International Journal of

Radiation Oncology

biology physics

www.redjournal.org

Clinical Investigation: Lymphoma

Radiation Therapy Administration and Survival inStage I/II Extranodal Marginal Zone B-Cell Lymphomaof Mucosa-Associated Lymphoid TissueAdam J. Olszewski, MD, and Amrita Desai, MD, MPH

The Cancer Center at Memorial Hospital of Rhode Island, Pawtucket, Rhode Island

Received Aug 5, 2013, and in revised form Nov 14, 2013. Accepted for publication Nov 14, 2013.

Summary

In this analysis of 7774early-stage mucosa-associ-ated lymphoid tissue lym-phoma cases from theSurveillance, Epidemiology,and End Results database,administration of radiationtherapy as part of initialtreatment was associatedwith excellent survival inmost sites of origin. The riskof lymphoma-related deathwas negligibly low with ra-diation therapy in cutaneousand ocular lymphomas. Sig-nificant disparities in the useof radiation therapy existwith regard to patient age,sex, race, and socioeconomicstatus.

Reprint requests to: Dr Adam J. Olszew

Medicine, Alpert Medical School of Brown Un

Pawtucket, RI 02860. Tel: (401) 429-3151; E

brown.edu

This study used the SEER database. The inte

these data are the sole responsibility of t

Int J Radiation Oncol Biol Phys, Vol. 88, No. 3

0360-3016/$ - see front matter � 2014 Elsevie

http://dx.doi.org/10.1016/j.ijrobp.2013.11.225

Purpose: To determine the factors associated with the use of radiation therapy and associatedsurvival outcomes in early-stage marginal zone lymphoma of the mucosa-associated lymphoidtissue (MALT).Methods and Materials: We extracted data on adult patients with stage I/II MALT lymphomadiagnoses between 1998 and 2010 recorded in the Surveillance, Epidemiology, and End Results(SEER) database. We studied factors associated with radiation therapy administration in a logis-tic regression model and described the cumulative incidence of lymphoma-related death (LRD)according to receipt of the treatment. The association of radiation therapy with survival wasexplored in multivariate models with adjustment for immortal time bias.Results: Of the 7774 identified patients, 36% received radiation therapy as part of the initialcourse of treatment. Older patients; black or Hispanic men; white, Hispanic, and black women;and socioeconomically disadvantaged and underinsured patients had a significantly lowerchance of receiving radiation therapy. Radiation therapy administration was associated with alower chance of LRD in most sites. In cutaneous, ocular, and salivary MALT lymphomas, the5-year estimate of LRD after radiation therapy was 0%. The association of radiation therapywith overall survival in different lymphoma sites was heterogeneous, and statistically significantin cutaneous (hazard ratio 0.45, PZ.009) and ocular (hazard ratio 0.47, P<.0001) locations aftermultivariate adjustment.Conclusions: Demographic factors are associated with the use of radiation therapy in MALTlymphoma. Clinicians should be sensitive to those disparities because the administration of ra-diation therapy may be associated with improved survival, particularly in cutaneous and ocularlymphomas. � 2014 Elsevier Inc.

ski, MD, Department of

iversity, 111 Brewster St,

-mail: adam_olszewski@

rpretation and reporting of

he authors. The authors

acknowledge the efforts of the Applied Research Program, National

Cancer Institute; Information Management Services Inc; and the SEER

program tumor registries in the creation and maintenance of the SEER

database as a research resource.

Conflict of interest: none.

, pp. 642e649, 2014

r Inc. All rights reserved.

Volume 88 � Number 3 � 2014 Radiation therapy in MALT lymphoma 643

Introduction

Extranodal marginal zone lymphomas of mucosa-associatedlymphoid tissue (MALT) are indolent B-cell malignancies ac-counting for about 4% of non-Hodgkin lymphoma cases in theUnited States (1). These tumors arise from marginal zone lym-phocytes residing in the mucosal membranes of different organsand typically present at an early stage, remain confined to theprimary area, and have low cause-specific mortality rate (2).Multiple studies corroborate the role of chronic stimulation byinfection or autoimmune disease in the pathogenesis of MALTlymphomas. Although sustained remissions have been achievedwith eradication of Helicobacter pylori infection in the majorityof gastric cases, studies have failed to consistently show similarresponses in other MALT lymphoma sites (3). External beamradiation therapy, excision, or chemotherapy constitute effica-cious treatment options, providing over 88% relative survival at 5years (4). Single-institution series report that the administrationof radiation therapy is associated with excellent progression-freesurvival, but it leads to significant site-specific toxicities (cata-racts, xerostomia, secondary malignancies) (5-7). Currently, ra-diation therapy is the most common treatment for patients withlocalized MALT lymphoma, based on retrospective data. Withthe advent of efficacious, relatively nontoxic systemic treatment,some authors have suggested excision alone or chemo-immunotherapy as the initial strategy (8, 9).

It remains unclear whether radiation therapy in MALT lym-phomas provides principally local control or whether it also has acurative potential, affecting patients’ survival. Inasmuch as norandomized studies have addressed the issue, we conducted anobservational analysis based on a population-derived cohort. Ourfirst objective was to identify factors associated with the use ofradiation therapy in stage IE/IIE MALT lymphomas of the com-mon anatomic sites. The second objective was to describe survivaloutcomes in patients who received radiation therapy as part oftheir initial treatment strategy.

Methods and Materials

Data source and cohort selection

This study was performed with the use of public, deidentifieddata and was not subject to additional human protection over-sight according to federal regulations. The procedures followedwere in accordance with the ethical standards of the institutionalcommittee on human experimentation and with the HelsinkiDeclaration. The Surveillance, Epidemiology, and End Results(SEER) database curated by the National Cancer Institute in-tegrates records from 18 cancer registries in the United States,currently representing 28% of the population. SEER mandates a98% case ascertainment rate and conducts continuous qualityassurance programs. The records include information on patientdemographics, socioeconomic factors, histology, clinical stageand primary site of the malignancy, cancer-directed treatment,and survival outcomes up to December 31, 2010, in the availableupdated submission (10). The completeness of vital status isachieved by active local follow-up and by linkage to nationaldeath registries. Information on treatment is limited to surgeryand radiation therapy administered as part of the first course oftherapy, defined as the initial treatment plan delineated in the

medical record. Tumor recurrences or treatments used at the timeof progression are not recorded. A decision to pursue watchfulwaiting or refusal of treatment is recorded as no initial therapy.The dataset does not contain data on the dose, field, fractionationscheme, and intent (curative or palliative) of radiation or on theuse of chemotherapy. There is also no information on response totreatment, toxicities, or duration of remission.

We extracted data on all adult cases of MALT lymphomadiagnosed between 1998 and 2010 (NZ10,446) using the Inter-national Lymphoma Epidemiology Consortium classification (11).This classification, based on the International Classification ofDiseases for Oncology, third edition (ICD-O-3) histology codes,facilitates accurate and reproducible studies of lymphomas basedon large registries without central histopathologic review. Tomaximize the specificity of diagnoses, we restricted our study tocases with definitive histopathologic confirmation, removing 38cases. We also excluded cases identified through autopsy or deathcertificate (nZ4), cases of stage III or IV lymphoma (nZ2571)cases with no recorded survival time (nZ53), and duplicate re-cords (nZ6).

We assigned race/ethnicity using a combination of SEER racevariable and the Hispanic identification algorithm developed bythe North American Association of Central Cancer Registries. Weapproximated the socioeconomic status by quartiles of the per-centage of persons living below poverty level, using census-derived data for patients’ county of residence.

The primary site of MALT lymphoma was determined by useof the ICD-O-3 topography codes. Categories were created for allsites with >200 cases and were defined as follows: skin (codesC44.0-C44.9, excluding C44.1), ocular (C69.0-C69.9, C44.1),salivary glands (C0.79-C0.89), thyroid gland (C73.9), other headand neck mucosa (C00.0-C06.9, C09.0-C14.8, C30.0-C32.9),breast (C50.0-C50.9), lung (C34.0-C34.9), stomach (C16.0-C16.9), and small/large bowel (C17.0-C20.9). Excisional surgerywas defined by the use of site-specific codes: skin, ocular, salivarygland, head-and-neck mucosa and stomach (excisional biopsy),lung, thyroid (at least lobectomy), breast (at least lumpectomy),and intestine (at least polypectomy).

Statistical analysis

We studied factors associated with the use of radiation therapy in amultivariate logistic model. Age was categorized into groups 20-49, 50-59, 60-69, 70-79, and �80 years. Local control, responserate, and recurrence-free survival were not accessible from theSEER database. Therefore, we used cumulative incidence oflymphoma-related death (LRD) as the primary descriptive survivalendpoint, calculated by the Cronin-Feuer method, which does notrely on death certificate records (12). LRD describes excessmortality related to cancer and its treatment, compared with ex-pected survival derived from life tables stratified by age, sex, race,and calendar year. Cumulative incidence of LRD is more infor-mative to clinicians and patients than net survival because it ac-counts for the differential risk of death from causes unrelated tomalignancy (13). Indeed, the majority of patients with MALTlymphoma die of such “competing” causes, and this risk increaseswith advancing age.

To explore the varying association of radiation therapy withsurvival in different MALT lymphomas while adjusting foravailable confounders, we used multivariate Cox models foroverall survival in each anatomic MALT site. Additionally, we

Table 1 Demographic and clinical characteristics of patients with stage IE/IIE MALT lymphoma according to anatomic site of origin

Variable Skin OcularSalivaryglands

Head/neckmucosa

Thyroidgland Lung Breast Stomach Bowel

Othersites Total

No. of cases 604 1111 699 312 205 576 241 2849 688 489 7774% 7.8 14.3 9.0 4.0 2.6 7.4 3.1 36.7 8.9 6.3 100AgeMedian 57 64 63 67 63 68 69 68 67 68 66IQR (45-70) (51-75) (50-73) (54-76) (51-73) (59-76) (58-77) (57-78) (57-77) (57-78) (54-77)

Sex, %Women 42.9 55 73.5 57.1 68.8 58.3 95.4 49.9 52.2 58.3 55.7Men 57.1 45 26.5 42.9 31.2 41.7 4.6 50.1 47.8 41.7 44.3Male/female ratio 1.3 0.8 0.4 0.8 0.5 0.7 0.05 1.0 0.9 0.7 0.8

Race/ethnicity, %White non-Hispanic 81.2 69.0 68.2 66.2 72.1 79.9 79.4 71.9 69.0 75.6 72.4White Hispanic 9.6 10.0 16.5 14.3 11.3 6.8 8.8 10.4 10.0 9.5 10.6Black 3.2 7.7 6.2 10.1 2.0 8.8 8.8 8.5 6.6 5.4 7.3Asian/other 6.0 13.4 9.1 9.4 14.7 4.6 2.9 9.2 14.3 9.5 9.6

United States region, %West 68.4 60.4 56.2 60.3 54.6 45.5 56.0 54.3 65.1 62.8 57.6East 16.2 16.6 14.0 15.7 10.7 23.1 19.1 16.3 13.4 17.8 16.4South 8.6 11.8 17.3 13.8 20.5 19.6 11.2 19.0 14.0 10.6 15.7Midwest 6.8 11.3 12.4 10.3 14.1 11.8 13.7 10.4 7.6 8.8 10.4

Geographic area, %Metropolitan 93.9 90.0 89.4 92.3 79.0 88.0 90.0 89.3 89.8 90.0 89.6Urban 5.6 9.5 9.1 6.4 17.6 10.6 8.3 9.4 9.0 9.2 9.2Rural includingAlaska

0.5 0.5 1.6 1.3 3.4 1.4 1.7 1.3 1.2 0.8 1.2

Ann Arbor stage, %IE 96.5 92.7 75.4 73.4 73.7 81.4 86.7 90.0 79.1 85.3 86.5IIE 3.5 7.3 24.6 26.6 26.3 18.6 13.3 10.0 20.9 14.7 13.5

B symptoms present, % 2.3 2.8 4.9 5.4 5.4 8.0 2.5 11.3 10.5 7.2 7.6Surgical excision, % 37.9 33.1 69.1 28.2 82.4 17.9 44.0 5.7 56.5 * *

Radiation therapy, % 50.5 67.2 43.2 47.5 47.5 10.6 41.4 28.0 11.4 31.1 35.9Radiation therapy afterexcision, %

45.0 63.0 40.8 36.4 47.3 4.9 39.6 15.5 4.4 * *

Follow-up of survivingpatients

Median, mo 46 58 61 60 48 48 55 58 48 53 55IQR, mo (23-79) (29-94) (31-93) (27-95) (22-87) (20-83) (26-83) (28-95) (24-90) (20-86) (26-91)

Abbreviations: CI Z confidence interval; IQR Z interquartile range; MALT Z mucosa-associated lymphoid tissue.

* Not calculated because of lack of a uniform definition of excisional surgery.

Olszewski and Desai International Journal of Radiation Oncology � Biology � Physics644

fitted analogous competing risk models for LRD based on deathcertificate attributions, with LRD defined as death ascribed to theindex malignancy according to SEER algorithms, or any eventwith lymphoma listed as the cause of death. Landmark analysiswas used to eliminate immortal time bias, excluding patients withsurvival shorter than 12 months, who may not have survived longenough to receive the intended radiation therapy (14). The pro-portional hazard assumption was tested with Schoenfeld residuals.For lymphoma sites with <10 events per variable, we usedforward-stepwise variable selection, thus limiting the number ofvariables in the model. Cases with missing covariate values wereexcluded from the models. All statistical tests are reported with 2-tailed P values and 95% confidence intervals (95% CI), at an alevel of 0.05 or lower, using SEER*Stat, version 8.0.4 (http://www.seer.cancer.gov/seerstat), Stata, version 12.1 (StatCorp LP,College Station, TX), and strs module for relative survival anal-ysis (Dickman PW, Coviello E, Hills M, version 1.3.8).

Results

Patient characteristics

We identified 7774 stage IE/IIE MALT lymphoma patients(Table 1). There were significant differences with regard to age,sex, race, and geographic distribution depending on the anatomicsite of origin, as described previously (4). Among the other non-subcategorized sites, the most common locations were connectivetissue (nZ182), genitourinary tract (nZ116), gastrointestinal or-gans (eg, liver, gallbladder, nZ66), and central nervous system,including meninges (nZ49). The use of surgical excision andradiation therapy significantly varied according to site. Surgicalexcision was performed most commonly in thyroid, salivary, in-testinal, and mammary sites. Postexcision radiation therapy wasprescribed in one-third of cases and was rare in pulmonary,

Table 2 Factors associated with use of radiation therapy instage IE/IIE MALT lymphoma in a multivariate logistic model

Variable Odds ratio 95% CI P

Age category20-49 y Reference50-59 y 1 (0.84-1.19) .9960-69 y 0.95 (0.80-1.13) .5770-79 y 0.70 (0.59-0.83) .000180 y 0.48 (0.40-0.59) <.0001

Sex and race/ethnicityMen

White non-Hispanic ReferenceHispanic 0.67 (0.50-0.89) .006Black 0.48 (0.33-0.69) .0001Asian/other 0.92 (0.70-1.20) .53

WomenWhite non-Hispanic 0.83 (0.73-0.94) .005Hispanic 0.62 (0.49-0.80) .0002Black 0.82 (0.62-1.08) .15Asian/other 1.23 (0.94-1.60) .13

Poverty level*

�8% Reference8-11% 0.93 (0.79-1.08) .3211-16% 0.76 (0.65-0.89) .0005>16% 0.71 (0.60-0.85) .0002

Geographic areaMetropolitan ReferenceUrban 0.91 (0.74-1.11) .34Rural including Alaska 0.52 (0.31-0.87) .01

United States regionWest ReferenceEast 1.19 (1.01-1.40) .04South 1.01 (0.85-1.20) .88Midwest 1.4 (1.15-1.70) .001

StageIE ReferenceIIE 0.83 (0.70-0.98) .03

B symptomsAbsent ReferencePresent 0.71 (0.58-0.89) .002Unrecorded 0.40 (0.36-0.46) <.0001

MALT primary siteStomach ReferenceSkin 2.57 (2.10-3.15) <.0001Ocular 5.76 (4.87-6.81) <.0001Salivary glands 2.5 (2.03-3.07) <.0001Thyroid 3.31 (2.39-4.58) <.0001Other head/neck 2.65 (2.05-3.42) <.0001Breast 2.28 (1.70-3.06) <.0001Lung 0.27 (0.20-0.36) <.0001Bowel 0.36 (0.28-0.47) <.0001

Year of diagnosis1998-2002 Reference2003-2005 0.86 (0.74-1.00) .052006-2008 0.91 (0.79-1.06) .232009-2010 0.90 (0.76-1.06) .20

Excision of primary site 0.65 (0.57-0.75) <.0001

Abbreviations: CI Z confidence interval; MALT Z mucosa-

associated lymphoid tissue.

* Expressed as percentage of persons living below the federal

poverty level in patient’s census track area.

Volume 88 � Number 3 � 2014 Radiation therapy in MALT lymphoma 645

intestinal, or gastric MALT lymphoma but was very common inocular locations. With a median follow-up time of 4.6 years, therelative survival for the entire cohort was 92.0% at 5 years and85.7% at 10 years. There were 1763 deaths, 30% of which wereascribed to lymphoma on death certificates, but in up to 20% ofevents the death was ascribed to a “cancer” of the organ of origin,“miscellaneous malignancy,” or an undefined cause.

Use of radiation therapy

In a multivariable logistic regression model, the use of radiationtherapy was associated not only with site and stage of lymphomabut also with patients’ age, sex, race, and socioeconomic status(Table 2). The treatment was administered in 46% of patientsyounger than 50 years but only in 29% of those older than 70years. The significant interaction between sex and race (overalltest for interaction, PZ.001) was incorporated in the model by useof separate categories for each group. In comparison with whitenon-Hispanic men, radiation therapy was less commonly admin-istered to Hispanic or black men and to all women except Asians.Patients living in the areas with higher poverty levels or in ruralcounties also had lower rates of radiation therapy administration.Patients were less likely to receive radiation therapy after surgicalexcision. The model had a c statistic of 0.76 and a c2 statistic of3.54 (PZ.90) on the Hosmer-Lemeshow goodness-of-fit test with10 groups, indicating adequate fit.

Because the unrecorded presence of H. pylori infection waslikely a major factor confounding the administration of radiationtherapy in gastric MALT lymphoma, we analyzed an analogousmodel excluding that anatomic location. We found no substantialdifference in the findings except for a less pronounced disparity inmen (Hispanic: odds ratio [OR] 0.73, 95% CI 0.50-1.06, PZ.09;black: OR 0.63, 95% CI 0.37-1.06, PZ.08) and an equally pro-nounced disparity in women (white: OR 0.74, 95% CI 0.62-0.88,PZ.0006; Hispanic, OR 0.66, 95% CI 0.48-0.90, PZ.008; black:OR 0.75, 95% CI 0.52-1.08, PZ.12; Asian: OR 1.27, 95% CI0.90-1.80, PZ.17).

Additionally, SEER has recorded insurance status at the timeof diagnosis since 2007. In a model limited to 2451 patients whoreceived diagnoses between 2007 and 2010, there was evidence ofsignificantly lower rates of radiation therapy administration inthose who were uninsured (OR 0.45, 95% CI 0.22-0.92, PZ.03)or who had state-sponsored Medicaid insurance (OR 0.55, 95% CI0.38-0.80, PZ.002). The racial and sex-related disparities per-sisted in that analysis.

Association of radiation therapy with survival

We calculated grouped estimates of 5-year risk of LRD in patientswith MALT lymphomas of each anatomic site (Table 3, Fig. 1).These estimates do not account for factors confounding treatmentselection, which is reflected by a lower risk of death fromcompeting causes in the treated group. Consequently, they shouldbe interpreted as aggregate outcomes derived from populationswith potentially disparate clinical indications, contraindications,and patient/physician preferences. In patients who received radi-ation therapy, the risk of LRD was very low in all MALT sitesexcept for intestinal and head and neck mucosa. In cutaneous,ocular, salivary, and thyroid MALT lymphomas, this risk wasessentially zero. Differences in mortality from competing causes

Table 3 Cumulative incidence of death related to lymphoma and competing causes in stage IE/IIE MALT lymphoma patients,stratified by receipt of radiation therapy

MALT primary site All patients Radiation therapy No radiation therapy

Lymphoma-related death at 5 y (95% CI)Skin 0% * 0% * 4.0% (1.2-9.9)Ocular 2.5% (0.7-6.3) 0% * 11.5% (6.5-18.1)Salivary glands 4.2% (1.7-8.4) 0% * 8.4% (4.2-14.2)Thyroid gland 2.1% (0.4-6.6) 0.4% (0-81.7) 4.8% (1.3-11.8)Head/neck mucosa 8.2% (3.7-15.2) 9.3% (3.4-19.0) 6.0% (0.9-18.3)Breast 9.3% (4.0-17.4) 4.9% (0.3-20.3) 12.8% (5.3-23.8)Lung 10.2% (6.2-15.3) 2.9% (0-34.8) 10.6% (6.3-16.1)Stomach 9.8% (7.9-11.9) 3.8% (1.4-8.1) 11.9% (9.6-14.5)Small/large intestine 9.4% (5.9-13.9) 8.2% (1.2-24.4) 9.4% (5.7-14.3)

Death from competing causes at 5 y (95%CI)Skin 9.9% (9.8-10.1) 8.7% (8.6-8.8) 10.9% (10.6-11.2)Ocular 13.3% (13.1-13.5) 12.7% (12.5-12.9) 14.6% (14.1-15.0)Salivary glands 11.2% (11.0-11.4) 9.8% (9.7-10.0) 12.2% (11.9-12.5)Thyroid gland 11.5% (11.1-11.9) 9.8% (9.5-10.1) 13.7% (12.9-14.5)Head/neck mucosa 14.3% (13.8-14.8) 11.4% (11.0-11.9) 17.5% (16.5-18.4)Breast 13.7% (13.2-14.1) 12.3% (11.8-12.8) 14.7% (14.0-15.5)Lung 14.3% (14.0-14.7) 13.8% (12.9-14.8) 14.4% (14.0-14.8)Stomach 15.7% (15.5-15.9) 12.0% (11.8-12.3) 17.0% (16.7-17.3)Small/large intestine 14.2% (13.8-14.6) 12.6% (11.7-13.5) 14.3% (13.8-14.7)

Abbreviations: CI Z confidence interval; MALT Z mucosa-associated lymphoid tissue.

* The confidence interval could not be calculated because of negative calculated values (adjusted to zero).

Olszewski and Desai International Journal of Radiation Oncology � Biology � Physics646

between anatomic locations were smaller, primarily reflectingvarying age distributions.

To further characterize the association between radiationtherapy use and survival, we fitted exploratory multivariable Coxmodels for the categorized MALT lymphoma sites in 6108 pa-tients who survived at least 12 months after diagnosis. The modelsminimized the immortal time bias, and bias related to availableconfounders: age, sex, race/ethnicity, stage, B symptoms, andexcisional surgery. The overall interaction test for heterogeneity ina model including all cases was significant, with PZ.005. Therewas no significant association of survival with race or socioeco-nomic status, but survival was better in women (HR 0.70, 95% CI0.63-0.79, P<.0001). In models fitted for each MALT lymphomasubtype, the association between radiation therapy use and sur-vival was statistically significant only for cutaneous and ocularlocations (Fig. 2). Conversely, the hazard ratio for the intestinaland head and neck mucosal sites did not suggest a favorablecorrelation, although the confidence intervals were wide. Therewas no significant interaction between radiation therapy and stage(I vs II), and the results were consistent in models limited to stageIE cases (data not shown). There were no substantial differences inanalogous competing risk models when death certificate meth-odology was used, although the estimates were much less precisebecause of a small number of LRD events (<20 in thyroid, skin,and breast lymphomas).

Discussion

In this cohort of patients with stage I/II MALT lymphoma, westudied the use of radiation therapy as part of the initial treatmentand the subsequent survival outcomes. The strengths of the anal-ysis include size of the cohortdthe largest reported to datedand

its comprehensive, population-based source, contrasting with priorretrospective case series. Although we could not describeprogression-free survival, we accounted for the competing risks ofdeath predominating in these indolent lymphomas, and thus oursite-specific survival estimates are of practical value when patientsare counseled about their prognosis (13).

Significant disparities in the use of radiation therapy exist withregard to age, sex, race/ethnicity, and socioeconomic and insurancestatus, after disease-related factors are controlled for. These in-equalities have not been described in prior studies of non-Hodgkinlymphoma and raise questions about barriers to therapy and aboutpatient and physician preferences. Considering the significant as-sociation of radiation therapy use with survival in MALT lym-phomas, efforts are warranted to minimize the disparities that do notarise from well-defined clinical contraindications.

The risk of LRD after radiation therapy was very low for thecutaneous, ocular, salivary, and thyroid sites. In MALT lym-phomas of the skin, a review of 132 published cases found thatradiation therapy provided local control in over 99% of cases, butrecurrences were as common as after excision alone (46% and43%, respectively) and were mostly confined to the skin (15). Ouranalysis suggests that patients undergoing radiation therapy, whoperhaps also receive regular surveillance and retreatment, have areassuringly very low excess mortality related to the lymphoma.The significant association of survival with administration of ra-diation therapy could not be explained by bias related to otherstudied characteristics. Although a causal interpretation of thisassociation would require a randomized study, our results supportradiation therapy as the primary modality in cutaneous MALTlymphoma.

Prior series of ocular lymphoma documented high local controlrates with radiation therapy but also a high rate of ophthalmologictoxicities: xerophthalmia, cataracts, and keratitis. Treatment with

0.0

0.1

0.2

0.3

0.0

0.1

0.2

0.3

0.0

0.1

0.2

0.3

0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5

sdnalg yravilaSralucOnikS

Thyroid gland Head and neck mucosa Breast

enitsetni egral / llamShcamotSgnuL

Death due to MALT lymphoma Death due to competing causes

Cum

ulat

ive

inci

denc

e of

dea

th

Years since diagnois

Radiation therapy Radiation therapy Radiation therapy

Radiation therapy Radiation therapy Radiation therapy

Radiation therapy Radiation therapy Radiation therapy

No radiation therapy No radiation therapy No radiation therapy

No radiation therapy No radiation therapy No radiation therapy

No radiation therapy No radiation therapy No radiation therapy

Fig. 1. Cumulative incidence of death as a result of lymphoma or other causes in patients with mucosa-associated lymphoid tissue(MALT) lymphomas, with or without radiation therapy as initial treatment. The estimates were calculated using expected mortalitystratified by age, sex, race, and calendar year; negative calculated increments were adjusted to zero.

Volume 88 � Number 3 � 2014 Radiation therapy in MALT lymphoma 647

doxycycline for Chlamydia infection showed efficacy in a Euro-pean multicenter study, but the association with the pathogen wasnot confirmed in American series, possibly reflecting a geographicvariation (16, 17). Goda et al (7) described 89 patients with stageIE ocular MALT lymphoma, with 99% complete response to ra-diation therapy and a 25% rate of recurrence, predominantlydistant. The cause-specific survival at 7 years was 96%. Latecomplications were reported in 45% of cases, including a 25% riskof grade 3 cataracts, although lens shielding reduced their inci-dence. The lower dose of radiation (25 Gy) in this study was alsoassociated with fewer complications and comparable diseasecontrol, although some authors favor higher doses of �30 Gy (18).Son et al (19) reported 100% cause-specific 5-year survival afterradiation therapy (21-45 Gy) in 43 patients and a low rate of latecomplications (4%) with the use of lens shielding and lacrimal

gland blocks. In another Korean series, chemotherapy alonecontrolled the lymphoma in only 56% of cases, requiringconsolidation radiation therapy (9). The study by Tanimoto et al(20) suggested improved progression-free survival (PZ.016) butnot overall survival (PZ.091) with radiation therapy. Althoughdistant recurrences appear to be common in ocular MALT lym-phoma, our analysis corroborates the association of radiationtherapy administration with improved survival in a large cohort.

With regard to salivary gland MALT lymphoma, Anacak et al(21) reported a series of 63 patients, of whom 41 were treated withradiation therapy alone or in combination with surgery orchemotherapy. Radiation therapy was associated with betterdisease-free survival (66% vs 34%), but the differences in overallor disease-specific survival were not significant. Similarly, we didnot detect a significant association with overall survival in 623

0.0 0.5 1.0 1.5 2.0

Skin

Ocular

Salivary

Thyroid

Head/neck

Breast

Lung

Stomach

Bowel

Primary site of the lymphoma

0.45 (0.24-0.82)

0.47 (0.34-0.66)

0.79 (0.53-1.17)

0.93 (0.40-2.17)

1.04 (0.58-1.86)

0.71 (0.38-1.32)

0.63 (0.30-1.30)

0.92 (0.74-1.13)

1.04 (0.53-2.03)

0.009

<0.0001

0.23

0.89

0.28

0.21

0.43

0.91

0.87

Hazard ratio (95% CI) P

BA

0.0 0.5 1.0 1.5 2.0

0.27 (0.08-0.96)

0.46 (0.24-0.87)

0.79 (0.40-1.56)

0.46 (0.11-1.86)

0.91 (0.39-2.15)

0.62 (0.22-1.74)

0.64 (0.19-2.11)

1.06 (0.71-1.58)

1.94 (0.83-4.55)

0.043

0.016

0.51

0.83

0.36

0.46

0.76

0.13

0.28

Subhazard ratio (95% CI) P

Fig. 2. (A) Association of overall survival with radiation therapy administration in stage I/II mucosa-associated lymphoid tissue lym-phomas of different primary sites. Hazard ratios are derived from multivariate Cox models with adjustment for age, sex, race/ethnicity, stage,B-symptoms, and excisional surgery. (B) Analogous models for subhazard of lymphoma-related death from competing risk regression.

Olszewski and Desai International Journal of Radiation Oncology � Biology � Physics648

cases. In other MALT lymphomas of the head and neck mucosa,we observed a higher risk of LRD in patients treated with radia-tion therapy (9.3% vs 6.0%) despite lower mortality from unre-lated causes (11.4% vs 17.5%). This unexpected finding maysuggest lower efficacy of local therapy, especially because the riskof LRD for this anatomic location is overall relatively high.Further research on MALT lymphomas arising from oropharyn-geal mucosa is necessary because they have not been specificallystudied and are typically pooled together with other histologictypes (22). In intestinal MALT lymphomas there was a similarlack of association with survival. The very low rates (w10%) ofthe use of radiation therapy in pulmonary and intestinal lym-phomas reflect a high selectivity of clinical decisions in thosesubtypes with relatively inferior prognosis.

There are several limitations to this SEER-database analysis.Because of lack of data, we could not study elements commonlyreported in radiation therapy studies: doses and fields of radiation,duration of local control, progression-free survival, or treatmenttoxicities. These factors are of essential value in the making ofclinical decisions. However, from a patient’s perspective, therisk of dying as a result of the malignancy may be particularlyimportant because recurrences of MALT lymphoma are treatableand may not necessarily have an impact on survival. Progression-free survival is less useful when the majority of deaths are unre-lated to cancer. The very low or even zero excess mortality afterradiation therapy in most MALT lymphomas seen in our studycompared with age-, sex- and race-matched populations is thusreassuring. Although we attempted to study the hazard of LRDusing information from death certificates, we found such infor-mation unreliable for MALT lymphomas because of the signifi-cant proportion of unrecorded or likely misclassified cause ofdeath attributions. Such misclassification and the resulting biashave been previously described (23, 24). Therefore, we primarilyrelied on an alternative methodology, using stratified tables ofexpected mortality to calculate the excess risk of LRD.

Despite multivariable adjustment, our analysis is descriptiverather than comparative because of the unavailability of someimportant confounders in the database. This limitation precludescausal interpretation of the results. Patients who do not receiveradiation therapy may have other adverse features related to in-dividual characteristics (performance status, contraindications,intercurrent events causing treatment delay) or to the disease(systemic progression, tumor bulk or location favoring

chemotherapy). Some patients with small lesions may have elec-ted excision alone, so our “untreated” group consists of a mix ofcases with favorable and unfavorable prognosis. The inability toaccount for systemic treatments with antibiotics or chemotherapyis another significant limitation. Similarly, we were unable toassess the toxicity of radiation therapy. A randomized trial wouldbe necessary to properly compare different treatment modalities,and our analysis indicates that it would need to account for thesite-specific heterogeneity of treatment effects.

In conclusion, the administration of radiation therapy is asso-ciated with a low risk of LRD in early-stage MALT lymphomas,and this association is particularly significant for cutaneous andocular sites. Women, black and Hispanic minorities, socioeco-nomically disadvantaged patients, and underinsured patients maybe at risk of not receiving radiation therapy. Our results enableimproved, precise prognostic counseling for patients with MALTlymphomas. They also have implications for future studies in thisdisease and for healthcare policies facilitating access to radiationtherapy.

References

1. Morton LM, Turner JJ, Cerhan JR, et al. Proposed classification of

lymphoid neoplasms for epidemiologic research from the Pathology

Working Group of the International Lymphoma Epidemiology Con-

sortium (InterLymph). Blood 2007;110:695-708.

2. Zinzani PL. The many faces of marginal zone lymphoma. Hematology

Am Soc Hematol Educ Program 2012;2012:426-432.

3. Zucca E, Dreyling M. ESMO Guidelines Working Group. Gastric

marginal zone lymphoma of MALT type: ESMO clinical recommen-

dations for diagnosis, treatment and follow-up. Ann Oncol 2009;

20(Suppl 4):113-114.

4. Olszewski AJ, Castillo JJ. Survival of patients with marginal zone

lymphoma: analysis of the Surveillance, Epidemiology, and End Re-

sults database. Cancer 2013;119:629-638.

5. Tsai HK, Li S, Ng AK, et al. Role of radiation therapy in the treatment

of stage I/II mucosa-associated lymphoid tissue lymphoma. Ann Oncol

2007;18:672-678.

6. Isobe K, Kagami Y, Higuchi K, et al. A multicenter phase II study of

local radiation therapy for stage IEA mucosa-associated lymphoid

tissue lymphomas: A preliminary report from the Japan Radiation

Oncology Group (JAROG). Int J Radiat Oncol Biol Phys 2007;69:

1181-1186.

Volume 88 � Number 3 � 2014 Radiation therapy in MALT lymphoma 649

7. Goda JS, Le LW, Lapperriere NJ, et al. Localized orbital mucosa-

associated lymphoma tissue lymphoma managed with primary radia-

tion therapy: Efficacy and toxicity. Int J Radiat Oncol Biol Phys 2011;

81:e659-e666.

8. Parbhakar S, Cin AD. Primary cutaneous B-cell lymphoma: Role of

surgery. Can J Plast Surg 2011;19:e12-e14.

9. Paik JS, Cho WK, Lee SE, et al. Ophthalmologic outcomes after

chemotherapy and/or radiotherapy in non-conjunctival ocular adnexal

MALT lymphoma. Ann Hematol 2012;91:1393-1401.

10. Surveillance, Epidemiology, and End Results (SEER) Program (www.

seer.cancer.gov) SEER*Stat Database: Incidence - SEER 18 Regs

Research Data þ Hurricane Katrina Impacted Louisiana Cases, Nov

2012 Sub (1973-2010 varying) - Linked To County Attributes - Total

U.S., 1969-2011 Counties, National Cancer Institute, DCCPS, Sur-

veillance Research Program, Surveillance Systems Branch, released

April 2013, based on the November 2012 submission, 2013.

11. Turner JJ, Morton LM, Linet MS, et al. InterLymph hierarchical

classification of lymphoid neoplasms for epidemiologic research based

on the WHO classification (2008): Update and future directions. Blood

2010;116:e90-e98.

12. Cronin KA, Feuer EJ. Cumulative cause-specific mortality for cancer

patients in the presence of other causes: A crude analogue of relative

survival. Stat Med 2000;19:1729-1740.

13. Eloranta S, Adolfsson J, Lambert PC, et al. How can we make cancer

survival statistics more useful for patients and clinicians: An illus-

tration using localized prostate cancer in Sweden. Cancer Causes

Control 2013;24:505-515.

14. Park HS, Gross CP, Makarov DV, et al. Immortal time bias: A frequently

unrecognized threat to validity in the evaluation of postoperative radio-

therapy. Int J Radiat Oncol Biol Phys 2012;83:1365-1373.

15. Senff NJ, Noordijk EM, Kim YH, et al. European Organization for

Research and Treatment of Cancer and International Society for

Cutaneous Lymphoma consensus recommendations for the manage-

ment of cutaneous B-cell lymphomas. Blood 2008;112:1600-1609.

16. Ferreri AJ, Govi S, Pasini E, et al. Chlamydophila psittaci eradication

with doxycycline as first-line targeted therapy for ocular adnexae

lymphoma: Final results of an international phase II trial. J Clin Oncol

2012;30:2988-2994.

17. Rosado MF, Byrne GE Jr., Ding F, et al. Ocular adnexal lymphoma: A

clinicopathologic study of a large cohort of patients with no evidence

for an association with Chlamydia psittaci. Blood 2006;107:467-472.

18. Stefanovic A, Lossos IS. Extranodal marginal zone lymphoma of the

ocular adnexa. Blood 2009;114:501-510.

19. Son SH, Choi BO, Kim GW, et al. Primary radiation therapy in pa-

tients with localized orbital marginal zone B-cell lymphoma of

mucosa-associated lymphoid tissue (MALT Lymphoma). Int J Radiat

Oncol Biol Phys 2010;77:86-91.

20. Tanimoto K, Kaneko A, Suzuki S, et al. Primary ocular adnexal

MALT lymphoma: A long-term follow-up study of 114 patients. Jpn J

Clin Oncol 2007;37:337-344.

21. Anacak Y, Miller RC, Constantinou N, et al. Primary mucosa-

associated lymphoid tissue lymphoma of the salivary glands: A

multicenter Rare Cancer Network study. Int J Radiat Oncol Biol Phys

2012;82:315-320.

22. MacDermed D, Thurber L, George TI, et al. Extranodal nonorbital

indolent lymphomas of the head and neck: Relationship between

tumor control and radiotherapy. Int J Radiat Oncol Biol Phys 2004;59:

788-795.

23. Percy C, Stanek E 3rd, Gloeckler L. Accuracy of cancer death cer-

tificates and its effect on cancer mortality statistics. Am J Public

Health 1981;71:242-250.

24. Hinchliffe SR, Abrams KR, Lambert PC. The impact of under and

over-recording of cancer on death certificates in a competing risks

analysis: A simulation study. Cancer Epidemiol 2013;37:11-19.