Radiotherapy for Non-Hodgkin’s lymphoma: still standard practice and not an outdated treatment option

FL represents the most common indolent NHL in the USA and Europe, accounting for an estimated 20-30 % of all NHL [9]. The genetic hallmark of FL, the translocation t(14;18)(q32;q21), results in the constitutive overexpression of the bcl 2 protein, impairing the normal germinal centre apoptotic programme [10].

Initial workup should include an adequate staging (including bone marrow examination) to ensure precise definition of the extent of the disease. Recently, the use of PET imaging - in clinical trials and clinical practice - has been recommended as the preferred option for staging and remission assessment of all FDG-avid NHLs, including follicular lymphoma by an International Consensus group [11]. In FL, the use of positron emission tomography (PET) scanning is particularly useful in patients presenting with early stage disease to confirm localized disease before initiating RT [7].

RT remains the current standard of care for patients with low-grade (grade 1-2), early stage FL (non-bulky stage I-II), which represent 10-15 % of all FL [7]. Other options in selected cases include observation, immunotherapy (rituximab) and chemotherapy. FL grade 3b is regarded as an aggressive lymphoma and treated like DLBCL.

Despite NCCN and ESMO guidelines endorsing RT as the preferred initial management [6, 7], RT use in patients with early-stage low-grade FL continues to decline. In the largest retrospective cohort study reported so far (35,961 patients), use of RT dropped from 37 % in 1999 to 24 % in 2012 (p < 0.0001), correlating with an increase of observation from 34 % to 44 % (p < 0.0001) over the same period [12]. In this paper, the use of RT was associated with a significant 12 % absolute improvement in the 5-year and a 14 % absolute improvement in the10-year overal survival (OS) rate. An identical 14 % absolute 10-year OS benefit in patients receiving RT was noted in an older population-based analysis from 1973 through 2004 from the SEER data set [13]. These data suggest that even within the context of modern systemic therapy, RT significantly contributes to survival for patients with early-stage low-grade FL and should remain standard practice.

These results have been challenged by a retrospective multicenter observational study (National LymphoCare Study). In this series, only 111 out of 474 patients (23.4 %) with stage I FL received RT as single modality treatment [14]. Excellent outcomes with different treatment approaches were achieved, suggestive of a potential PFS benefit for systemic therapy plus RT or systemic therapy plus rituximab over RT alone, but without any OS difference [15]. However, several flaws make it difficult to draw any conclusions on treatment choices from this retrospective analysis. The National LymphoCare study was not designed to compare different therapeutic options in early-stage FL, and selection of treatment was not based on predefined entry criteria [10, 16]. Furthermore, a potential conflict of interest might arise from the partial funding of this study by pharmaceutic companies (Genentech and Biogen).

The dogma of early-stage, indolent lymphoma as a pure localized disease (amenable to cure with RT) has to be challenged, as nearly half of patients with early-stage FL will relapse within ten years, almost exclusively in distant non-irradiated areas [17]. It should be noted that with the increasing integration of PET-CT into initial management, improved staging accuracy may translate into improved disease control and survival rates for early-stage FL, by excluding patients with occult, distant disease [18]. For the latter, lymphoma remains the leading cause of death [19].

In this context, rituximab represents an ideal systemic drug with a low toxicity profile, aiming to eliminate circulating lymphoma cells as well as distant subclinical involvement, and would complement nicely RT, which targets macroscopic disease only. Furthermore, rituximab proved as a radiosensitizer in vitro, enhancing radiation-induced apoptosis and cell growth delay [20]. Prospective studies are therefore urgently needed, to answer the question if single-agent rituximab, rituximab combined with involved-field radiotherapy (IFRT) (sequential and/or concomitant), or even radioimmunotherapy offers any advantage over IFRT alone [21]. So far, we are only aware of a single prospective, multicentric phase II-study (MIR trial, 85 patients) in the field. The treatment included a first sequence of rituximab (4 weekly cycles; 375 mg/m² body surface), a four week treatment gap with a restaging CT / planning CT of the involved region in week seven, and followed by another block of rituximab (4 weekly cycles) concurrently with an IFRT of 40 Gy for macroscopic tumor or 30 Gy in case of a complete remission after rituximab induction [22]. Preliminary results were presented in abstract form in 2012; the authors stated that this combined treatment was well tolerated; 2-year PFS was similar to historical data with large field RT without the accompanying toxicity and was superior to historical data of IFRT only [23]. Mature results are pending, however.

In current practice, we consider that RT should not be omitted outside clinical trials. For patients with low-grade, early-stage FL, general dose guidelines are 24-30 Gy and 24-36 Gy according to NCCN and ESMO recommendations [6, 7].

The British National Lymphoma Investigation (BNLI) randomized study showed that 24 Gy in 12 fractions was as effective for local control as 40 Gy in 20 fractions in terms of overall response and within field progression [24]. This attractive 24 Gy dose recommendation has not been universally implemented so far, as the trial presents some shortcomings, including a greater heterogeneity both in terms of histological diagnosis (FL grade 1-2 making only 59 % of the 289 patients with indolent lymphoma) and in the evaluation of treatment response (no consequent 3D imaging during follow-up).

Very low-dose radiotherapy (4 Gy) has been evaluated in a Dutch phase II study of 109 patients with recurrent indolent lymphoma (98 cases with FL), with an overall response rate of 92 %. The subgroup of 67 patients with complete response (CR) showed a median time to progression of 25 months and a median time to local progression of 42 months [25].

To address prospectively the issue of very low-dose RT, the FORT trial randomized 614 sites (in 548 patients) to receive either 24 Gy or 4 Gy. Results clearly showed that 4 Gy was inferior to 24 Gy in terms of time to local progression (HR of 3.4; 95 % CI: 2.09-5.55; p < 0.001) [26]. The FORT trial has been criticized not only for presenting similar flares to the previous dosage study (40 Gy vs 24 Gy) as mentioned above, but also for not stratifying patients and assessing local response according to the initial size of lesions; the non-inferiority of the 4 Gy regimen might not be excluded for smaller FL lesions [27]. To illustrate their point further, the same investigators highlight the encouraging results achieved with orbital lymphoma, where the 2x2 Gy schedule appears to be quite a valuable option, with two retrospective studies reporting a local PFS of 100 % at 2 years [28, 29].

Taken together, 24 Gy can be considered as the lowest standard dose for curative treatment of most patients with low grade, early-stage FL, until a lower dose (that is in between 4 Gy and 24 Gy) is explored properly [16].

For patients with asymptomatic, low-tumor burden stage III-IV disease, FL is considered as a chronic disease; a “watch and wait” strategy is usually recommended. The natural course of disease is characterized by spontaneous regression in 10-20 % of cases [7]. Systemic treatment (rituximab ± chemotherapy) should only be initiated when a patient presents with indications for treatment (Groupe d’Etude des Lymphomes Folliculaires (GELF) criteria) [6]. Future research protocols will also integrate second-generation anti-CD20 antibodies (ofatumumab and obinutuzumab), lenalidomide, mTOR inhibitors (temsirolimus and everolimus) Bcl-2 or Bruton’s tyrosine kinase (BTK) inhibitors.

Local palliation with low-dose RT (2x2 Gy) may be used in selected cases. This treatment regimen provides effective symptomatic relief for tumor bulk of all sizes, with an overall response rate of 81 % [30]. Should the patient progress again locally without significant systemic progression of disease, local treatment with 2x2 Gy can be repeated. This low-dose RT regimen remains an excellent option in the palliative and relapse setting, especially for patients with poor performance status, and should be practiced more frequently [16].

The most common type of aggressive B cell lymphoma is DLBCL, accounting for approximately 30 % of NHLs diagnosed annually [31]. Gene expression profiling studies have revealed significant heterogeneity within DLBCL. Incoroporation of this information into treatment algorithms awaits further investigation [6].

PMBCL is relatively uncommon, comprising around 3 % of all NHL, and up to 10 % of DLBCL. It most often presents in young adults, with a median age of 35 [53]. Gene expression profiling is distinct from other types of DLBCL [54]. In common with nodular sclerosing Hodgkin lymphoma (HL) in the mediastinum, it is thought to originate from transformed thymic B-cells, and there is an intermediate entity which lies between these two types, currently termed mediastinal gray-zone lymphoma [55].

In the absence of randomized trials, optimal first-line treatment for patients with PBMCL is more controversial than other types of NHL, especially in the rituximab era [6]. The recently completed CALGB study (NCT00118209) comparing R-CHOP versus DA-R-EPOCH will provide useful information wether any particular chemotherapy regimen is superior.

The use of consolidation RT for PMBCL has been an historical standard of care, based upon poor results following chemotherapy alone prior to rituximab; furthermore the very poor outcomes for patients who develop recurrent disease highlighted the need to maximize cures at first attempt [56]. Considering the potential long-term toxicities of mediastinal RT, some investigators have elected to omit consolidation RT [56, 57]. The ongoing IELSG-37 study (NCT01599559) is currently evaluating the important question whether RT can be omitted in PMBCL patients, who have become “PET-negative” at the end of initial rituximab-chemotherapy regimens.

PCNSL accounts for approximatively 3 % of all primary CNS tumors. It is an aggressive form of NHL that develops within the brain, spinal cord, eye, or leptomeninges, without evidence of systemic involvement [6]. Ninety percent of non-HIV-associated PCNSL cases are of the diffuse large B-cell type, characterized by lymphoid clustering around small cerebral vessels [58]. The brain parenchyma is involved in more than 90 % of cases. Leptomeningeal involvement may occur in up to 30 % of patients. Ocular involvement may develop independently in ten to 20 % of patients.

The CHOP regimen induces responses of brief duration in patients with primary CNS lymphoma. This inefficacy is probably because the metabolite of cyclophosphamide, phosphoramide mustard, and doxorubicin are not able to cross the blood–brain barrier [59]. Rituximab has also poor CNS penetration because of its large size. Methotrexate (MTX) is currently the most effective single agent in PCNSL. Combining antimetabolites such as MTX and cytarabine (ara-C) constitute the backbone of most anti-PCNSL regimens with proven efficacy in prospective trials [60]. In a phase II study conducted by the International Extranodal Lymphoma Study Group (IELSG), 79 patients were assigned to 4 courses of MTX (3.5 g/m²) alone or combined with ara-C (4 doses of 2 g/m²), in both arms followed by whole-brain irradiation (WBRT). The addition of ara-C resulted in significantly improved response (CRR: 46 % vs 18 %; p = 0.006) and better outcome (3-year OS: 46 % vs 32 %; p = 0.07) compared with high-dose MTX alone, with manageable hematologic toxicity and uncommon nonhematologic side effects [61]. Another study highlighted the importance of ara-C dose, suggesting that 4 doses of 2 g/m² is an appropriate choice [62]. Although it was not addressed in a phase 3 trial, the MTX-ara-C combination is currently the most commonly used treatment regimen outside clinical trials.

Methotrexate can also be administered as high-dosed monotherapy. The superiority of high-dose methotrexate over RT alone has never been demonstrated in a randomized trial [63]. Although not formally compared in a randomised trial, results of several studies suggest that the combination of high-dose methotrexate with RT is better than RT alone, in terms of increasing the proportion of long-term survivors (5-year survival 20–50 %) and OS by two to four times (median 30–72 months) [63]. There has been so far only one large prospective phase III study comparing consolidation WBRT to observation alone after high-dose methotrexate [64]. The unmet primary endpoint for non-inferiority and the high rate of protocol violations (only 318 out of 551 patients treated per protocol) prevent unfortunately any reliable conclusions to be drawn from this study [65].

In current practice, chemotherapy will usually be followed by consolidation RT (or high-dose chemotherapy) as initial treatment to maximize response and improve outcome; WBRT up to 24-36 Gy is recommended, without a boost, according to NCCN guidelines [6]. However, it has to be stressed that consolidation WBRT is associated with high rates of neurotoxicity, especially in patients > 60 years of age, and therefore often omitted. Several small retrospective series suggest that some elderly patients in CR after primary chemotherapy could be watchfully observed without OS impairment. To delay WBRT until relapse appears to be an acceptable strategy considering the increased risk of disabling neurotoxicity in these patients [60].

To reduce neurotoxicity, investigators explored reduced-dose WBRT (23.4 Gy in 13 fractions) for those patients achieving CR after induction chemotherapy with rituximab, methotrexate, procarbazine, and vincristine (R-MPV). Consolidation cytarabine was given after RT [66]. This phase II study was associated with a 2-year PFS rate of 77 %, median PFS of 7.7 years, and 3-year OS rate of 87 %. This R-MPV regimen with or without reduced-dose WBRT is currently tested by the Radiation Therapy Oncology Group (RTOG 1114 trial).

Building on the same induction immunochemotherapy (R-MPV regimen), consolidation WBRT can be replaced by a novel consolidation high-dose chemotherapy with thiotepa, busulfan and cyclophosphamide (TBC regimen) and autologous stem cell transplantation (HDC-ASCT). This strategy has been tested in a prospective, single-arm, phase II study (32 patients). Following R-MPV, objective response rate was 97 %, and 26 (81 %) patients proceeded with HDC-ASCT. There were 3 treatment-related deaths. Among all patients (N = 32), with a median follow-up of survivors of 45 months, 5-year overall survival (OS) and progression-free survival (PFS) were 81 % and 79 %, respectively. There was no evidence of neurotoxicity thus far. As stated by the authors, this intensive treatment should still be considered experimental. Moreover, the R-MPV regimen has not been formally tested alone, without a consolidation strategy such as WBRT or HDC-ASCT [67].

Outside clinical trials, outcome of CNS lymphoma remains globally dismal. About one third of patients with primary CNS lymphoma will present with disease that is refractory to first-line treatment. Even for patients achieving initial CR, about half of them will relapse. For refractory or recurrent disease, there are few treatment options available, and no established standard of care. For patients who previously achieved a long-lasting response to high-dose methotrextate, retained chemosensitivity might be assumed, and a rechallenge with methotrexate attempted. Conventional chemotherapy might be proposed as well. For selected patients, specific salvage therapies might be explored, including high-dose chemotherapy and ASCT [68], or stereotactic radiosurgery [69].

Historically, RT has been associated with significant risks of late toxicities and second malignancies in long-term survivors; relevant data are mostly derived from retrospective HL series [70‐72].

The French-United Kingdom study of 4122 5-year survivors of various childhood cancers between 1942 and 1986 (median follow-up: 26 years) found that the average dose of radiation to the heart was linearly associated with the risk of cardiac mortality, with the incidence increasing by 60 % for every 1-Gy increase in mediastinal radiation dose [73, 74]. In a similar north-american cohort study restricted to female participants who had received chest irradiation for their childhood cancer diagnosis (RT treatment between 1970 and 1986), the cumulative incidence of breast cancer by age 50 years was 35 % among HL survivors [75]. At this point, it has to be stressed that these late-toxicity figures apply to patients treated for childhood cancers with ancient techniques, and cannot be directly extrapolated to mostly adult patients treated in these days with modern RT.

Different strategies have been implemented to reduce RT toxicities. As normal tissue complication rate is a function of dose and volume, RT therapeutic ratio should improve both with lower treatment doses and smaller target volumes. For NHL, there is now increasing evidence that tradional doses were higher than necessary for disease control; some investigators recommend now no more than 30 Gy for aggressive NHL or 24 Gy for indolent lymphomas [76].

Regarding target volumes for aggressive NHL, optimal imaging allowed to evolve from IFRT (encompassing the pre-chemotherapy involved node chains, based on anatomical landmarks) to involved site RT (ISRT) (target volumes reduced to cover the pre-chemotherapy involved nodes only) [3]. Such a strategy is based on the assumption that chemotherapy already eradicated adjacent or regional microscopic disease and ISRT targets the identifiable pre-chemotherapy disease. Clinical judgment in conjunction with the best available imaging is used to contour a clinical target volume (CTV) that will accommodate the uncertainties in defining the prechemotherapy gross tumor volume (GTV) in each individual case. In the situation where pre-chemotherapy imaging cannot be fused with the post-chemotherapy planning CT scan, allowances should be made for the uncertainty of the contouring and differences in positioning by including a larger volume in the CTV [3].

The proposed reduced late toxicity with involved-site or involved-node RT (INRT) is yet unproven; in the absence of recent long-term toxicity data, dose–volume metrics of organs at risk (OAR) will provide a surrogate measure of toxicity risk. In an Australian study comparing IFRT with INRT using conventional technique (parallel-opposed anterior and posterior photon beams), INRT allowed to reduce the proportion of tissue receiving higher RT doses (V95%) by a factor of 1.9. Regarding major organs at risk, the greatest benefit was seen when analyzing the dose received by 50 % of these organs (D50), which was reduced by half for the heart and the lungs. Keeping now the same INRT target volume and comparing between conventional technique and volumetic modulated arc therapy (VMAT), VMAT resulted in further relative reductions in cardiac dose, but without improvement in coronary artery dose-volume metrics, and at the expense of increased low-dose exposure to lungs [77].

A larger retrospective study (150 patients, 73 % with HL and 27 % with NHL) evaluated recently the risk of radiation pneumonitis (RP) after mediastinal RT with intensity modulated radiotherapy (IMRT) [78]. The overall incidence of any RP (RTOG grades 1-3) was 14 % among the entire group. However, patients who received salvage chemotherapy or transplantation for relapsed or refractory disease were at greater risk: the incidence of RP among those patients was 25 % versus 10 % among those who received consolidative RT only for newly diagnosed disease. In both groups, half patients developing pneumonitis required a steroid course (RTOG grade 3). Interestingly, disease bulk, history of smoking, pre-RT pulmonary function test values, and history of bleomycin toxicity did not predict RP. Regarding dosimetric factors, while the V20 (irradiated volume receiving ≥ 20 Gy) predicted RP risk, the volume of lung receiving lower doses of radiation, especially V5 (irradiated volume receiving ≥ 5 Gy), was the most powerful predictor of RP. If IMRT allows for a better conformality of the higher doses of radiation, these results are achieved at the costs of delivering a low-dose bath to large volumes of lung, leading obviously to a clinically meaningful lung injury. The risk of RP approaches 35 %, when >55 % of the total lung receives 5 Gy in the treatment of HL or NHL with IMRT.

Different IMRT and VMAT refinements have been recently described, to optimize dose delivery and better spare OAR. Traditional IMRT beam arrangements involving 9 fixed and equally distributed beams produce a low-dose bath to the all surrounding critical organs; RT planning should therefore not be automated, and beam angles should be carefully selected on an individual basis, leading for example to a 5-beam anterior-posterior weighted “butterfly” coplanar beam arrangement [79]. A single arc VMAT plan can also be replaced with a more sophisticated three-arcs VMAT (“butterfly” VMAT), with two anterior and posterior coplanar arcs of 60°, complemented with 1 no-coplanar arc of 60° [80].

Another possibility to further reduce the target volume and thus the irradiated volume is the use of gating techniques such as deep-inspiration breath-hold. Combined with IMRT, it can greatly reduce radiation exposure of the coronary arteries, heart, and lungs in patients with mediastinal HL [81]. Despite the obvious advantage to reduce side effects related to the heart and the coronary arteries, an open question is the impact of intensity modulation techniques on the incidence of second malignancies. While the application of VMAT techniques can result in better dose spearing of the thyroid gland, the heart and the coronary ostia, deep-inspiration VMAT was actually shown to be inferior to classical parallel opposed treatment techniques with regard to second cancer induction; VMAT can increase for example the cumulative absolute risk of breast cancer induction by 100 %. VMAT should therefore be cautiously implemented in clinical practice [82].

Last, proton therapy may offer significant and clinically relevant advantages to spare important OAR’s, and has been endorsed by NCCN as an appropriate treatment modality for NHL [6]. While several published studies have evaluated the use of proton therapy for HL, data regarding patients treated with proton therapy for NHL are scarce. Early outcomes appear favorable, but longer follow-up is needed before drawing any definitive conclusions [83].