Primary diffuse large B-cell lymphoma of the tonsil

Is a higher radiotherapy dose required?




The purpose was to evaluate the prognostic factors and treatment outcome of Indian patients with primary diffuse large B-cell lymphoma (DLBCL) of the tonsil treated at a single institution.


In all, 121 patients with DLBCL of the tonsil, treated at the Tata Memorial Hospital, Mumbai, India, from January 1990 to December 2002, were included. The median age was 45 years and the majority of patients (68%) were males. Systemic symptoms were present in 12% of patients; 28% presented with stage I and 67% had stage II disease. Treatment consisted of a combination of chemotherapy (CTh) and radiotherapy (RT) for the majority of patients (69.4%). Among those receiving RT, 64% received an RT dose of ≥45 Gy.


After a median follow-up of 62 months, disease-free survival (DFS) and overall survival (OS) were 66.4% and 81.6%, respectively. Significant prognostic factors included: WHO performance score ≥2 (OS: 72.1% vs 95.6%, P = .016), bulky tumors (OS: 68.5% vs 86.9%, P = .001), presence of B-symptoms (OS: 36.7% vs 79.6%, P < .001), and Ann Arbor stage. On multivariate analysis; WHO performance score ≥2 (hazard ratio [HR], 4.27; 95% confidence interval [CI], 1.20–15.12), and B symptoms (HR, 6.27; 95% CI, 2.38–16.48), retained statistical significance. CTh + RT resulted in a significantly better outcome than those treated with CTh alone (OS: 85.7% vs 70.7%, P = .008). The complete response (P = .053), DFS (P = .039), and OS (P = .014) rates were significantly better for patients receiving an RT dose ≥45 Gy.


Tumor bulk, WHO performance score, the presence of B symptoms, and Ann Arbor stage significantly influence outcome. A combined modality treatment, consisting of CTh and RT (with an RT dose of ≥45 Gy), results in a satisfactory outcome in patients with this uncommon neoplasm. Cancer 2007; 110:816–23. © 2007 American Cancer Society.

Malignant lymphoma is primarily a disorder of the lymph nodes; however, 24% to 48% of all non-Hodgkin lymphomas (NHL) may arise from extranodal sites, and there appears to be an increasing incidence of such lymphomas during the past few decades.1 Approximately 10% of patients with NHL present with extranodal disease in the head and neck region.2 Furthermore, more than half of these head and neck lymphomas occur in the Waldeyer ring,2, 3 and 40% to 50% of these arise from the tonsil.4, 5 These lymphomas occur predominantly in elderly males and present as a tonsillar swelling, cervical adenopathy, dysphagia, odynophagia, or with a sore throat. The majority of tonsil lymphomas are of B-cell origin and the most common histological type is diffuse large B-cell lymphoma (DLBCL).6, 7 Most patients present with localized disease: stage I in 12% to 42%, stage II in 36% to 60%, and stage III or IV disease in 10% to 20% or less.6–9 Treatment approaches that have been used include radiotherapy (RT) alone, chemotherapy (CTh) alone, or a combination of both (CTh + RT).5, 10 Most series in the literature have reported results of treating lymphomas of the Waldeyer ring as a group and the resultant diversity of primary site, histology, presentation, therapeutic modality, and outcome makes it difficult to draw meaningful conclusions from these studies. In an attempt to further define the prognostic factors, appropriate management, and treatment outcome for patients with primary DLBCL of the tonsil, we reviewed our experience with this particular extranodal lymphoma.


One hundred and twenty-one patients with DLBCL of the tonsil were treated at the Tata Memorial Hospital from January 1990 to December 2002. All patients were of Indian origin. The age at presentation ranged from 11 to 74 years with a median age of 45 years. There were 82 males and 39 females (male-to-female ratio 2.1:1). The majority of patients (57%) presented with cervical lymphadenopathy. The other common presenting symptoms were growth or ulcer on the tonsil (38%), dysphagia or foreign-body sensation (36%), odynophagia or otalgia (20%), change in voice (5.8%), and presentation with only ‘B’ symptoms (1.6%). Thirty-three (27%) patients had lesions that were clinically or radiologically determined to be over 7 cm in size (bulky tumors). Pathologic specimens of all patients were evaluated by experienced pathologists at our institute and classified according to the REAL classification system.11 All patients included in this study were diagnosed on the basis of histopathology and immunohistochemistry to have DLBCL. The patients were clinically staged according to the Ann Arbor system. Staging investigations included complete blood cell counts, serum biochemistry, bone marrow aspirate and trephine biopsy, chest x-ray, ultrasonography of the abdomen and pelvis, or computed tomography (CT) scan of the thorax, abdomen, and pelvis, and CT scan of the head and neck region. Thirty-four (28%) patients had stage I disease, 81 (67%) had stage II disease, whereas 6 (5%) had stage III/IV disease (stage III, 4; stage IV, 2). Systemic symptoms (B-symptoms) were present in only 15 (12%) patients, consisting of fever (80%), weight loss (60%), and night sweats (15%).

Eighty-four (69.4%) patients were treated with a combination of CTh and RT, whereas the remaining 37 (30.6%) received CTh alone. The distribution of treatment choice by stage is shown in Table 1. The various CTh regimens used were CHOP (cyclophosphamide 750 mg/M2, doxorubicin 50 mg/M2, vincristine 1.4 mg/M2, and prednisolone 100 mg) in 65.3% (n = 79), COP (cyclophosphamide 750 mg/M2, vincristine 1.4 mg/M2, and prednisolone 40 mg) in 24.8% (n = 30), MACOP-B (methotrexate 400 mg/M2, doxorubicin 50 mg/M2, cyclophosphamide 350 mg/M2, vincristine 1.4 mg/M2, prednisolone 75 mg, and bleomycin 10 U/M2) in 8.3% (n = 10), and other CTh regimens in 1.6% (n = 2) patients. The financial status of the patient influenced the choice of chemotherapy. External beam radiation therapy (EBRT) was delivered using 60 cobalt γ-rays or 6MV photons using conventional bilateral parallel opposed portals. The portals included the entire Waldeyer ring and the lymphatic drainage areas (bilateral level Ib to level V neck nodes). The EBRT dose ranged from 30 to 54 Gy with a median dose of 45 Gy. Despite the wide dose range, the use of 2 fractionation schedules predominated; 56% (n = 47) of the 84 patients treated with CTh + RT received an EBRT dose of 45 Gy in 25 fractions, whereas 29% (n = 24) were irradiated to a dose of 40 Gy in 20 fractions.

Table 1. Treatment by Stage
StageTreatment type
Chemotherapy aloneChemo + radiotherapy
I9 (26.5%)25 (73.5%)
II25 (30.9%)56 (69.1%)
III and IV3 (50%)3 (50%)
Total37 (30.6%)84 (69.4%)

Statistical Analysis

Follow-up information was abstracted from patients' hospital records and from an existing database of patients attending the Joint Lymphoma Clinic. Consecutive patients with involvement of the tonsils were considered for this retrospective study based on the following selection criteria: 1) primary symptoms and majority of the tumor bulky localized in 1 or both tonsils; 2) histopathologic examination and immunohistochemistry confirming the diagnosis of DLBCL type of NHL. Complete response (CR) rates, disease-free survival (DFS), and overall survival (OS) were the endpoints reviewed in this study. The definitions of CR, partial response (PR), stable disease (SD), and progressive disease (PD) provided by the WHO criteria for reporting results12 were used. The chi-square (χ2) test was used to compare CR rates. OS was calculated from date of registration to the date of death due to any cause. Univariate analysis for DFS and OS rates were done using the Kaplan-Meier method and prognostic factors were compared using the log-rank test. Multiple-covariate analysis was performed using the stepwise Cox proportional hazards regression model. The hazard ratio (HR) with the 95% confidence interval (CI) was calculated for the treatment groups.


After a median follow-up of 62 months (range, 2–176 months) for surviving patients, 86 patients were alive and without disease, 17 were alive with disease, 14 patients had died due to disease, 2 patients had died as a result of chemotherapy-related febrile neutropenia, 1 elderly gentleman had succumbed to a second cancer (squamous carcinoma of the base tongue) 6 years after radiotherapy, and 1 young man had died in a traffic accident. The complete response rate to therapy was 81.8%. The 10-year DFS and OS were 66.4% and 81.6%, respectively (Fig. 1).

Figure 1.

(A) Disease-free survival. (B) Overall survival (Kaplan-Meier).

Prognostic Factors

Various prognostic factors were analyzed to establish their influence on response rates and the survival of patients with DLBCL of the tonsil. On univariate analysis, patients with lesions that were clinically or radiologically determined to be over 7 cm in size (bulky) had a significantly poorer outcome as compared with those with smaller (nonbulky) tumors. The CR, DFS, and OS rates for these 2 groups were 66.7% vs 87.5% (P = .015), 50.6% vs 72.7% (P = .003), and 68.5% vs 86.9% (P = .013), respectively. Male patients had better CR rates as compared with females (86.6% vs 71.8%, P = .045); however, this did not translate into any difference in DFS or OS (Table 2). Other patient-related factors that had a statistically significant influence on survival were: WHO performance score ≥2 (OS: 72.1% vs 95.6%, P = .016), presence of B-symptoms (OS: 36.7% vs 79.6%, P < .001), and the Ann Arbor stage (Table 2; Fig. 2). The CR (P = .002), DFS (P = .005), and OS (P = .012) rates were 97.1%, 90.7%, and 100% for stage I; 77.8%, 59.5%, and 74% for stage II; and 50%, 0%, and 0% for stage III and IV disease, respectively. When these factors were entered into multivariate analysis using the stepwise logistic regression model to determine independent prognostic variables, a WHO performance score of ≥2 (HR, 4.27; 95% CI, 1.20–15.12; P = .024) and the presence of B symptoms (HR, 6.27; 95% CI, 2.38–16.48; P < .001) retained statistical significance (Table 3).

Figure 2.

Overall survival for prognostic factors (n = 121). (A) Stage. (B) Performance score (WHO). (C) Presence of B symptoms. (D) Size of tumor (>7 cm = bulky).

Table 2. Prognostic Factors for Clinical Outcome
Prognostic factorNo.Complete response rate (CR)P10-year disease-free survival (DFS)P10-year overall survival (OS)P
  1. COP indicates cyclophosphamide, vincristine, and prednisolone; CHOP, cyclophosphamide, doxorubicin, vincristine, and prednisolone.

 ≤30 y2885.7%.3885.0%.10985.0%.857
 >30 y9380.6%61.1%80.2%
WHO performance score
B symptoms
Tumor size
 Bulky, >7 cm3366.7%.01550.6%.00368.5%.001
Ann Arbor stage
 III and IV650%0%0%
 Chemotherapy alone3759.5%<.00135.6%<.00170.7%.008
 Chemotherapy + radiotherapy8491.7%78.4%85.7%
Radiotherapy dose
 <45 Gy3060%.0159.7%.03978.9%.014
 ≥45 Gy5482%91.0%88.2%
Table 3. Multiple-Covariate Cox Regression Analysis for Prognostic Factors
VariableDisease-free survivalOverall survival
  1. CI indicates confidence interval; HR, hazard ratio.

WHO performance score ≥2.1111.810.87–3.79.0244.271.20–15.12
Presence of B symptoms.0402.421.04–5.62<.0016.272.38–16.48
Presence of bulky lesions, ≥7 cm.0122.481.22–5.01.1302.360.85–6.53
Treatment with chemotherapy alone. no radiotherapy<.0017.073.31–15.09.0452.711.02–7.19
Radiotherapy dose <45 Gy.0059.651.98–46.98.0079.891.88–52.02

Treatment Outcome

Ninety-nine of the 121 patients had a CR to treatment resulting in an overall CR rate of 81.8%. At the time of analysis 12 (12.1%) of the patients who had achieved CR had recurred. Eight of these patients had failed at the site of initial disease, 3 patients developed disseminated disease and involvement of the bone marrow, whereas 1 patient had developed a recurrence at a new site altogether.

The 84 (69.4%) patients who were treated with a combination of multiagent chemotherapy and radiation therapy had a significantly better outcome than the 37 (30.6%) patients managed with chemotherapy alone (Fig. 3). The CR, DFS, and OS rates for these 2 treatment groups were: 91.7% vs 59.5% (P < .001), 78.4% vs 35.6% (P < .001), and 85.7% vs 70.7% (P = .008), respectively. The type of therapy used (CTh alone vs CTh + RT) retained statistical significance when entered in the multivariate analysis along with other prognostic factors. The HR for death in the chemotherapy alone group was 2.71 (95% CI, 1.02–7.19). In the subgroup of patients who received combined modality therapy, 54 patients (64%) had received an EBRT dose ≥45 Gy, whereas 30 (36%) had received a dose <45 Gy. The CR (96.3% vs 83.3%, P = .053), DFS (91% vs 59.7%, P = .039), and OS (88.2% vs 78.9%, P = .014) rates were found to be significantly superior for those who had received a dose ≥45 Gy (Table 2). The hazard ratio for death in the subgroup that received a radiotherapy dose of <45 Gy was 9.89 (95% CI, 1.88–52.02) (Table 3). On the other hand, the chemotherapy regimen used did not seem to significantly influence the outcome. The CR (P = .096), DFS (P = .881), and OS (P = .577) rates were 84.8%, 70%, and 85.6% for patients who received CHOP chemotherapy vs 80%, 60.9%, and 78.6% for those who received COP vs 66.7%, 67.5%, and 72.9% for patients who received other chemotherapy regimens, respectively (Table 2). A total of 83 of the 121 patients attained a CR to the initial CTh. Sixty-one (73.5%) of these patients received adjuvant RT after completion of CTh, whereas 26.5% (n = 22) received no further treatment. The DFS was significantly better for patients treated with adjuvant radiotherapy (96.2% vs 54.4%, P < .001); however, there was no difference in the OS (98.2% vs 95%) (Fig. 4).

Figure 3.

Overall survival by treatment type (Kaplan-Meier). (A) Chemotherapy alone (n = 37) vs chemotherapy plus radiotherapy (n = 84). (B) Radiotherapy dose ≥45 Gy (n = 54) vs <45 Gy (n = 30).

Figure 4.

Disease-free survival for patients attaining complete response to chemotherapy: adjuvant radiotherapy (n = 61) vs no radiotherapy (n = 22).


Most of the Waldeyer ring lymphomas reported in the literature are of B-cell origin.6, 8, 9, 13, 14 Furthermore, the vast majority of these lymphomas, ranging from 67% to 96%, have been reported to be of high to intermediate grade.6, 15 The Sheffield Lymphoma Group, in their 30-year experience with extranodal lymphomas of the head and neck region, found the most common histologic subtype to be DLBCL.16 Although T-cell and NK/T-cell lymphomas may make up a sizeable fraction of nasopharyngeal NHL, the most common histologic subtype to involve the tonsils is, by far, DLBCL.14–18 Ezzat et al.14 reported 84% of their patients to have DLBCL, whereas Qin et al.18 found 64% of the tonsil lymphomas in their series to be of the DLBCL subtype. Furthermore, a majority of these patients (60%–70%) present with early stage (stages I and II) disease.6, 7 Gao et al.19 reported 69.6% of their patients to have localized (stage I or II) disease, whereas in the series by Mohammadianpanah et al.17 18% of patients had stage I and 67% had stage II disease.

From January 1990 to December 2002, 148 patients with NHL involving the tonsil were treated at the Tata Memorial Hospital, Mumbai. The vast majority of these patients (81.7%) had the DLBCL histologic subtype. To avoid the confounding influences of the other histologies, we only included the 121 patients with primary DLBCL of the tonsil in this analysis.

In our current study we found a relatively younger cohort of patients as compared with the reported literature. The median age at presentation was 45 years and 23% of our patients were under 30 years of age. The majority of our patients presented with early stage disease, 28% had stage I, and 67% had stage II disease.

Treatment policies for Waldeyer ring lymphomas have included RT with or without CTh for limited stage disease and aggressive CTh with or without RT for advanced stage disease. Authors have reported5-year survival rates of 60% to 80% for patients with early stage disease.8, 9 Qin et al.18 reported a 5-year OS of 84% for stage I-II tonsil NHL, whereas Gao et al.20 reported an OS of 65% for these patients. In a study by Gao et al.19 larger size of the tumor, poor performance status, presence of systemic symptoms, and a higher Ann Arbor stage were indicative of a poor prognosis. Other studies have also found size and stage of disease to be poor prognostic factors.6, 7, 21 In our current series, WHO performance score ≥2, the presence of B-symptoms, tumor bulk >7 cm, and higher Ann Arbor stage were associated with inferior outcome (Table 2). On multivariate analysis, a performance score ≥2 and the presence of ‘B’ symptoms retained their poor prognostic significance (Table 3). The 10-year DFS for patients with stage I or stage II disease was 90.7% and 59.5%, respectively.

The majority of our patients (69.4%) were treated with a combination of CTh and RT, resulting in a DFS and OS of 78.4% and 85.6%, respectively, which was superior to the group receiving CTh alone (DFS: 35.6%, P < .001; OS: 70.7%, P = .008). Furthermore, even patients who attained CR after chemotherapy benefited significantly from the addition of adjuvant RT (96.2% vs 54.4%, P < .001). Similarly, Gao et al.20 also reported a significantly improved DFS (P = .046) with a combined modality treatment in their series of patients. Furthermore, Ezzat et al.14 demonstrated a significantly better event-free survival for the combination of CTh + RT. In a previously reported large randomized trial of 316 patients with stage I disease of Waldeyer ring, patients were randomized to receive RT alone, CTh alone, or a combination of both. Failure-free survival (FFS) and OS at 5 years of follow-up were significantly superior in the group receiving combined modality therapy compared with RT or CTh alone (FFS, 83% vs 48% vs 45%; OS, 90% vs 56% vs 58%).5 Similar improvements in FFS and OS have been reported by the Eastern Cooperative Oncology Group (ECOG) with the addition of RT after full-course CTh in limited-stage aggressive NHL.10

The subset analysis of our patients treated with a combination of CTh and RT revealed that patients receiving an RT dose of ≥45 Gy had a statistically significant improvement in CR (P = .01), DFS (P = .039), and OS (P = .014) rates over those treated with a lower dose. Although there has been no report to date on the most appropriate RT dose for primary NHL of the tonsil, there are a few studies that have suggested a benefit with the use of higher RT doses in patients with nasal and nasopharyngeal NHL.22–24 Isobe et al.23 reported a local failure rate of 67% for patients with nasal-type NK/T-cell lymphoma treated with an RT dose of less than 50 Gy as compared with 27% (P = .038) for 50 Gy or more. In another study, Koom et al.22 used a median dose of 45 Gy and reported a local failure rate of 64% for an RT dose less than 45 Gy vs 38% (P = .02) for 45 Gy or higher. Furthermore, in a previously reported series on NHL of the nasopharynx from our institute, we found a significant improvement in the CR (72% vs 21%, P = .00001), DFS (68% vs 23.3%, P = .00001), and OS (68.3% vs 31%, P = .00001) rates using an RT dose of ≥45 Gy.24 Similarly, in this study we observed that the escalation of RT doses to 45 Gy and above may result in an improved locoregional control and OS for patients with DLBCL of the tonsil.

In conclusion, we note from this relatively large single-institution study that primary DLBCL of the tonsil is best managed with a combination of CTh and RT. The CTh should consist of a CHOP or CHOP-like regimen, followed by involved field radiotherapy (IFRT) to a dose of ≥45 Gy with conventional fractionation. We also report that DLBCL of the tonsil involves a larger proportion of young patients (median age of 45 years) in the Indian subcontinent as compared with the rest of the world, where it usually presents in the sixth decade of life. Considering the young age and good prognosis of these patients, the possibility of using 3D-conformal radiotherapy or intensity-modulated radiotherapy to limit the late toxicities of treatment must be kept in mind if radiotherapy doses are to be escalated.