Advertisement

Survival benefit associated with treating follicular lymphoma in patients 80 years or older

Published:November 06, 2019DOI:https://doi.org/10.1016/j.jgo.2019.10.003

      Abstract

      Objectives

      To evaluate the overall survival benefit associated with follicular lymphoma (FL)-directed therapy among patients diagnosed with FL at 80+ years.

      Patients and methods

      This retrospective cohort study utilized the linked Surveillance, Epidemiology and End Results-Medicare dataset to identify patients 80+ years, diagnosed with FL between 2000 and 2013. We identified FL-directed treatments based on published guidelines. We utilized a propensity-score matched sample to compare treated and untreated groups who had similar observed characteristics. We reported the median overall survival time and the 3-year restricted mean survival time (RMST) of the study groups as well as the hazard ratio (HR) of death associated with treatment receipt.

      Results

      We identified 3705 older patients with FL (mean [SD] age, 84 [3.6] years). Over a median follow-up of 2.9 years, 68% of the sample received FL-directed therapy and the most common regimen was rituximab monotherapy (N = 768, 21%). The matched sample included 2306 patients. The median overall survival for the treated group was 4.31 years (95% confidence interval [CI], 4.00–4.61) compared to 2.86 years (95% CI, 2.59–3.16) for the untreated group. The 3-year RMST for the treated group was 2.36 years (95% CI, 2.30–2.41), while it was 2.05 years (95% CI, 1.98–2.11) for the untreated group. Treatment was associated with a 23% reduction in the hazards of death (HR: 0.77, 95% CI: 0.70–0.85; p < .001).

      Conclusion

      FL-directed therapy was associated with improved survival among patients diagnosed with FL at 80+ years. These findings can support treatment decision-making for individuals diagnosed with FL at older ages.

      Keywords

      1. Introduction

      Follicular lymphoma (FL) is the second most common form of non-Hodgkin's lymphoma (NHL) [
      • Ganti AK
      • Bociek RG
      • Bierman PJ
      • Enke CA
      • Vose JM
      • Armitage JO
      Follicular lymphoma: expanding therapeutic options.
      ]. It represents 35% of adult NHL in the US [
      • Ganti AK
      • Bociek RG
      • Bierman PJ
      • Enke CA
      • Vose JM
      • Armitage JO
      Follicular lymphoma: expanding therapeutic options.
      ]. The treatment options and outcomes of FL patients have improved considerably in the past years owing to the improved supportive modalities, introduction of rituximab, and effective sequential and combination therapies [
      • Swenson WT
      • Wooldridge JE
      • Lynch CF
      • Forman-Hoffman VL
      • Chrischilles E
      • Link BK
      Improved survival of follicular lymphoma patients in the United States.
      ,
      • Liu Q
      • Fayad L
      • Cabanillas F
      • et al.
      Improvement of overall and failure-free survival in stage IV follicular lymphoma: 25 years of treatment experience at The University of Texas MD Anderson Cancer Center.
      ,
      • Conconi A
      • Motta M
      • Bertoni F
      • et al.
      Patterns of survival of follicular lymphomas at a single institution through three decades.
      ,
      • Tan D
      • Horning SJ
      • Hoppe RT
      • et al.
      Improvements in observed and relative survival in follicular grade 1-2 lymphoma during 4 decades: the Stanford University experience.
      ]. Multiple first-line regimens are available for managing FL, including regimens suggested for older or frail patients [
      • Network N.C.C.
      NCCN clinical practice guidelines in oncology (NCCN guidelines®) non-Hodgkin's lymphomas, Version 1.2016.
      ]. Also, depending on disease burden and patient characteristics, clinicians may choose initial observation until progression before initiating treatment in what is known as the watchful waiting approach [
      • Solal-Céligny P
      • Bellei M
      • Marcheselli L
      • et al.
      Watchful waiting in low–tumor burden follicular lymphoma in the rituximab era: results of an F2-study database.
      ]. However, the survival benefit associated with treatment receipt among patients 80 years or older remains unclear.
      Evaluating the benefit of FL-directed therapy is particularly important among older patients for several reasons. First, the US population continues to age and estimates indicate that individuals aged 80 years and older will increase by 67% between 2012 and 2030 [
      • Ortman JM
      • Velkoff VA
      • Hogan H
      An aging nation: the older population in the United States.
      ]. Second, although the incidence rate of NHL has stabilized in recent years [
      • Ward EM
      • Ma J
      • Jemal A
      • et al.
      Annual report to the nation on the status of cancer, 1975–2014, featuring survival.
      ], it has increased by >50% in the past decades, and it increased even more among individuals older than 60 years [
      • Thieblemont C
      • Coiffier B
      Lymphoma in older patients.
      ]. Third, recent estimates of FL incidence rates showed that 51% of all FL cases occur in individuals 65 years or older [
      • Noone A
      • Howlader N
      • Krapcho M
      • et al.
      SEER cancer statistics review, 1975-2015.
      ]. Fourth, based on the Follicular Lymphoma International Prognostic Index (FLIPI), advanced age (>60 years) is an independent negative prognostic factor [
      • Solal-Celigny P
      • Roy P
      • Colombat P
      • et al.
      Follicular lymphoma international prognostic index.
      ].
      However, the optimal management of FL patients 80 years or older remains unclear [
      • Nabhan C
      • Byrtek M
      • Rai A
      • et al.
      Disease characteristics, treatment patterns, prognosis, outcomes and lymphoma-related mortality in elderly follicular lymphoma in the United States.
      ]. Physicians' decision to administer chemotherapy to older patients is complex and involves considering the tumor burden, patients' comorbidities and health status, and the availability of social support [
      • Wan-Chow-Wah D
      • Monette J
      • Monette M
      • et al.
      Difficulties in decision making regarding chemotherapy for older cancer patients: a census of cancer physicians.
      ]. Furthermore, the administration of therapies at effective doses can be hindered by the presence of comorbidities or impaired health status among older patients [
      • Nabhan C
      • Smith SM
      • Helenowski I
      • et al.
      Analysis of very elderly (≥ 80 years) non-hodgkin lymphoma: impact of functional status and co-morbidities on outcome.
      ]. Finally, there is a paucity of chemotherapy safety and efficacy data in patients 80 years or older as they are typically not included in clinical trials [
      • Hiddemann W
      • Kneba M
      • Dreyling M
      • et al.
      Frontline therapy with rituximab added to the combination of cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) significantly improves the outcome for patients with advanced-stage follicular lymphoma compared with therapy with CHOP alone: results of a prospective randomized study of the German Low-Grade Lymphoma Study Group.
      ,
      • Pfreundschuh M
      • Trümper L
      • Österborg A
      • et al.
      CHOP-like chemotherapy plus rituximab versus CHOP-like chemotherapy alone in young patients with good-prognosis diffuse large-B-cell lymphoma: a randomised controlled trial by the MabThera International Trial (MInT) Group.
      ,
      • Ghielmini M
      • Schmitz S-FH
      • Cogliatti SB
      • et al.
      Prolonged treatment with rituximab in patients with follicular lymphoma significantly increases event-free survival and response duration compared with the standard weekly× 4 schedule.
      ,
      • Marcus R
      • Imrie K
      • Belch A
      • et al.
      CVP chemotherapy plus rituximab compared with CVP as first-line treatment for advanced follicular lymphoma.
      ,
      • Talarico L
      • Chen G
      • Pazdur R
      Enrollment of elderly patients in clinical trials for cancer drug registration: a 7-year experience by the US Food and Drug Administration.
      ,
      • Lewis JH
      • Kilgore ML
      • Goldman DP
      • et al.
      Participation of patients 65 years of age or older in cancer clinical trials.
      ,
      • Hutchins LF
      • Unger JM
      • Crowley JJ
      • Coltman CA
      • Albain KS
      Underrepresentation of patients 65 years of age or older in cancer-treatment trials.
      ].
      This study fills a gap in the literature by examining the survival benefit associated with treatment receipt among older patients with FL. We identified a cohort of FL patients 80 years or older and evaluated the overall survival of patients who received FL-directed therapy and compared them to patients who did not. We identified therapies recommended by the National Comprehensive Cancer Network (NCCN) guidelines published between 1997 and 2016 [
      • Network N.C.C.
      NCCN clinical practice guidelines in oncology (NCCN guidelines®) non-Hodgkin's lymphomas, Version 1.2016.
      ]. We controlled for a range of sociodemographic and clinical factors that might influence treatment receipt and survival or survival alone, including comorbidity and health status.

      2. Patients and Methods

      2.1 Data Source and Study Population

      We conducted this retrospective cohort study using the Surveillance, Epidemiology and End Results (SEER)-Medicare database, which links data from the SEER registry to the insurance claims of eligible Medicare enrollees. The SEER program collects cancer incidence data from cancer registries that cover approximately 35% of the US population [
      • National Cancer Institute
      Overview of the SEER program.
      ]. We identified incident FL cases from SEER data, which also provides demographic and tumor-related information, such as stage at diagnosis and lymphoma grade. Medicare claims were used to identify the utilization of health services, including the receipt of FL-directed therapy.
      The study sample consisted of FL cases newly-diagnosed between 2000 and 2013 with associated Medicare claims data from 1999 until 2014. We included patients who were 80 years or older at the time of diagnosis and with continuous enrollment in Medicare Parts A and B in the 12 months prior to diagnosis (baseline period). Patients were excluded for the following reasons: health maintenance organization (HMO) enrollment in the baseline period; missing diagnosis date; FL diagnosis on autopsy; <60 days of survival after FL diagnosis. Included patients were followed from the year prior to diagnosis until death or censoring. Following diagnosis, patients were censored if they enrolled in an HMO; lost Medicare Parts A or B or if they were alive at the end of the study period (12/31/2014).

      2.2 Treatment Receipt

      Treatment receipt was defined as receiving any FL-directed therapy as recommended by the NCCN guidelines published between 1997 and 2016 [
      • Network N.C.C.
      NCCN clinical practice guidelines in oncology (NCCN guidelines®) non-Hodgkin's lymphomas, Version 1.2016.
      ]. Older versions of the guidelines were requested from the NCCN. We used the National Drug Code (NDC) and the Healthcare Common Procedures Coding System (HCPCS) codes to capture claims for any drug product listed in the NCCN guidelines. The receipt of radiation therapy was captured using the Current Procedural Terminology (CPT) codes associated with radiation. For treatment regimens that consisted of multiple agents, such as RCHOP (rituximab with cyclophosphamide, doxorubicin, vincristine, and prednisone), we developed an algorithm that captured claims for each of the constituent medications within a specified time period and output the multi-agent treatment regimen. Patients with claims for any FL-directed therapy were referred to as the treated group. Patients with no claims for FL-directed therapy were referred to as the untreated group.

      2.3 Overall Survival

      Overall survival (OS) was the outcome of interest. OS was estimated from the date of FL diagnosis until death or censoring. Date of death was obtained from Medicare records, and date of censoring was anchored to the censoring reasons described above.

      2.4 Covariates

      We evaluated sociodemographic, clinical and health status factors that could be associated with treatment receipt and/or survival. Sociodemographic factors included age, race, gender, marital status, urban residence, geographic region (Northeast, Midwest, South, and West), diagnosis period (early, 2000–2005; middle, 2006–2010; and late, 2010–2013), and state buy-in, which is a proxy for socioeconomic disadvantage. Available clinical factors included Ann Arbor stage at diagnosis, FL grade (grade 1–2 vs. grade 3), and Charlson Comorbidity Index (CCI). The CCI score was estimated based on the presence of comorbid conditions in the baseline period in claims data. Comorbid conditions were counted and weighted depending on their expected impact on non-cancer mortality [
      • Charlson ME
      • Pompei P
      • Ales KL
      • MacKenzie CR
      A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.
      ,
      • Deyo RA
      • Cherkin DC
      • Ciol MA
      Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases.
      ]. Health status was assessed using a proxy for poor performance status that was developed previously in other cancer sites [
      • Davidoff AJ
      • Zuckerman IH
      • Pandya N
      • et al.
      A novel approach to improve health status measurement in observational claims-based studies of cancer treatment and outcomes.
      ]. The proxy measure of poor performance status was defined as the presence of at least one claim for hospitalization, skilled nursing facility stay, oxygen use, walking aids or a wheelchair in the baseline period. Finally, we included the utilization of preventive services in the baseline period as a proxy for the healthy user effect, which is the tendency of patients using preventive services to engage in other healthy behaviors [
      • Shrank WH
      • Patrick AR
      • Brookhart MA
      Healthy user and related biases in observational studies of preventive interventions: a primer for physicians.
      ]. Patients were indicated to utilize preventive services if they had at least one claim for any of the following services in the baseline period: flu vaccine; pneumococcal vaccine; prostate, breast or colorectal cancers screening tests; bone mineral density test or a visit to a primary care provider.

      2.5 Statistical Analysis

      We used two-sample t-tests to compare the mean age and mean follow-up time of the treated and untreated groups. Chi-squared tests were used to compare categorical variables across the comparison groups.
      In order to reduce bias from systematic differences between the treated and the untreated groups, we conducted 1:1 propensity-score matching to identify a cohort in which treated and untreated patients were similar on observed characteristics associated with treatment receipt and survival or survival alone [
      • Austin PC
      An introduction to propensity score methods for reducing the effects of confounding in observational studies.
      ,
      • Austin PC
      • Grootendorst P
      • Anderson GM
      A comparison of the ability of different propensity score models to balance measured variables between treated and untreated subjects: a Monte Carlo study.
      ]. Factors associated with treatment receipt and survival were identified using a logistic regression model of treatment receipt and a Cox proportional-hazards (PH) model, respectively. Accordingly, propensity scores were estimated using a logistic regression model that included the following variables: age, race, gender, marital status, state buy-in, geographic region, CCI, performance status, preventive services use, and diagnosis period. We used a greedy matching algorithm to achieve satisfactory exact matching while retaining the maximum number of cases and controls possible [
      • Parsons L.
      Reducing bias in a propensity score matched-pair sample using greedy matching techniques.
      ]. After matching, we obtained two cohorts of equal sample size and with similar observed characteristics. We tested the goodness of matching by comparing the characteristics of the two samples using Chi-square tests and standardized mean differences.
      To answer the study question, we compared the survival rates of the treated and the untreated groups post matching, using Kaplan-Meier (KM) survival estimates [
      • Kaplan EL
      • Meier P
      Nonparametric estimation from incomplete observations.
      ]. We assessed the statistical significance of the difference in survival rates using the Wilcoxon and log-rank tests. For both groups, we reported the median survival time as well as the survival probability in each of the first 5 years after diagnosis. Additionally, we compared the restricted mean survival time (RMST) of both groups at 3 years [
      • Royston P
      • Parmar MK
      Restricted mean survival time: an alternative to the hazard ratio for the design and analysis of randomized trials with a time-to-event outcome.
      ]. Finally, we estimated a Cox PH model to determine the hazards ratio of death among the treated compared to the untreated groups [
      • Cox DR
      Regression models and life-tables.
      ]. The PH assumption was not met so we included an interaction between time and treatment receipt in the Cox model.
      In exploratory subgroup analysis using the unmatched sample, we evaluated the benefit associated with treatment among four subgroups defined by comorbidity (CCI of zero or one/CCI of two or higher) and proxy for poor performance status (yes/no). Within each of the four subgroups, we examined differences between treated and untreated groups using the unadjusted KM survival curves. Also, we conducted Cox regression analyses controlling for all the factors included in the propensity score model, except for the variable defining the subgroup, i.e. CCI or proxy for poor performance status.
      In all analyses, an alpha of p ≤ .05 was used as the threshold for statistical significance. Statistical analyses were conducted using SAS 9.4 (SAS Institute, Cary, NC). This study was approved by the local institutional review board.

      3. Results

      The full study sample included 3705 patients, who met the study inclusion criteria. Overall, 68% (N = 2519) of the sample received FL-directed therapy and 32% (N = 1186) did not (Table 1). The median age of the study sample was 84 (interquartile range [IQR], 81–87) years and the median follow-up time was 2.9 (IQR, 1.2–5.5) years. Patients were predominantly white (92%), and females represented 60% of the study sample. The median follow-up time of the treated group was 3.3 (IQR, 1.4–5.9) years with treatment initiated within a median of 61 days. The most common first-line treatment received was rituximab monotherapy (N = 768, 21%). The median follow-up time of the untreated group was 2.2 (IQR, 0.84–4.6) years, with 70% of this group dying during the follow-up period. The treated and untreated groups showed statistically significant differences in sociodemographic and clinical factors. The treated group had larger proportions of younger patients, males, and married patients. The untreated group had larger percentages of patients with CCI value of two or higher (25% vs. 17%, p < .001) and with proxy indicators of poor performance status (35% vs. 27%, p < .001). The treated group had larger proportions of stage III/IV patients (45% vs. 33%, p < .001) and grade 3 patients (19% vs. 10%, p < .001). Also, the treated group was more likely to utilize preventive services in the baseline period (81% vs. 76%, p < .001).
      Table 1Characteristics of follicular lymphoma patients aged 80 years and older, diagnosed between 2000 and 2013 (N = 3705).
      Patient characteristicsOverallUntreatedTreatedP value
      N (%)
      Percentages may not sum up to 100% due to rounding.
      N (%)
      Percentages may not sum up to 100% due to rounding.
      N (%)
      Percentages may not sum up to 100% due to rounding.
      Sample size (row %)3705 (100%)1186 (32%)2519 (68%)
      Age, mean (SD), year84 (3.6)85 (4.0)84 (3.4)<.001
      Based on two-sample t-test.
      Follow-up time, mean (SD), year3.71 (3.03)3.07 (2.87)4.01 (3.05)<.001
      Based on two-sample t-test.
      Age group<.001
       80–842214 (60)620 (52)1594 (63)
       85–891134 (31)398 (34)736 (29)
       90+357 (10)168 (14)189 (8)
      Gender.018
       Female2209 (60)740 (62)1469 (58)
       Male1496 (40)446 (38)1050 (42)
      Race/ethnicity.091
       Non-Hispanic White3400 (92)1079 (91)2321 (92)
       Non-Hispanic African American67 (2)31 (3)36 (1)
       Hispanic128 (3)42 (4)86 (3)
       Other110 (3)34 (3)76 (3)
      Married<.001
       No1854 (50)645 (54)1209 (48)
       Yes1608 (43)452 (38)1156 (46)
       Missing243 (7)89 (8)154 (6)
      Urban residence.409
       No396 (11)134 (11)262 (10)
       Yes3309 (89)1052 (89)2257 (90)
      State buy-in<.001
       No3378 (91)1048 (88)2330 (93)
       Yes327 (9)138 (12)189 (8)
      Diagnosis period.011
       2000–20051520 (41)458 (39)1062 (42)
       2006–20101481 (40)471 (40)1010 (40)
       2011–2013704 (19)257 (22)447 (18)
      Charlson Comorbidity Index<.001
       02089 (56)599 (51)1490 (59)
       1891 (24)287 (24)604 (24)
       2+725 (20)300 (25)425 (17)
      Ann Arbor stage<.001
       11182 (32)474 (40)708 (28)
       2624 (17)188 (16)436 (17)
       3792 (21)206 (17)586 (23)
       4732 (20)187 (16)545 (22)
       Unknown375 (10)131 (11)244 (10)
      Poor performance status proxy<.001
       No2617 (71)769 (65)1848 (73)
       Yes1088 (29)417 (35)671 (27)
      Preventive services utilization<.001
       No759 (20)285 (24)474 (19)
       Yes2946 (80)901 (76)2045 (81)
      Follicular lymphoma grade<.001
       1–23118 (84)1071 (90)2047 (81)
       3587 (16)115 (10)472 (19)
      Region.291
       Northeast823 (22)251 (21)572 (23)
       Midwest538 (15)189 (16)349 (14)
       South851 (23)278 (23)573 (23)
       West1493 (40)468 (39)1025 (41)
      a Percentages may not sum up to 100% due to rounding.
      b Based on two-sample t-test.
      The propensity score-matched sample included 2306 patients divided evenly between treated (N = 1153) and untreated (N = 1153) individuals. The results of the logistic regression model used for propensity score matching are shown in Supplementary Fig. 1. Within the matched sample, there were no statistically significant differences between the two groups in terms of the sociodemographic or clinical factors used in matching. The characteristics of the matched sample are shown in Supplementary Table 1.
      Fig. 1 illustrates the KM survival curves of the treated and untreated, matched groups. The treated group had a median overall survival of 4.31 years (95% confidence interval [CI], 4.00–4.61), while the untreated group had a median overall survival of 2.86 years (95% CI, 2.59–3.16). There was a statistically significant difference between the two survival curves based on the Wilcoxon (p < .001) and log-rank tests (p < .001). The survival probabilities at years one through five post diagnosis for both groups are shown in Supplementary Table 2. At each year post diagnosis, the treated group had a higher survival probability compared to the untreated group. For example, the 3-year survival probability among the treated group was 0.61 (95% CI, 0.58–0.64) and 0.48 (95% CI, 0.45–0.51) for the untreated group. The 5-year survival probability was 0.43 (95% CI, 0.40–0.46) among the treated, and 0.34 (95% CI, 0.31–0.37) among the untreated. Also, the 3-year RMST for the treated group was 2.36 years (95% CI, 2.30–2.41), while it was 2.05 years (95% CI, 1.98–2.11) for the untreated group. The unadjusted KM survival results within the subgroups defined by comorbidity status and the proxy for poor performance status are shown in Fig. 2. Treatment receipt was associated with improved survival in each subgroup (p < .001 for all log-rank tests).
      The results of the Cox PH models are shown in Table 2. Based on the main model, treatment was associated with a 23% reduction in the hazards of death (hazard ratio [HR], 0.77; 95% CI, 0.70–0.85). When we interacted treatment receipt with time, the baseline HR was 0.59 (95% CI, 0.51–0.69) with an increase of 0.008 for each additional month after diagnosis (95% CI, 0.004–0.011). In each of the four subgroups, treatment was associated with a reduction in the hazard of death (Table 2).
      Fig. 1
      Fig. 1Kaplan-Meier curve of overall survival for a matched sample of patients who received FL-directed therapy and those who did not (N = 2306).
      Abbreviation: FL, follicular lymphoma; CI, confidence interval.
      Fig. 2
      Fig. 2The unadjusted Kaplan-Meier survival curves stratified by treatment receipt among subgroups defined by Charlson Comorbidity Index score and a proxy indicator for poor performance status.
      (A) Patients with Charlson Comorbidity Index score of 0 or 1 (N = 2980).
      (B) Patients with CCI score of 2+ (N = 725).
      (C) Patients with no proxy indicator of poor performance status (N = 2617).
      (D) Patients with a proxy indicator of poor performance status (N = 1088).
      * Per data use agreement, cell sizes less than 11 cannot be reported or derived.
      Table 2The association between treatment receipt and the risk of mortality among the matched sample of treated and untreated FL patients and among subgroups defined by comorbidity and proxy for poor performance status.
      Sample and modelSample sizeHazard ratio95% CIP value
      Full matched sample2306
       Model 1: Assuming proportional hazards
      Treatment receipt0.7680.696–0.848<.001
       Model 2: Assuming non-proportional hazards
      Treatment receipt0.5930.511–0.687<.001
       Treatment receipt*Time (in months)1.0081.004–1.011<.001
      Subgroup analysis by CCI score
      Patients with CCI score of 0 or 1
      Controlling for age, race, sex, marital status, state buy-in, region, Ann Arbor stage, proxy for poor performance status, preventive services utilization in the pre-diagnosis period, diagnosis period.
      2980
       Model 1: Assuming proportional hazards
      Treatment receipt0.7770.703–0.859<.001
       Model 2: Assuming non-proportional hazards
      Treatment receipt0.6060.523–0.702<.001
       Treatment receipt*Time (in months)1.0071.004–1.010<.001
      Patients with CCI score of 2
      Controlling for age, race, sex, marital status, state buy-in, region, Ann Arbor stage, proxy for poor performance status, preventive services utilization in the pre-diagnosis period, diagnosis period.
      725
       Model 1: Assuming proportional hazards
      Treatment receipt0.6800.569–0.813<.001
       Model 2: Assuming non-proportional hazards
      Treatment receipt0.5980.462–0.775<.001
       Treatment receipt*Time (in months)1.0050.998–1.012.185
      Subgroup analysis by proxy for poor performance status
      Patients with no proxy indicators of poor performance status
      Controlling for age, race, sex, marital status, state buy-in, region, Ann Arbor stage, Charlson Comorbidity Index, preventive services utilization in the pre-diagnosis period, diagnosis period.
      2617
       Model 1: Assuming proportional hazards
      Treatment receipt0.8140.730–0.907<.001
       Model 2: Assuming non-proportional hazards
      Treatment receipt0.6590.561–0.775<.001
       Treatment receipt*Time (in months)1.0061.002–1.009<.001
      Patients with proxy indicators of poor performance status
      Controlling for age, race, sex, marital status, state buy-in, region, Ann Arbor stage, Charlson Comorbidity Index, preventive services utilization in the pre-diagnosis period, diagnosis period.
      1088
       Model 1: Assuming proportional hazards
      Treatment receipt0.6500.561–0.753<.001
       Model 2: Assuming non-proportional hazards
      Treatment receipt0.5320.433–0.654<.001
       Treatment receipt*Time (in months)1.0071.002–1.012.008
      Abbreviation: CCI, Charlson Comorbidity Index; CI, confidence interval.
      a Controlling for age, race, sex, marital status, state buy-in, region, Ann Arbor stage, proxy for poor performance status, preventive services utilization in the pre-diagnosis period, diagnosis period.
      b Controlling for age, race, sex, marital status, state buy-in, region, Ann Arbor stage, Charlson Comorbidity Index, preventive services utilization in the pre-diagnosis period, diagnosis period.
      Table 3 illustrates the most frequent initial treatments received by the treated group, stratified by the CCI score and the proxy indicator for poor performance status. Overall, rituximab monotherapy was the most commonly utilized first-line treatment among the treated sample (31%) and within subgroups defined by the CCI score and performance status.
      Table 3The most frequently received treatments within the matched sample, stratified by the Charlson Comorbidity Index score and the proxy indicator for poor performance status (N = 1153).
      Matched treated groupCCI of 0 or 1CCI of 2+No proxy indicator of poor performance statusWith proxy indicator of poor performance status
      N (%)N (%)N (%)N (%)N (%)
      Sample size1153875278758395
      Treatment regimen
       RM359 (31)264 (30)95 (34)227 (30)132 (33)
       Radiation335 (29)258 (29)77 (28)221 (29)114 (29)
       RCVP143 (12)102 (12)41 (15)94 (12)49 (12)
       RCHOP98 (9)84 (10)14 (5)61 (8)37 (9)
       BM-RM51 (4)38 (4)13 (5)39 (5)12 (3)
       CVP42 (4)*
      Per data use agreement, cell sizes <11 cannot be reported or derived.
      *
      Per data use agreement, cell sizes <11 cannot be reported or derived.
      31 (4)11 (3)
      Abbreviations and acronyms: BM, Bendamustine; CCI, Charlson Comorbidity Index; CVP, cyclophosphamide, vincristine and prednisone; RCHOP, rituximab, cyclophosphamide, doxorubicin, vincristine and prednisone; RCVP, rituximab, cyclophosphamide, vincristine and prednisone; RM, rituximab monotherapy.
      a Per data use agreement, cell sizes <11 cannot be reported or derived.

      4. Discussion

      There is limited evidence regarding the treatment benefit among patients diagnosed with FL at 80 years or older. Patients in this age group are likely to present with comorbidities and declined performance status. However, the availability of effective and relatively tolerable therapeutic options warrant an examination of treatment effectiveness in this frail patient group. We evaluated the overall survival of patients who received FL-directed therapy and compared them to patients who did not in a cohort of individuals diagnosed with FL at 80 years or older. After we accounted for measurable differences between the treated and the untreated groups, we found that patients who received FL-directed therapy exhibited better survival compared to patients who did not. We found similar results among clinically important subgroups defined by factors that can complicate the decision to offer treatment, i.e., high comorbidity and poorer physical health status.
      In our sample, one out of three patients did not receive treatment by the end of the study follow-up. Individuals who received treatment differed from those who did not based on demographic and clinical factors. In multivariable analysis, the following factors were inversely associated with treatment receipt: older age, CCI score of two or higher, and the proxy indicator for poor performance status. These factors are expected to discourage physicians from providing FL-directed therapy. We also found that treatment rates were lower among patients with state buy-in, which is a proxy for limited income. On the other hand, treatment receipt was higher among married patients, which may reflect the importance of social support for treatment receipt. Treatment receipt was also higher among individuals with higher lymphoma stage and grade. Interestingly, we found that treatment receipt was higher among patients who utilized preventive services in the year prior to diagnosis. This might suggest that patients engaged in healthy behavior were more likely to receive FL-directed therapy. The most common systemic treatment received by this population was rituximab monotherapy, as expected, given the lower toxicity risk with this agent.
      A few studies in the literature have discussed the importance of studying the treatment patterns and outcomes of FL/NHL patients 80 years and older [
      • Nabhan C
      • Byrtek M
      • Rai A
      • et al.
      Disease characteristics, treatment patterns, prognosis, outcomes and lymphoma-related mortality in elderly follicular lymphoma in the United States.
      ,
      • Nabhan C
      • Smith SM
      • Helenowski I
      • et al.
      Analysis of very elderly (≥ 80 years) non-hodgkin lymphoma: impact of functional status and co-morbidities on outcome.
      ,
      • Thieblemont C
      • Grossoeuvre A
      • Houot R
      • et al.
      Non-Hodgkin’s lymphoma in very elderly patients over 80 years. A descriptive analysis of clinical presentation and outcome.
      ] One of these studies evaluated treatment benefit in this older patient group [
      • Nabhan C
      • Byrtek M
      • Rai A
      • et al.
      Disease characteristics, treatment patterns, prognosis, outcomes and lymphoma-related mortality in elderly follicular lymphoma in the United States.
      ]. The study by Nabhan et al. used data from the National LymphoCare Study, which is a prospective study of FL patients diagnosed between 2004 and 2007 [
      • Nabhan C
      • Byrtek M
      • Rai A
      • et al.
      Disease characteristics, treatment patterns, prognosis, outcomes and lymphoma-related mortality in elderly follicular lymphoma in the United States.
      ]. However, the authors did not find a survival benefit associated with FL treatment among patients older than 80 years [
      • Nabhan C
      • Byrtek M
      • Rai A
      • et al.
      Disease characteristics, treatment patterns, prognosis, outcomes and lymphoma-related mortality in elderly follicular lymphoma in the United States.
      ]. The study included a small sample of patients older than 80 years and the authors acknowledged the limited power to detect differences between the comparison groups among individuals older than 80 years. The large sample size of our study allowed us to evaluate the survival benefit overall and among clinically-important subgroups. The Nabhan et al. study included valuable clinical measures that are not available in claims data, such as lactate dehydrogenase (LDH) and hemoglobin levels, and bone marrow involvement. However, it lacked information on comorbidity, which, according to our results, is an important confounder associated with treatment receipt and survival in this older patient population. Additionally, our study included patients diagnosed over a longer time period compared to the LymphoCare study (i.e., 2000–2013 vs. 2004–2007) [
      • Friedberg JW
      • Taylor MD
      • Cerhan JR
      • et al.
      Follicular lymphoma in the United States: first report of the National LymphoCare study.
      ]. This longer time period allowed us to consider additional treatment options as treatment patterns and guidelines evolved over time.
      There are multiple limitations to our study, many of which are typical for studies using administrative claims data. First, we were not able to identify and control for the Groupe D'Etude des Lymphomes Folliculaires (GELF) criteria [
      • Brice P
      • Bastion Y
      • Lepage E
      • et al.
      Comparison in low-tumor-burden follicular lymphomas between an initial no-treatment policy, prednimustine, or interferon alfa: a randomized study from the Groupe d’Etude des Lymphomes Folliculaires. Groupe d’Etude des Lymphomes de l’Adulte.
      ] or the British National Lymphoma Investigation (BNLI) criteria [
      • Ardeshna K
      • Smith P
      • Norton A
      • et al.
      Long-term effect of a watch and wait policy versus immediate systemic treatment for asymptomatic advanced-stage non-Hodgkin lymphoma: a randomised controlled trial.
      ], which assess tumor burden and indicate if immediate treatment initiation is necessary. Nonetheless, we do not expect that this limitation would change the qualitative conclusion of our study. Holding other factors constant, treated patients were expected to have higher tumor burden, yet they demonstrated improved survival. Second, owing to the lack of laboratory results and the number of nodal areas information, we did not have the FLIPI prognostic score [
      • Solal-Celigny P
      • Roy P
      • Colombat P
      • et al.
      Follicular lymphoma international prognostic index.
      ]. While the data lacked this composite measure, we were able to control for stage and age, which are component measures used to estimate the FLIPI. Third, even though cases and controls were similar on observed demographic and clinical factors after matching, residual confounding might still be present particularly due to the lack of clinical measures (e.g. GELF criteria) and the lack of a direct measure of performance status, such as the Eastern Cooperative Oncology Group (ECOG) performance status. While we lacked these key clinical measures, we controlled for important potential confounders, such as stage and CCI. Also, we included a previously developed proxy for poor performance status, which was expected to reduce selection bias related to treatment receipt and improve covariate control [
      • Davidoff AJ
      • Zuckerman IH
      • Pandya N
      • et al.
      A novel approach to improve health status measurement in observational claims-based studies of cancer treatment and outcomes.
      ]. Lastly, the limited follow-up time of the study might have prevented us from identifying patients who received treatment after the end of the study, especially for patients diagnosed in recent years. However, among the matched sample, the majority of the untreated group (70%) died by the end of the study. Additionally, patients who remained alive (30%) had a relatively long follow-up time (median = 3.7 years) compared to the full sample.
      Overall, we observed that, in a cohort of FL patients aged 80 years or older, FL-directed therapy was associated with an overall survival benefit, which persisted in clinically important subgroups. Our results provide novel information regarding a vulnerable, under-studied population as well as baseline data on treatment effectiveness for future comparisons. With newer FL therapies currently on the market, it is important to conduct studies that examine the survival benefit associated with newer therapies for older adults. Additional evidence regarding the comparative effectiveness of FL therapies would provide important information to support a complex treatment decision for a vulnerable population.
      The following are the supplementary data related to this article

      Author Contribution

      Study concepts: Husam Albarmawi, Jean Yared
      Study design: Husam Albarmawi, Eberechukwu Onukwugha, Karen N. Keating, Sreevalsa Appukkuttan
      Data acquisition: Eberechukwu Onukwugha
      Quality control of data and algorithms: Eberechukwu Onukwugha
      Data analysis and interpretation: Husam Albarmawi, Eberechukwu Onukwugha, Karen N. Keating, Sreevalsa Appukkuttan, Jean Yared
      Statistical analysis: Husam Albarmawi, Eberechukwu Onukwugha
      Manuscript preparation: Husam Albarmawi, Eberechukwu Onukwugha, Jean Yared
      Manuscript editing: Eberechukwu Onukwugha, Karen N. Keating, Sreevalsa Appukkuttan, Jean Yared
      Manuscript review: Husam Albarmawi, Eberechukwu Onukwugha, Karen N. Keating, Sreevalsa Appukkuttan, Jean Yared

      Declaration of Competing Interest

      Mr. Albarmawi, and Dr. Yared declare that they have no conflict of interest; Dr. Onukwugha reports grant funding from Bayer US LLC and Pfizer, Inc. as well as consulting fees from Novo Nordisk; Dr. Keating, Dr. Appukkuttan are employees of Bayer US LLC.

      Acknowledgements

      The study team thanks Teddy Dunning for the research assistance that he provided. Specifically, he conducted a literature review and provided an initial draft of the manuscript Introduction section. We also thank the Pharmaceutical Research Computing staff at the University of Maryland School of Pharmacy for their assistance with preparing the analytic files that we used for the statistical analyses.
      This study used the linked SEER-Medicare database. The interpretation and reporting of these data are the sole responsibility of the authors. The authors acknowledge the efforts of the National Cancer Institute; the Office of Research, Development and Information, CMS; Information Management Services (IMS), Inc.; and the Surveillance, Epidemiology, and End Results (SEER) Program tumor registries in the creation of the SEER-Medicare database. The collection of cancer incidence data used in this study was supported by the California Department of Public Health as part of the statewide cancer reporting program mandated by California Health and Safety Code Section 103885; the National Cancer Institute‘s Surveillance, Epidemiology and End Results Program under contract HHSN261201000140C awarded to the Cancer Prevention Institute of California, contract HHSN261201000035C awarded to the University of Southern California, and contract HHSN261201000034C awarded to the Public Health Institute; and the Centers for Disease Control and Prevention’s National Program of Cancer Registries, under agreement # U58DP003862-01 awarded to the California Department of Public Health. The ideas and opinions expressed herein are those of the author(s) and endorsement by the State of California Department of Public Health, the National Cancer Institute, and the Centers for Disease Control and Prevention or their Contractors and Subcontractors is not intended nor should be inferred.

      Source of Funding

      Financial support for this study was provided by Bayer US LLC under Research Agreement [201603311353]. The funding agreement ensured the authors' independence in designing and conducting the study, collecting, managing, analyzing and interpreting the data, preparing, reviewing and approving of the manuscript, and the decision to submit the manuscript for publication.

      References

        • Ganti AK
        • Bociek RG
        • Bierman PJ
        • Enke CA
        • Vose JM
        • Armitage JO
        Follicular lymphoma: expanding therapeutic options.
        Oncology (Williston Park). 2005; 19 (discussion 228, 233–216, 239): 213-228
        • Swenson WT
        • Wooldridge JE
        • Lynch CF
        • Forman-Hoffman VL
        • Chrischilles E
        • Link BK
        Improved survival of follicular lymphoma patients in the United States.
        J Clin Oncol. 2005; 23: 5019-5026
        • Liu Q
        • Fayad L
        • Cabanillas F
        • et al.
        Improvement of overall and failure-free survival in stage IV follicular lymphoma: 25 years of treatment experience at The University of Texas MD Anderson Cancer Center.
        J Clin Oncol. 2006; 24: 1582-1589
        • Conconi A
        • Motta M
        • Bertoni F
        • et al.
        Patterns of survival of follicular lymphomas at a single institution through three decades.
        Leuk Lymphoma. 2010; 51: 1028-1034
        • Tan D
        • Horning SJ
        • Hoppe RT
        • et al.
        Improvements in observed and relative survival in follicular grade 1-2 lymphoma during 4 decades: the Stanford University experience.
        Blood. 2013; 122: 981-987
        • Network N.C.C.
        NCCN clinical practice guidelines in oncology (NCCN guidelines®) non-Hodgkin's lymphomas, Version 1.2016.
        • Solal-Céligny P
        • Bellei M
        • Marcheselli L
        • et al.
        Watchful waiting in low–tumor burden follicular lymphoma in the rituximab era: results of an F2-study database.
        J Clin Oncol. 2012; 30: 3848-3853
        • Ortman JM
        • Velkoff VA
        • Hogan H
        An aging nation: the older population in the United States.
        United States Census Bureau, Economics and Statistics Administration, US …, 2014
        • Ward EM
        • Ma J
        • Jemal A
        • et al.
        Annual report to the nation on the status of cancer, 1975–2014, featuring survival.
        J Natl Cancer Inst. 2017; 109
        • Thieblemont C
        • Coiffier B
        Lymphoma in older patients.
        J Clin Oncol. 2007; 25: 1916-1923
        • Noone A
        • Howlader N
        • Krapcho M
        • et al.
        SEER cancer statistics review, 1975-2015.
        National Cancer Institute, Bethesda, MD2018
        • Solal-Celigny P
        • Roy P
        • Colombat P
        • et al.
        Follicular lymphoma international prognostic index.
        Blood. 2004; 104: 1258-1265
        • Nabhan C
        • Byrtek M
        • Rai A
        • et al.
        Disease characteristics, treatment patterns, prognosis, outcomes and lymphoma-related mortality in elderly follicular lymphoma in the United States.
        Br J Haematol. 2015; 170: 85-95
        • Wan-Chow-Wah D
        • Monette J
        • Monette M
        • et al.
        Difficulties in decision making regarding chemotherapy for older cancer patients: a census of cancer physicians.
        Crit Rev Oncol Hematol. 2011; 78: 45-58
        • Nabhan C
        • Smith SM
        • Helenowski I
        • et al.
        Analysis of very elderly (≥ 80 years) non-hodgkin lymphoma: impact of functional status and co-morbidities on outcome.
        Br J Haematol. 2012; 156: 196-204
        • Hiddemann W
        • Kneba M
        • Dreyling M
        • et al.
        Frontline therapy with rituximab added to the combination of cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) significantly improves the outcome for patients with advanced-stage follicular lymphoma compared with therapy with CHOP alone: results of a prospective randomized study of the German Low-Grade Lymphoma Study Group.
        Blood. 2005; 106: 3725-3732
        • Pfreundschuh M
        • Trümper L
        • Österborg A
        • et al.
        CHOP-like chemotherapy plus rituximab versus CHOP-like chemotherapy alone in young patients with good-prognosis diffuse large-B-cell lymphoma: a randomised controlled trial by the MabThera International Trial (MInT) Group.
        Lancet Oncol. 2006; 7: 379-391
        • Ghielmini M
        • Schmitz S-FH
        • Cogliatti SB
        • et al.
        Prolonged treatment with rituximab in patients with follicular lymphoma significantly increases event-free survival and response duration compared with the standard weekly× 4 schedule.
        Blood. 2004; 103: 4416-4423
        • Marcus R
        • Imrie K
        • Belch A
        • et al.
        CVP chemotherapy plus rituximab compared with CVP as first-line treatment for advanced follicular lymphoma.
        Blood. 2005; 105: 1417-1423
        • Talarico L
        • Chen G
        • Pazdur R
        Enrollment of elderly patients in clinical trials for cancer drug registration: a 7-year experience by the US Food and Drug Administration.
        J Clin Oncol. 2004; 22: 4626-4631
        • Lewis JH
        • Kilgore ML
        • Goldman DP
        • et al.
        Participation of patients 65 years of age or older in cancer clinical trials.
        J Clin Oncol. 2003; 21: 1383-1389
        • Hutchins LF
        • Unger JM
        • Crowley JJ
        • Coltman CA
        • Albain KS
        Underrepresentation of patients 65 years of age or older in cancer-treatment trials.
        N Engl J Med. 1999; 341: 2061-2067
        • National Cancer Institute
        Overview of the SEER program.
        (Accessed November 4, 2018)
        • Charlson ME
        • Pompei P
        • Ales KL
        • MacKenzie CR
        A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.
        J Chronic Dis. 1987; 40: 373-383
        • Deyo RA
        • Cherkin DC
        • Ciol MA
        Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases.
        J Clin Epidemiol. 1992; 45: 613-619
        • Davidoff AJ
        • Zuckerman IH
        • Pandya N
        • et al.
        A novel approach to improve health status measurement in observational claims-based studies of cancer treatment and outcomes.
        J Geriatric Oncol. 2013; 4: 157-165
        • Shrank WH
        • Patrick AR
        • Brookhart MA
        Healthy user and related biases in observational studies of preventive interventions: a primer for physicians.
        J Gen Intern Med. 2011; 26: 546-550
        • Austin PC
        An introduction to propensity score methods for reducing the effects of confounding in observational studies.
        Multivar Behav Res. 2011; 46: 399-424
        • Austin PC
        • Grootendorst P
        • Anderson GM
        A comparison of the ability of different propensity score models to balance measured variables between treated and untreated subjects: a Monte Carlo study.
        Stat Med. 2007; 26: 734-753
        • Parsons L.
        Reducing bias in a propensity score matched-pair sample using greedy matching techniques.
        (Paper presented at: Proceedings of the twenty-sixth Annual SAS users group international conference 2001)2001
        • Kaplan EL
        • Meier P
        Nonparametric estimation from incomplete observations.
        J Am Stat Assoc. 1958; 53: 457-481
        • Royston P
        • Parmar MK
        Restricted mean survival time: an alternative to the hazard ratio for the design and analysis of randomized trials with a time-to-event outcome.
        BMC Med Res Methodol. 2013; 13: 152
        • Cox DR
        Regression models and life-tables.
        in: Breakthroughs in statistics. Springer, 1992: 527-541
        • Thieblemont C
        • Grossoeuvre A
        • Houot R
        • et al.
        Non-Hodgkin’s lymphoma in very elderly patients over 80 years. A descriptive analysis of clinical presentation and outcome.
        Ann Oncol. 2008; 19: 774-779
        • Friedberg JW
        • Taylor MD
        • Cerhan JR
        • et al.
        Follicular lymphoma in the United States: first report of the National LymphoCare study.
        J Clin Oncol. 2009; 27: 1202-1208
        • Brice P
        • Bastion Y
        • Lepage E
        • et al.
        Comparison in low-tumor-burden follicular lymphomas between an initial no-treatment policy, prednimustine, or interferon alfa: a randomized study from the Groupe d’Etude des Lymphomes Folliculaires. Groupe d’Etude des Lymphomes de l’Adulte.
        J Clin Oncol. 1997; 15: 1110-1117
        • Ardeshna K
        • Smith P
        • Norton A
        • et al.
        Long-term effect of a watch and wait policy versus immediate systemic treatment for asymptomatic advanced-stage non-Hodgkin lymphoma: a randomised controlled trial.
        The Lancet. 2003; 362: 516-522