCR).¹⁸ Recently, it was shown that BL has a reproducible gene expression profile that can be used to identify and distinguish molecular BL from other subtypes.^19,20 

Newer methodologies, such as identification of genetic changes in tumor cells and genome-wide gene expression profiling, will become increasing important for identifying biologically distinct subtypes and therapy targets. Therefore, whenever possible, following proper diagnostic classification of an individual patient, appropriate material should be preserved for future research (e.g., purified tumor cells, shock-frozen tumor tissue).

In large, multicenter studies, EFS rates of 60% to more than 80% were achieved, even for children with advanced-stage T-LBL (Table 2 ).^5,6,21,–23 Currently, the most frequently used treatment regimens are the LSA₂-L₂ protocol in numerous modified forms and the Berlin-Frankfurt-Münster (BFM) group strategy, which was originally designed to treat children with ALL. Both protocols are divided into phases of induction, consolidation, reintensification, and maintenance, and include corticosteroids, vincristine (VCR), anthracyclines, L-asparaginase (L-Asp), cyclophosphamide (CP), methotrexate (MTX), cytarabine, 6-mercaptopurine (6-MP), and 6-thioguanine. The main differences between the protocols are earlier application of L-Asp and high-dose (HD) MTX (5 g/m² intravenous over 24 hours) in the BFM regimen. Treatment duration for both regimens was 18 to 24 months. Repeated continuation courses, including CP and anthracycline until the end of therapy, are part of the LSA₂-L₂ protocol, while maintenance includes only oral 6-MP and MTX in the BFM strategy. Most relapses occur early, and late relapses are rare. In patients with T-LBL most relapses occur during the first 12 months after diagnosis, suggesting that the duration of maintenance can be reduced.⁵ 

The contribution of individual drugs to patient cure is largely unknown due to the shortness of randomized trials. For L-Asp, the effect in T-LBL was demonstrated in the POG-8704 trial when patients did or did not receive weekly L-Asp × 20 after induction.²¹ With backbone BFM protocols, no additive effect on outcome was observed for HD cytarabine in the consolidation phase nor for intensification blocks up-front of the BFM induction protocol.^22,24 Currently, randomized BFM-based trials are ongoing to determine whether dexamethasone instead of prednisone during induction can further improve outcome of patients with LBL, as was observed in children with ALL. Results of a COG trial that examined the impact of HD MTX during consolidation and up-front intensification with CP and anthracycline are pending.

Treatment intensity is mainly stratified according to stages I and II versus stages III and IV. Children with stage I/II are rare. Most have pB-LBL and achieve EFS rates higher than 90% with reduced-intensity (omission of reintensification in the BFM protocol) and full-length maintenance therapy. Whether treatment can be reduced further is difficult to determine. In a POG trial, a 24-week maintenance in addition to a 9-week induction was beneficial for patients with LBL, although their EFS rate was only 63%.⁴ This suggests that their biological similarity to ALL is more important than their low tumor burden and that they might, therefore, benefit from an ALL-type treatment, including maintenance.

For patients with overt CNS disease, 18 to 24 Gy cranial irradiation (CRT), in addition to LSA₂-L₂ or BFM chemotherapy, is highly effective in preventing CNS recurrences.^2,5,6 For CNS-negative patients, treatment that includes intrathecal MTX and systemic HD MTX (0.5–5 g/m²) but no CRT is sufficient CNS protection in children with stage I/II disease.^2,4,5 Although a randomized trial is not available, cumulative evidence suggests that preemptive CRT can also be omitted for CNS-negative patients with advanced-stage disease. In study NHL-BFM 95, including 4 × HD MTX (5 g/m² intravenous over 24 hours) and 11 doses of intrathecal MTX but no CRT, disease-free and CNS relapse-free survival of patients was not significantly inferior to the historic control group of the preceding trials NHL-BFM 86 and 90, in which patients received prophylactic CRT.²³ HD MTX also appears to be efficacious in preventing testes relapse.⁵ 

Chemotherapy strategies for B-NHL are tailored to the biological and clinical characteristics of BL and are also efficacious for patients with other B-NHLs, especially DLBCL. The prototype therapy courses of the currently most-used treatment regimens were developed in the St. Jude Total B program, the French LMB, and the German-Austrian-Switzerland BFM NHL studies. Based on the extremely high proliferative activity of BL, a basic principle is to maintain cytotoxically active drug concentrations over a period that is sufficient to affect as many lymphoma cells as possible during the vulnerable active cell cycle, using either fractionated administration or continuous infusion.²⁵ Other principles are combining drugs with different mechanisms of action and few overlapping toxicities; high-dose intensity over time by keeping between-treatment intervals short; and efficient CNS-directed therapy to address the strong tendency for invasion of the CNS, especially that of BL. Therapeutic strategies that adhere to this principle of rapidly repeated 4- to 7-day courses composed of corticosteroids, VCR, CP or ifosfamide, HD MTX, cytarabine, doxorubicin, etoposide, and triple drug (MTX/cytarabine/corticosteroid) intrathecal therapy resulted in EFS rates up to 90% in large, multicenter studies (Table 2 ).^7,–9,26 Evidence for the importance of CP, VCR, and MTX was derived from early studies on BL in Africa. Although randomized comparisons are lacking, evidence for the effect of the MTX dose can be derived from the significant relapse reduction in patients with advanced-stage disease and high tumor mass after a multifold increase in MTX dose from 0.5 g/m² to 3.0, 5.0, and 8.0 g/m².^7,8 The efficacy of HD cytarabine combined with etoposide was demonstrated by remission induction in patients who failed conventional therapy.²⁷ 

The current highly efficacious regimen correlates with considerable acute toxicity which could not be decreased by post-chemotherapeutic granulocyte colony-stimulating factor (G-CSF).²⁸ Patients with advanced-stage disease are at significant risk of roughly 3% to die from treatment-related complications, especially during the first phase of therapy.^7,8,26,29 Severe oro-intestinal mucositis, caused primarily, but not solely, by HD MTX, and severe neutropenia are the most important acute toxicities, synergistically promoting serious infections. Metabolic disturbances of acute tumor cell lysis syndrome (ATLS) is another serious threat in the first days of treatment that decreased, but did not completely disappear, after the introduction of a cytoreductive pre-phase, usually consisting of corticosteroid and low-dose cyclophosphamide and urate oxidase to prevent or treat ATLS.³⁰ 

In most North American studies, treatment intensity was stratified according to St. Jude stage, while in the French LMB and BFM studies, additional criteria (e.g., resectability, tumor mass, and CNS involvement) were explored (Table 3 ). The main goal of the recent multicenter trials FAB/LMB-96 (based on the French LMB-89 protocol) and NHL-BFM-95 was to reduce treatment burden (Table 3 ).^26,29,31,32 Patients with localized resected tumors have nearly 100% EFS with two 5-day therapy courses and may not need MTX at all. A favorable balance of efficacy and toxicity was observed with 4 courses of therapy that included 3 g/m² MTX given intravenously over 3 hours (FAB/LMB group B) and 1 g/m² MTX given intravenously over 4 hours (BFM group R2) in patients with unresected lymphoma of intermediate risk (group B in the FAB/LMB-96 trial, group R2 in the NHL-BFM 95 trial; Table 3 ). In the FAB/LMB-96 trial, intermediate-risk patients (group B) with response to the 7-day cytoreductive prophase COP (cyclophosphamide/vincristine/prednisone) and the first course COPADM were randomized in a factorial design into 4 arms, 2 receiving half-dose cyclophosphamide in the second induction course and 2 with omission of the maintenance course M. pEFS was equal with reduced-dose cyclophosphamide and omission of course M. In both trials, attempts to reduce treatment burden in high-risk patients failed, however. Data from the NHL/BFM-95 study showed that toxicity and antitumor efficacy of MTX depends on duration of exposure to the drug. A 4-hour MTX infusion is less toxic than a 24-hour infusion but is also less efficacious.²⁶ The required efficacy of MTX appears to differ according to the patient risk for failure. For patients in the low- and intermediate-risk groups R1 and R2, 1 g/m² MTX over 4 hours was as efficient and less toxic than the 24-hour infusion. Mucositis grades III/IV and infection grade III were observed after 6% and 2% of the treatment courses. Patients at high risk for relapse (groups R3+R4) benefit from higher MTX efficacy grades in terms of dose and exposure duration. Regarding treatment duration, there is little rationale for more than 6 (BFM) to 8 (FAB/LMB-96) courses, even for high-risk patients, as roughly one-third of relapses occur during therapy.^7,8,26,29 

With chemotherapy alone, including systemic MTX in risk-adapted doses of 0.5 to 8 g/m² combined with triple-drug intrathecal therapy, CNS relapse is rarely seen in patients without overt CNS disease at diagnosis.^{2,7,8,26,29,32} Intrathecal therapy may even be dispensable in patients with resected localized tumors except those with head and neck tumors.² Outcomes of patients with overt CNS involvement are inferior to those of patients who are CNS-negative with advanced-stage disease, including BM involvement.^2,7,8,26,29 There are no randomized trials testing the role of cranial irradiation in patients with CNS-positive B-NHL. In recent studies patients with CNS-positive B-NHL had EFS rates of 70% with LMB and BFM protocols that included 3 g/m² HD cytarabine; 8 g/m² and 5 g/m² HD MTX intravenously over 4 and 24 hours, respectively; and intensive intrathecal triple-drug therapy applied via lumbar puncture (LMB) or fractionated intra-ventricularly (BFM).^8,26,29 These results were comparable with the outcome of CNS-positive patients receiving in addition cranial irradiation in a previous LMB study.⁷ Testicular relapse is rare with therapy strategies that include HD MTX.^7,8 

EFS rates of 65–75% were achieved with different therapeutic strategies including LSA₂-L₂-type protocols and short-pulse B-NHL–type protocols (Table 2 ).^11,–14,33 Treatment durations of these regimens varied between 2 and 5 months and 2 years. Most tumor failures occur within the first 15 months after diagnosis, although late relapse has been observed with all treatment strategies. Some patients experience progression very early in treatment.

Due to the heterogeneous nature of treatment regimens, only limited conclusions can be drawn about the roles of their individual components. Alkylating agents, HD MTX, and etoposide are main components of most regimens, but were absent in the 6-drug APO regimen of the POG-8704 trial at the expense of a high cumulative dose of doxorubicin.³³ Thus, one might hypothesize that doxorubicin, VCR, and steroids are key component drugs. An intriguing observation was the efficacy of vinblastine in multiply recurrent ALCL.¹⁵ The role of vinblastine in front-line therapy is currently under investigation.

The criteria for stratifying treatment intensity for ALCL are less well established than for other NHL subtypes. Patients with complete resected stage I appear to require only short treatment of three 5-day courses.¹² In an ongoing trial, patients were allocated to the high-risk group based on the presence of at least 1 risk factor (skin involvement, mediastinal mass, visceral involvement of the lung, liver, or spleen).³⁴ 

ALCL has a moderate tendency for CNS invasion. The incidence of CNS relapse was low without preemptive CRT in all studies.^2,11,–14 A recent randomized trial demonstrated that HD MTX (3 g/m² intravenously over 3 hours) was sufficient to protect the CNS in CNS-negative patients in the absence of additional intrathecal chemotherapy.³⁴ For the rare patients with overt CNS disease, 18 to 24 Gy CRT may be an option in addition to HD MTX, HD cytarbine, and intrathecal triple-drug therapy.^12,14 To date, testicular involvement has not been reported.

Initial surgery is primarily diagnostic to provide sufficient biopsy material. Patients with completely resected small localized B-NHL and ALCL have an EFS of nearly 100% after 2 and 3 short chemotherapy courses, respectively. The impact of resection on outcome of these patients is unknown, however. In advanced stages, gross sectional surgery does not contribute to cure but may delay the onset of chemotherapy and is, therefore, obsolete.

There is little evidence for a beneficial role of local irradiation. Across all entities, patients with localized disease have excellent outcomes with chemotherapy alone. Patients in advanced stages who still experience relapses often have widespread disease, making application of radiotherapy difficult and toxic. Nevertheless, since local sites are a frequent site of recurrence, there may be select patients who would benefit from local irradiation. The question is how those patients can be accurately identified, while protecting other patients from the late risks of irradiation (see below).

Further refinement of the balance between treatment burden and risk of failure is warranted. However, reduction of treatment intensity may be dangerous in the absence of a salvage strategy with proven efficacy. Most relapses occur early, leaving patients with highly refractory tumors and pre-existing morbidity that results from ongoing or recently completed first-line therapy. Early identification of those patients who will fail current treatment and alteration of their treatment may enhance their chances of survival. Thus, identifying the parameters that predict current treatment outcome with high accuracy is a major task. Within the established treatment subgroups, the histologic subtype has no adverse prognostic impact, except for PMLBL³⁵ and, possibly, the lymphohistiocytic variant of ALCL.^14,36 Table 4  depicts parameters that have prognostic significance for current treatments in addition to St. Jude stage. Many of the parameters are too limited in predictive strength to provide the basis for major alteration of current treatment, however.

The availability of new methodologic tools will greatly enhance our ability to distinguish additional biologically distinct subtypes beyond histology and can enable application of subtype-specific therapies. However, the application of such methods (e.g., gene expression profiling) is often impeded by rather trivial causes, including a lack of appropriately preserved tumor material. A lack of tumor material is also a major limitation in the availability of cytogenetic analyses. Preliminary data suggest that secondary chromosomal aberrations in patients with BL are associated with a high risk of failure (Table 4 ).^37,38 

The kinetics of treatment response is a strong prognostic parameter in many malignancies and is underexplored in childhood NHL. The primary reason for this may be methodological difficulties in reproducibly evaluating the kinetics due to the complexity of lymphoma manifestations. Incomplete tumor regression during induction is a frequent observation. Its affect on the subsequent course differs, however, since the tumor remnant may be a result of resistant lymphoma or persistent fibrous or necrotic tissue. Conventional imaging and second-look surgery have limited value in distinguishing patients who are prone to subsequent progress and those for whom the tumor remnant represents the fibrous-necrotic waste of the initial lymphoma with no impact on outcome. The only exception is unequivocal viable tumor on second-look OP.^7,8,26,29,32 Functional imaging using positron emission tomography with F18 fluorodeoxy-D2-glucose (FDG-PET) may provide a new tool for early assessment of treatment response. In aggressive adult NHL, negative FDG-PET after the first two courses of chemotherapy was associated with a 2-year EFS of 82% compared with a EFS of 43% for patients with positive FDG-PET.³⁹ Whether these findings generalize to childhood NHL is unclear, since FDG-PET avidity differs significantly between lymphoma subtypes.⁴⁰ Furthermore, to determine treatment guidelines, prospective evaluation of prognostic accuracy will be necessary in the context of a given treatment. Monitoring residual clonal lymphoma cells in the blood and/or BM by means of aberrant immunophenotype- or PCR-based identification of specific fusion gene products may be an alternative tool for evaluating the kinetics of treatment response.^41,42 

The chance of patient survival after relapse from current front-line protocols is poor, although there might be differences according to NHL subtypes. However, prospective controlled clinical trials on relapsed pediatric NHL are rare. One of the more frequently used salvage regimen may be ICE (ifosfamide, carboplatin, etoposide),⁴³ although reports on results in relapsed pediatric patients with NHL are rare. With the combination dexamethasone, etoposide, cisplatin, HD cytarabine, and L-Asp (DECAL), complete remission was achieved in 40% of patients while EFS at 2 years was 33%.⁴⁴ Results were not broken down for NHL subtypes, however. With ALL relapse protocols, survival rates of less than 30% were achieved for children with relapse of LBL.^45,46 Similarly, survival rates of only 10% to 15% were reported for children with relapsed or refractory BL/B-ALL, with therapy approaches consisting of dose-intense chemotherapy followed by autologous or allogeneic hematopoietic stem-cell transplantation (HSCT).^29,45,47 Prospective studies are needed to determine whether newer therapy options such as anti–B-cell–specific monoclonal antibodies have a role in the treatment of these patients. A more favorable survival rate of 69% at 3 years was reported for 41 relapsed ALCL patients.¹⁵ They were treated with courses of CCNU, vinblastine, bleomycin or cytarabine followed by autologous HSCT in some of the patients. An intriguing observation was that some patients with follow-up relapses achieved complete remission with weekly vinblastine 6 mg/m² as a single agent. However, this retrospective study included patients from as far back as the 1970s. For patients with ALCL who failed current front-line protocols, less favorable survival rates were reported.^11,12 Early occurrence conferred the highest risk for failure to salvage therapy.¹⁵ However, even in those high-risk patients, including patients with follow-up relapse after autologous HSCT, long-term remissions were observed after allogeneic HSCT.⁴⁸ Currently, prospective clinical trials are ongoing to test the feasibility and efficacy of new treatment strategies for children with recurring ALCL, including monoclonal anti-CD30 antibodies.

Table 5  summarizes candidates for new treatment options for childhood NHL. Allogeneic HSCT was effective in children with recurring ALCL refractory to chemotherapy. In the BFM series, the only survivors of T-LBL relapse received allogeneic HSCT. Therefore, allogeneic HSCT, as part of front-line treatment, may be an option for improving outcomes for the highest-risk patients with ALCL and T-LBL. Monoclonal antibodies are a new category of treatment options, and rituximab (anti-CD20) is established in the treatment of adult patients with B-NHL. It is not clear, however, that monoclonal antibodies will have similar beneficial effects in children with B-NHL due to differences in the baseline treatment outcome and differences in biological features. The addition of rituximab to CHOP chemotherapy was beneficial for bcl-6–negative but not bcl-6–positive adult patients with DLBCL.⁴⁹ However, most children with B-NHL, in particular BL and DLBCL, are bcl-6 positive.⁵⁰ The effects of rituximab in childhood B-NHL is being investigated in ongoing studies. A new generation of purine nucleoside analogs and purine nucleoside phosphorylase inhibitors will expand the spectrum of chemotherapy, especially for T-cell neoplasms. More specifically targeted future treatment options include disease-specific kinase inhibitors, such as Alk inhibitors, and drugs that interfere with the constitutive activation of the nuclear factor-kappa B (NF-κ B) pathway in distinct subtypes, such as PMLBL.⁵¹ The availability of new drugs to prevent severe mucositis is a new option to ameliorate the predominant severe orointestinal toxicity of chemotherapy, especially for patients with advanced-stage B-NHL.

Acknowledgments

The helpful comments of Ian Magrath, MD, FRCP, FRCPath (Brussels, Belgium), are gratefully acknowledged.