Persistent Disparities Among Patients With T-Cell Non-Hodgkin Lymphomas and B-Cell Diffuse Large Cell Lymphomas Over 40 Years: A SEER Database Review

Abstract

Our population-based analysis for T-cell and B-cell non-Hodgkin lymphoma (NHL) in the novel therapeutic agent era showed worse median overall survival (OS) for male subjects, older patients, and ethnic minorities for both T-cell and B-cell NHL. Despite novel therapeutics targeting T cells, there has been no statistically significant improvement in OS over time, in contrast to the significant improvement in B-cell NHL OS, presumably as a result of the advent of monoclonal antibodies.

Background:

As of 2013, more than 550,000 people are living with non-Hodgkin lymphoma (NHL).

Patients and Methods:

We undertook a large Surveillance Epidemiology and End Results (SEER) based analysis to describe outcome disparities in different subgroups of aggressive T-cell and B-cell NHL patients, with a focus on various ethnicities.

Results:

The final analysis included 7662 patients with T-cell and 84,910 with B-cell NHL. Survival analysis revealed that male sex and increasing age were independent predictors of worse overall survival (OS; P < .001). For aggressive T-cell NHL, there was no significant improvement in median OS between 1973 and 2011 (P = .081), and ethnic minorities (Asians, Hispanics, and African Americans) had significantly worse OS than whites (P < .001). There were similar trends for age, sex, and race for diffuse large B-cell NHL, but a significant improvement in median OS was seen over time (P < .001).

Conclusion:

These results are the first to elicit outcomes in a broad classification of ethnic minorities and underscore the urgency for development of novel therapeutics, especially in T-cell NHL. In addition, in-depth studies of disease biology and health care utilization are required for better triage of health care resources, especially for ethnic minorities.

Introduction

Non-Hodgkin lymphoma (NHL) encompasses a diverse group of malignant neoplasms derived from B cells, T cells, or natural killer cells. It is estimated that more than 558,000 individuals in the United States are living with NHL in 2013 and the incidence of NHL is increasing among men.^1,2 Age, sex, and race/ethnicity have been shown to affect overall survival (OS) among patients with multiple myeloma and different lymphoid malignancies, including chronic lymphocytic leukemia and follicular B-cell NHL.^3–5 Previous studies utilizing the Surveillance, Epidemiology, and End Results (SEER) database have focused on a specific NHL, such as diffuse large B-cell lymphoma.^6–8 The few published studies evaluating outcome disparities in aggressive NHL are of limited relevance to current patients in view of significant changes in the standard treatments of these disorders over the last few decades.^9–13 None of the studies has evaluated disparities in outcomes of patients with T-cell NHL.

In the 1980s, combination chemotherapy with cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP regimen) was the standard treatment for NHL.^14,15 The addition of rituximab to CHOP in 1997 led to an improvement in OS in patients with B-cell NHL.¹⁶ New effective therapies for T-cell NHL are currently available, including romidepsin and pralatrexate. The impact of these new agents on patient outcomes, particularly among different population subgroups, remains to be fully characterized.¹⁷ We undertook a Surveillance Epidemiology and End Results (SEER)-based analysis to describe outcome disparities among different subgroups of aggressive T-cell and B-cell diffuse large cell lymphoma (DLCL) patients, with a focus on different racial/ethnic subgroups.

Patients and Methods

Data Source

The SEER public database 1973 to 2011 (SEER 18), based on the November 2011 submission, was used for the analysis (http://www.seer.cancer.gov).

Study Population

All cases of primary T-cell NHL and B-cell DLCL reported to the SEER cancer registry were evaluated. The following cases were excluded: if diagnosis of NHL was made at death certificate or autopsy, if patient’s age was less than 18, if there were no follow-up records, or if there was no documentation of age at diagnosis, sex, or race/ethnicity.

Variable Definitions

The primary end point was an analysis of differences in OS, defined as the period of time from lymphoma diagnosis to death, among patients with T-cell NHL or B-cell DLCL from different races and ethnicities. For patients who died from a cause other than lymphoma, the data were censored at the time of their death. For patients who were still alive, OS was censored at whichever came first, their last follow-up or December 31, 2011. Data collected from the SEER databases included registry identification, age, race/ethnicity, and year of diagnosis. Mutually exclusive race/ethnicity categories were defined as white, African American, Asian, Hispanic, and Native American. On the basis of the year of diagnosis, patients were stratified into 4 cohorts (1973–1991, 1992–2001, 2002–2006, and 2007–2011) to study the impact of the evolution of NHL therapy. For T-cell NHL, the subtype information was collected using the International Classification of Diseases (ICD)-0–3 codes. The subtypes included in our analysis were peripheral T-cell lymphoma not otherwise specified (NOS; 9702/3), angioimmunoblastic T-cell lymphoma (9705/3), anaplastic large cell lymphoma (ALCL; 9714/3), and hepatosplenic gamma-delta cell lymphoma (9716/3). Only cases of B-cell DLCL were included in the analysis among the B-cell lymphoma cases (9680/3).

Statistical Analysis

Cox proportional hazard models were used to evaluate the association between patient characteristics and OS. All multivariate models included SEER registry ID as a stratification variable. Survival time was censored at 10 years for all analyses. Hazard ratios (HR) and 95% confidence intervals (CI) were generated, with an HR of < 1.0 indicating reduced mortality. Pairwise interactions were examined using stratified models and were tested by comparing corresponding likelihood ratio statistics between the baseline and nested Cox proportional hazard models that included the multiplicative product terms. Departure of the proportional hazard assumption of Cox models was examined graphically using log–log survival curves and smoothed plots of weighted Schoenfeld residuals.¹⁸ For the analysis of survival disparities among various racial/ethnic subgroups of NHL patients, whites were used as the reference. All statistical tests were 2 sided and utilized SAS v9.4 software (SAS Institute) with a 2-sided significance level of .05.

Results

The final analysis included 7662 patients with T-cell NHL (58.9% male) and 84,910 patients with B-cell DLCL (53.7% male) as per the inclusion criteria described above. Patient characteristics are shown in Table 1. The percentage of patients with T-cell NHL and B-cell DLCL who were 65 or older was 44.5% and 54.2%, respectively. Racial/ethnic minorities (African American, Hispanic, Asian, and Native American) comprised 31.3% and 22.9% in T-cell and B-cell DLCL patients, respectively. Within the T-cell NHL group, T-cell lymphoma NOS was the most common subtype (51.3%). Median age at diagnosis for T-cell NHL patients showed significant variability, with African Americans and Hispanics having the youngest (54 years) and Asians the oldest (65 years) median age at diagnosis (Figure 1).

Table 1.

Patient Characteristics

Characteristic	T-Cell NHL (N = 7662)	B-Cell NHLa (N = 84,910)
Sex
Male	4509 (58.9%)	45,559 (53.7%)
Female	3153 (41.2%)	39,351 (46.3%)
Age
18–44 years	1562 (20.4%)	12,084 (14.2%)
45–54 years	1161 (15.2%)	11,183 (13.2%)
55–64 years	1526 (19.9%)	15,661 (18.4%)
65–74 years	1625 (21.2%)	20,120 (23.7%)
≥75 years	1788 (23.3%)	25,862 (30.5%)
Race
White	5261 (68.7%)	65,424 (77.1%)
African American	968 (12.6%)	5490 (6.5%)
Asian	592 (7.7%)	5612 (6.6%)
Hispanic	800 (10.4%)	8048 (9.5%)
Native American	41 (0.5%)	336 (0.4%)
Year of Diagnosis
1973–1991	278 (3.6%)	13,027 (15.3%)
1992–2001	2008 (26.2%)	22,507 (26.5%)
2002–2006	2591 (33.8%)	23,698 (27.9%)
2007–2011	2785 (36.4%)	25,678 (30.2%)
Subtype (T-Cell NHL Only)
Peripheral T-cell lymphoma, NOS	3930 (51.3%)
Angioimmunoblastic T-cell lymphoma	1202 (15.7%)
Anaplastic large-cell lymphoma	2469 (32.2%)
Hepatosplenic gamma-delta-cell lymphoma	61 (0.8%)

Abbreviations: NHL = non-Hodgkin lymphoma; NOS = not otherwise specified; SEER = Surveillance, Epidemiology, and End Results.

^aRepresenting diffuse large B-cell NHL as the aggressive B-cell subtype.

Figure 1.

Median Age at Diagnosis for Patients With Aggressive T-Cell NHL by Race/Ethnicity

Abbreviation: NHL = non-Hodgkin lymphoma.

Survival Analysis: Sex and Age

Survival analysis revealed that female patients had a better median OS than male subjects for T-cell NHL (3.3 years vs. 2.3 years; HR, 0.876; 95% CI, 0.824, 0.932; P < .001) and B-cell DLCL (6.4 years vs. 4.2 years; HR, 0.808; 95% CI, 0.793, 0.824; P < .001; Figure 2A and Figure 3A). Increasing age was a predictor of worse OS independent of sex, race, or year of diagnosis (Figure 2B and Figure 3B). T-cell NHL patients aged 18 to 44 years had a median OS of 9.4 years (HR, 0.345; 95% CI, 0.312, 0.382) compared to patients aged ≥ 75 years with a median OS of 1.0 years (P < .001). A similar difference was seen in the B-cell DLCL as well, with median OS of 10+ years in 18 to 44 years age group versus 1.4 years in the ≥ 75 year age group (P < .001). Of note, T-cell NHL patients had worse OS compared to B-cell NHL patients for each of the age cohorts, and the absolute difference was most pronounced in the 55-to-64-year group (2.9 years vs. 8.1 years).

Figure 2.

(A) Overall Survival for T-Cell NHL by Sex, (B) Age at Diagnosis, (C) Patient Race/Ethnicity Subset, (D) Year of Diagnosis, and (E) Histologic Subtype

Abbreviation: NHL = non-Hodgkin lymphoma.

Figure 3.

(A) Overall Survival for B-Cell DLCL by Sex, (B) Age at Diagnosis, (C) Patient Race/Ethnicity, and (D) Year of Diagnosis

Abbreviation: DLCL = diffuse large cell lymphoma.

Survival Analysis: Race/Ethnicity

A significant interaction was noted between patient race and median OS (P < .001) for both NHL subgroups. In T-cell NHL, multivariate analysis showed African American patients had the worst median OS compared to whites (reference group; 1.7 years vs. 3.1 years, HR, 1.292; 95% CI, 1.176, 1.419; P < .001). All the other nonwhite races had an inferior median OS compared to whites as well (Table 2, Figure 2C). Similarly, in B-cell DLCL, whites had the best median OS of all races (5.8 years), and among all the nonwhite races, Hispanics had the worst median OS (2.8 years; Table 2, Figure 3C). For each of the racial subgroups, patients with B-cell DLCL had a superior outcome compared to T-cell NHL patients, although the magnitude of difference noted was largest for Native Americans and smallest for Asians.

Table 2.

Outcomes of Patients With T-Cell NHL and B-Cell NHL by Demographic Characteristics, SEER Data, 1973–2011

Characteristic	T-Cell NHL (N = 7662)		B-Cell NHLa (N = 84,910)
	Median (95% CI) (Years)b	HR (95% CI)c	Median (95% CI) (Years)b	HR (95% CI)c
Sex
Male	2.3 (2.1, 2.5)	1 (Reference)	4.2 (4.0, 4.3)	1 (Reference)
Female	3.3 (3.0, 3.8)	0.876 (0.824, 0.932)	6.4 (6.3, 6.7)	0.808 (0.793, 0.824)
P valueb		<.001		<.001
Age
18–44 years	9.4 (8.0, 10.0+)	0.345 (0.312, 0.382)	10+	0.309 (0.298, 0.320)
45–54 years	5.9 (4.9, 7.2)	0.432 (0.390, 0.478)	10+	0.376 (0.363, 0.389)
55–64 years	2.9 (2.5, 3.3)	0.583 (0.535, 0.636)	8.1 (7.8, 8.4)	0.466 (0.453, 0.480)
65–74 years	1.9 (1.7, 2.2)	0.705 (0.650, 0.764)	4.6 (4.3, 4.8)	0.614 (0.599, 0.629)
≥75 years	1.0 (0.9, 1.1)	1 (Reference)	1.4 (1.4, 1.5)	1 (Reference)
P valueb		<.001		<.001
Race
White	3.1 (2.8, 3.3)	1 (Reference)	5.8 (5.7, 6.0)	1 (Reference)
African American	1.7 (1.4, 2.1)	1.292 (1.176, 1.419)	3.3 (3.0, 3.6)	1.261 (1.212, 1.312)
Asian	2.6 (1.9, 3.8)	1.109 (0.968, 1.271)	2.9 (2.7, 3.3)	1.031 (0.983, 1.081)
Hispanic	1.9 (1.6, 2.4)	1.215 (1.087, 1.357)	2.8 (2.6, 3.1)	1.154 (1.111, 1.198)
Native American	2.5 (1.1, 6.8)	0.836 (0.528, 1.324)	4.4 (3.2, 6.0)	1.210 (1.034, 1.414)
P valueb		<.001		<.001
Year of Diagnosis
1973–1991	3.8 (2.7, 5.3)	0.816 (0.697, 0.956)	2.4 (2.3, 2.6)	1.098 (1.068, 1.129)
1992–2001	2.6 (2.3, 2.9)	1 (Reference)	3.9 (3.8, 4.2)	1 (Reference)
2002–2006	2.5 (2.3, 2.8)	0.998 (0.926, 1.075)	6.4 (6.3, 6.7)	0.841 (0.820, 0.863)
2007–2011	2.8 (2.4, 3.3)	0.978 (0.901, 1.061)	7.0 (6.8, 7.3)	0.846 (0.822, 0.870)
P valueb		.081		<.001
Subtype
Mature T cell	2.2 (1.9, 2.3)	1 (Reference)
Angioimmunoblastic T cell	3.0 (2.5, 3.5)	0.865 (0.794, 0.943)
Anaplastic large cell	3.8 (3.3, 4.4)	0.766 (0.714, 0.822)
Hepatosplenic gamma-delta cell	0.7 (0.5, 1.1)	1.959 (1.424, 2.695)
P value		<.001

Abbreviations: CI = confidence interval; HR = hazard ratio; NHL = non-Hodgkin lymphoma; SEER = Surveillance, Epidemiology, and End Results.

^aRepresenting diffuse large B-cell NHL as aggressive B-cell subtype.

^bAdjusted median overall survival was calculated based on nonstratified Cox regression model including all variables as covariates.¹⁹

^cBased on Cox proportional hazard model including all variables in the table and marital status (yes vs. no), US born, first malignancy, and stratified by SEER registries.

Survival Analysis: Year of Diagnosis

To explore the effect of any improvement in therapeutic strategies and their beneficial impact on patient outcomes, median OS was analyzed for patients diagnosed over successive time periods (1973–1991, 1992–2001, 2002–2006, 2007–2011) for both NHL subgroups. Smaller time periods were used for more recent years in order to capture the effects of more recently approved treatments. For T-cell NHL, no change in median OS was found over the 38-year span that was evaluated (P = .081; Table 2, Figure 2D). On the other hand, for B-cell DLCL, the median OS significantly improved over time from 2.4 years in 1973 to 1991 to 7 years in 2007 to 2011 (P < .001; Table 2, Figure 3D).

The SEER database was interrogated for the different subtypes of T-cell NHL. This analysis revealed that the best median OS was observed in patients with ALCL (3.8 years, HR, 0.766; CI 95% 0.714, 0.822) compared to the median OS in patients with T-cell lymphoma NOS (reference, median OS of 2.2 years). Patients with hepatosplenic gamma-delta T-cell subtype had the worst median OS of 0.7 years (HR, 1.959; 1.424, 2.695; Table 2, Figure 2E).

Discussion

To our knowledge, this is the largest and most comprehensive evaluation of outcomes in patients belonging to different racial subgroups, including minorities, with T-cell NHL and B-cell DLCL in the United States.

In our analysis, we found that women had a statistically significant better median OS compared to men in T-cell NHL (3.3 years vs. 2.3 years, P < .001) and B-cell DLCL (6.4 years vs. 4.2 years, P < .001). This improved survival by multivariate analysis was independent of race/ethnicity, year of diagnosis, or NHL subtype (P < .001). This finding is consistent with previous reports in NHL as well as other cancers.^20–22 The improved survival in women is thought to be multifactorial, with hypotheses including earlier time of diagnosis and fewer comorbid conditions in women.^20,22 Alcohol consumption and tobacco use have been associated with worse prognosis in lymphoid malignancies.^22,23

For hematologic malignancies, advanced age (defined as ≥ 60 years) has been identified as a poor prognostic factor and has been included in the International Prognostic Index to help risk stratification and to determine optimal therapeutic options for patients with certain NHL subtypes.²⁴ Consistent with the International Prognostic Index criteria and previous reports, the median OS for T-cell and B-cell DLCL patients in our study decreased significantly with increasing age.^21,25 This finding was most pronounced when comparing T-cell patients aged 18 to 44 years with a median OS of 9.4 years compared to patients aged ≥ 75 years with a median OS of 1.0 years (P < .001). This was true of B-cell DLCL patients as well, where those 18 to 44 and 45 to 54 years of age had a median OS of 10+ years compared to 1.4 years for patients aged ≥ 75 years (P < .001). Inclusion of monoclonal antibodies (rituximab) in B-cell NHL treatment seems to have made a difference in patient outcomes and has been shown to be safe and effective in older patients as well.^16,26,27 Still, on the basis of our analyses, it seems that the magnitude of benefit, even in B-cell NHL, for elderly patients is significantly less and requires further investigation into specific therapeutic strategies in prospective clinical trials.

Few studies on the impact of race on outcomes among patients with hematologic malignancies have been completed. These studies have focused on white and African American patients.^10,13,28 Our study compared the outcomes of patients with NHL from 5 different racial/ethnic groups, including white, African American, Asian, Hispanic, and Native American. Our data showed that white patients had significantly longer median OS compared to all other races for T-cell and B-cell DLCL, with the exception of Native American patients with T-cell NHL, where the difference did not reach statistical significance (P < .001). The differences in OS in T-cell NHL with an OS of 3.1 years among whites compared to an OS of 1.7 years among African American patients (P < .001). Among B-cell DLCL patients, the most pronounced difference in median OS was between whites (5.8 years) and Hispanics (2.8 years) (P < .001). The median OS values among nonwhite patients with B-cell DLCL were fairly similar for Hispanic patients (2.8 years), Asian patients (2.9 years), and African American patients (3.3 years), but all were significantly shorter than white patients (5.8 years; P < .001). Another finding of interest regarding race/ethnicity in T-cell NHL was that Asian and white patients were significantly older (65 years and 64 years, respectively) than Hispanic (56 years) and African American (54 years) patients at the time of diagnosis. The reason for this difference may lie in disease biology. A similar significant disparity between median ages at diagnosis has been noted for racial/ethnic minorities and whites in other lymphoid malignancies as well.^3,29 An advantage of our analysis is the utilization of more stratified racial subgroups, which is a better representation of the current racial and ethnic makeup of the United States and shows significant differences in the demographic profile and outcomes in T-cell NHL and B-cell NHL.

The reasons behind these disparities in outcomes may be complex and multifactorial and could plausibly include differences in access to health care. Independent of access issues, studies have shown that nonwhite patients seek medical care less often than white patients.^30–33 This delayed treatment could affect patient outcomes. In addition to race, the effect of socioeconomic status has also been explored. Lower socioeconomic status has been shown to be associated with lower access to and worse quality of integrated treatment which contribute to poor long-term follow-up cancer care.^34–37 This contribution of differences in access to care and disease biology on outcome was evaluated in a study of NHL patients by Pulte and colleagues.²¹ In this study, survival was lower in African American patients than whites, and the disparities in outcomes were more pronounced among patients treated more recently compared to patients treated in previous decades. This study included all patients with NHL, and it should be noted that African American patients were more likely to be diagnosed with T-cell NHL than whites, which may contribute to the differences in outcomes.³⁸ However, this difference cannot be entirely explained by the T-cell NHL diagnosis as shown in our data with the differences present in other nonwhite racial subgroups and in patients with B-cell NHL as well. Ongoing studies are needed to further delineate the etiologies of these outcome differences and implement strategies to reduce these significant disparities.

Our analysis showed a significant improvement in median OS for patients with B-cell DLCL over time. Patients diagnosed with B-cell DLCL during the 1973–1991 period had a median OS of 2.4 years compared to patients diagnosed between 2007 and 2011 who had a median OS of 7.0 years (P < .001). Differences in survival were also evident when comparing patients diagnosed in 1992–2001 to those diagnosed in 2002–2006 where median OS changed from 3.9 years to 6.4 years, respectively (P < .001). The US Food and Drug Administration (FDA) approval of rituximab in 1997 significantly improved survival in patients with B-cell NHL.^16,39,40 Prior studies have reported on the disparate effect of rituximab treatment in racial/ethnic minorities, but have been limited as a result of short follow-up and only African American patients making up the nonwhite cohort. Shah et al¹¹ evaluated 1- and 3-year relative survival in patients with advanced B-cell DLCL. They evaluated patients treated in 2002–2005 before rituximab was approved as a first-line treatment (pre-rituximab) and compared them to patients treated in 2006–2009 (post-rituximab). African American patients, as well as young male patients, showed no improvement in relative survival rates.

For patients with T-cell NHL the median OS remains poor and there has been no significant improvement over the last 38 years. T-cell NHL is a heterogenous group of lymphomas that are much less common than B-cell DLCL.⁴¹ The T-cell subtype ALCL has been shown to have the best median OS of 3.8 years. A meta-analysis of front-line anthracycline chemotherapy in T-cell NHL showed that patients with ALCL have the best response profile with a complete response rate of 65.8%.⁴² In a phase 2 study of 58 patients with relapsed ALCL, novel therapies, including brentuximab–vedotin, a CD30 antibody–drug conjugate, showed promise, with an 86% objective response rate and a 57% complete response with a median duration of 12.6 months and 13.2 months, respectively.⁴³ As a result of the promising results of this study, the FDA granted accelerated approval to brentuximab–vedotin for the treatment of systemic ALCL after failure of at least 1 prior multiagent chemotherapy regimen.⁴⁴ A phase 3 study, ECHELON-2 (NCT01777152), is ongoing.⁴⁵

The major strength of our study is the use of a large population-based data set that reflects the racial/ethnic makeup of the current United States population. The number of patients and the racial/ethnic subgroups included is far larger and more up to date than previous studies. The inclusion of patients with T-cell and B-cell NHL highlights the important similarities, such as outcome disparities among nonwhite patients. It also highlights important differences, such as the significant improvement in median OS in B-cell NHL since the introduction of rituximab as a novel therapy versus the lack of any improvement in T-cell NHL outcomes over time. One of the limitations of SEER-based analysis is the lack of information on clinical data and treatment or treatments provided.

Conclusion

Despite these limitations, our study shows that significant disparities in outcomes exist among racial/ethnic minorities with B-cell DLCL and T-cell lymphomas. These disparities have persisted despite the development of highly effective treatment for B-cell DLCL (rituximab) over the last 20 years. Future studies, especially conducted prospectively, are required to determine the exact causes of these disparities and to formulate strategies to ameliorate them.

Clinical Practice Points.

• NHL encompasses a diverse group of malignant neoplasms affecting approximately 558,000 people in the United States.

• Survival disparities among different racial/ethnic groups has been seen in multiple myeloma and different lymphoid malignancies, including chronic lymphocytic leukemia and follicular B-cell NHL.

• Since 1973 until at least 2011, there has been no improvement in the OS of patients with T-cell NHL.

• Ethnic and racial minorities had worse median OS for both T-cell and B-cell NHL.

• There is a need for in-depth studies of disease biology and health care utilization for better triage of health care resources, especially for ethnic minorities.

• There is a need for novel therapies in the treatment of T-cell lymphomas.