Abstract
Background
Mental health disorders may decrease quality of life and survival in patients with cancer. Little is known about the survival implications of mental health disorders in patients with diffuse large B-cell lymphoma (DLBCL). We aimed to evaluate the impact of pre-existing depression and/or anxiety on survival in a cohort of older DLBCL patients.
Methods
Using the Surveillance, Epidemiology, and End Results-Medicare database, we identified patients ≥67 years old diagnosed with DLBCL in the United States between 01/01/2001 and 12/31/2013. We used billing claims to identify patients with pre-existing depression and/or anxiety prior to their DLBCL diagnosis. We compared 5-year overall survival (OS) and lymphoma-specific survival (LSS) between these patients and those without pre-existing mental disorders with Cox proportional analyses adjusting for sociodemographic and clinical characteristics, including DLBCL stage, extranodal disease, and B symptoms.
Findings
Among 13,244 patients with DLBCL, 15.8% had depression and/or anxiety, 52.8% was female, and 94.1% was White. The median follow-up for the cohort was 2.0 years (IQR 0.4–6.9 years). Patients with pre-existing mental health disorders had inferior 5-year OS. While survival differences between mental health disorders were modest, those with depression had the worst survival (HR 1.37, 95%CI 1.27–1.47) followed by those with depression and anxiety (HR 1.23, 95% CI 1.08–1.41) and anxiety alone (HR 1.17, 95%CI 1.06–1.29). Individuals with pre-existing mental health disorders also had lower 5-year LSS, with depression conferring the greatest effect (HR 1.37, 95%CI 1.28–1.47).
Interpretation
Pre-existing depression and/or anxiety within 24 months prior to DLBCL diagnosis worsens prognosis for patients with DLBCL. Our data underscore the need for universal and systematic mental health screening for this population, as mental health disorders are manageable and improvements in this prevalent comorbidity may affect LSS and OS.
INTRODUCTION
Diffuse large B-cell lymphoma (DLBCL) is the most common lymphoma subtype, with an incidence of approximately 6 cases per 100,000 per year.1 It is an aggressive hematologic malignancy that most commonly presents in older adults with a median age of 66 years at diagnosis.1 Several factors—including stage, age, presence of extranodal disease, and performance status—are known to impact survival;2 however, the majority of these prognostic factors are non-modifiable. In contrast, comorbidities such as underlying mental health disorders are both potentially prognostic and modifiable and thus warrant closer investigation in the context of this deadly blood cancer.3
In patients with solid malignancies, mental health disorders have been associated with lower quality of life as well as decreased overall and cancer-specific survival.4–7 Physiologic factors, such as immune dysregulation, and delays in treatment have been suggested as potential links between mental health disorders and mortality in this population.5,8 Moreover, improvement in depression has been associated with prolonged survival.8 In contrast, the implications of mental health disorders for patients with hematologic malignancies are less well understood. The psychological burden of hematologic malignancies is high, and this can conspire with pre-existing mental illness to increase psychiatric comorbidity. While recent literature has highlighted the increased prevalence of mental health disorders in this population,9–11 only few and mostly small-scale studies have examined the survival implications of mental health disorders in patients with blood cancers.12–17
We sought to better understand the impact that pre-existing mental health disorders may have on the survival of patients with DLBCL. Given existing literature on solid malignancies and the psychosocial stressors of having an aggressive lymphoma, we hypothesized that underlying depression and/or anxiety would be associated with decreased overall survival (OS) and lymphoma-specific survival (LSS).5–7
METHODS
Study design, datasets, and participants
We conducted a retrospective cohort study using the Surveillance, Epidemiology, and End Results cancer registry linked to Medicare claims (SEER-Medicare, 2020 linkage). SEER collects demographic, clinical, cancer-related treatment, and survival data from 20 population-based cancer registries throughout the United States, representing approximately 48% of the US population. The sociodemographic characteristics of patients in the SEER registries are similar to the general population of the United States.18 Medicare is a federally-funded health insurance program in the United States that provides coverage for the vast majority (~ 94%) of individuals ≥ 65 years. A few individuals younger than 65 years are also eligible for Medicare if they have end-stage renal disease (ESRD) or long-term disability. These individuals are not considered representative of the younger than 65-year-old general population and their patterns of cancer care differ significantly from the general population. Accordingly, most SEER-Medicare analyses do not typically include these individuals.19 Medicare data includes billing claims for inpatient care, outpatient visits, physician services, and durable medical equipment across the healthcare spectrum.19 For the linked SEER-Medicare database, approximately 95% of individuals 65 years and older in the SEER registries are matched to their corresponding Medicare data.18 The Dana-Farber/Harvard Cancer Center Office for Human Research Studies reviewed our study and approved it as nonhuman subject research.
The study cohort included patients diagnosed with DLBCL as their only cancer diagnosis from January 1, 2001 to December 31, 2013. Diagnosis was determined from claims data using the International Classification of Disease for Oncology, 3rd edition (ICD-O-3) codes for DLBCL (9684/3, 9680/3, and 9688/3). We focused exclusively on patients with DLBCL because inclusion of all hematologic malignancies would introduce high levels of heterogeneity that could substantially confound our analysis. We required patients to be at least 67 years of age at time of DLBCL diagnosis to ensure a minimum of two years of Medicare enrollment to identify claims for depression and anxiety prior to DLBCL diagnosis. There is precedence of using this 24-month look-back period in SEER-Medicare.20 We excluded patients diagnosed with DLBCL at time of death or autopsy, those who were enrolled in Medicare for ESRD or long-term disability, and those who had another malignancy concurrently. We also excluded those without continuous enrollment in Medicare Parts A and B or those with any health maintenance organization enrollment 24 months prior to DLBCL diagnosis to time of death to ensure that we had complete claims data capturing all their healthcare utilization. This exclusion was necessary as missing data on depression and/or anxiety could lead to misclassification and threaten validity of this analysis. SEER included three new registries (New York, Massachusetts, and Idaho) in its most recent data linkage. Data from these registries are still being harmonized with SEER-Medicare and currently lack relevant data for this study, such as tumor stage and other patient demographics; as a result, we excluded these registries from our analysis. Our cohort assembly is illustrated in Figure 1.
Figure 1:
Cohort Assembly
Outcomes
Survival:
The primary outcomes were OS and LSS. Survival time was tracked from date of DLBCL diagnosis until the end of December 2018. We used the SEER database to determine the date and cause of death. We used SEER data to determine whether each patient died from DLBCL, another cause, or was alive at the end of the survival follow-up period.
Mental health disorders:
The key independent variables (i.e., exposure variables) were pre-existing depression or anxiety. We focused on depression and anxiety as these are relatively common mental health disorders in the general population. We defined a diagnosis of pre-existing depression or anxiety as at least one inpatient or two outpatient claims in the 24 months prior to DLBCL diagnosis. This time frame was based on precedent in the literature.20 Claims were based on the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes (appendix p 1).
Covariates:
We incorporated the following demographic variables from the SEER database into our analysis: age (67–69, 70–74, 75–79, 80+ years), sex, race (White, Non-White), ethnicity, marital status (Married, Unmarried, Unknown), urban residence (Metro, Urban, Rural), SEER Registry region (Northeast, South, Midwest, West), income, and education. Income and education were based on where individuals resided at time of their DLBCL diagnosis (census tract or the US postal code [ZIP code]) and not individual-level income.19 Income was categorized based on the census tract poverty indicator, which categorizes median household income into percentages below the federal poverty level based on the census tract of the diagnosis address. We dichotomized education into one of two groups: those with less than 25% of people in a census tract below high school equivalency and those with 25% or greater of the people in a census tract below high school equivalency. When census tract data was unavailable, we used ZIP code data to determine education. We also obtained disease characteristics including year of DLBCL diagnosis, disease stage, presence of extranodal disease, B symptoms (i.e., presence of fevers, night sweats, or weight loss), and comorbidity. Of note, data regarding B symptoms was not collected in the SEER-Medicare database prior to 2004. We assessed comorbidity using the Deyo adaptation of the Charlson comorbidity index, which calculates scores based on ICD-9 diagnosis and procedure codes during one-year prior to cancer diagnosis.21 Scores were categorized into 0, 1, and 2+. We recategorized patients with unknown race into the Non-White subgroup; this comprised only 0.4% of our cohort.
Statistical analysis
We described the distribution of patients with DLBCL with pre-existing mental disorders (depression and/or anxiety) or without pre-existing mental disorders by their demographic characteristics and disease characteristics. We used chi-squared tests to assess for differences in variables between the groups. We then assessed OS and LSS with the Kaplan-Meier method and compared these outcomes for patients in 4 categories: (1) no pre-existing mental illness, (2) depression, (3) anxiety, and (4) depression and anxiety using the Log-rank test. Multivariable Cox proportional hazards regression was then performed to determine associations between mental illness and the outcomes of OS and LSS. The regression analyses determined the effects of pre-existing depression, pre-existing anxiety, and co-occurring depression and anxiety relative to no mental illness. These analyses adjusted for covariates including age, sex, race, ethnicity, marital status, income, education, region, urbanicity, year of diagnosis, lymphoma stage, presence of extranodal disease, B symptoms, and Charlson comorbidity score given existing literature suggesting they may confound the relationship between mental health disorders and survival.7,22,23 In light of the Sex and Gender Equity in Research (SAGER) guidelines, we disaggregated our Cox regression analyses for both OS and LSS by sex. Based on data suggesting significant association between income and mental health disorders,5 we conducted a moderation analysis to determine if there was variation in the effect of mental health disorders and the outcomes of OS and LSS for different income levels. Statistical significance was defined as p values of less than 0·05. All statistical analyses were conducted with SAS software, version 9.4 (SAS Institute, Cary, NC) and R software version 4.0.5.
Role of the funding source
The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.
RESULTS
We identified 13,244 patients diagnosed with diffuse large B-cell lymphoma between January 1, 2001 and December 31, 2013 who were eligible for this study. The median age at time of diagnosis was 79 years (interquartile range: 73–84). Slightly over half of the cohort was female (52.8%) and married (54.6%), and the majority of patients were white (94.1%). Patients with advanced stage disease comprised 48.1% of the cohort and the majority (61.5%) had nodal disease. Among those who had B symptoms assessed and recorded, 28% had B symptoms present. Additional cohort characteristics are listed in Table 1.
Table 1:
Baseline characteristics
| Characteristic | Entire Cohort | No Mental Health Disorder | Any Mental Health Disorder | p-value |
|---|---|---|---|---|
| Total | 13,244 | 11,150 | 2,094 | |
| Age at Diagnosis, yrs | ||||
| Mean (SD) | 78.7 (6.9) | 78.7 (6.9) | 79.1 (6.8) | 0.009 |
| Median (IQR) | 79 (11) | 78 (11) | 79 (10) | 0.006 |
| Sex | ||||
| Male | 6,256 (47.2%) | 5,545 (49.7%) | 711 (34.0%) | <0.0001 |
| Female | 6,988 (52.8%) | 5,605 (50.3%) | 1,383 (66.1%) | |
| Race * | ||||
| White | 12,468 (94.1%) | 10,453 (93.7%) | 2,015 (96.2%) | <0.0001 |
| Non-White | 776 (5.9%) | 697 (6.3%) | 79 (3.8%) | |
| Ethnicity | ||||
| Non-Hispanic | 12,807 (96.7%) | 10,769 (96.6%) | 2,038 (97.3%) | 0.08 |
| Hispanic | 437 (3.3%) | 381 (3.4%) | 56 (2.7%) | |
| Income (census-level) | ||||
| <5% below FPL | 3,805 (28.7%) | 3,240 (29.1%) | 565 (27.0%) | 0.40 |
| 5% to <10% below FPL | 3,799 (28.7%) | 3,181 (28,5%) | 618 (29.5%) | |
| 10% to <20% below FPL | 3,194 (24.1%) | 2,685 (24.1%) | 509 (24.3%) | |
| 20% to 100% below FPL | 1,439 (10.9%) | 1,204 (10.8%) | 235 (11.2%) | |
| Unknown | 1,007 (7.6%) | 840 (7.5%) | 167 (8.0%) | |
| Education * | ||||
| <25% with <12 years of education | 11,819 (89.2%) | 9,941 (89.2%) | 1,878 (89.7%) | 0.47 |
| ≥25% with <12 years of education | 1,425 (10.8%) | 1,209 (10.8%) | 216 (10.3%) | |
| Marital Status | ||||
| Married | 7,228 (54.6%) | 6,242 (56.0%) | 986 (47.1%) | <0.0001 |
| Non-Married/Other | 5,310 (40.1%) | 4,306 (38.6%) | 1,004 (48.0%) | |
| Unknown | 706 (5.3%) | 602 (5.4%) | 104 (5.0%) | |
| SEER Registry Region | ||||
| Northeast | 3,037 (22.9%) | 2,502 (22.4) | 545 (25.6) | 0.0005 |
| Midwest | 3,029 (22.9%) | 2,532 (22.7%) | 497 (23.7%) | |
| South | 1,938 (14.6%) | 1,624 (14.6%) | 314 (15.0%) | |
| West | 5,240 (39.6%) | 4,492 (40.3%) | 748 (35.7%) | |
| Urbanicity | ||||
| Metro | 11,054 (83.5%) | 9,296 (83.4%) | 1,758 (84.0%) | 0.80 |
| Urban | 1,899 (14.3%) | 1,607 (14.4%) | 292 (13.9%) | |
| Rural | 291 (2.2%) | 247 (2.2%) | 44 (2.1%) | |
| Charlson Comorbidity Score | ||||
| 0 | 6,232 (47.1%) | 5,521 (49.5%) | 711 (34.0%) | <0.0001 |
| 1 | 3,359 (25.4%) | 2,821 (25.3%) | 538 (25.7%) | |
| 2+ | 3,653 (27.6%) | 2,808 (25.2%) | 845 (40.4%) | |
| DLBCL Stage | ||||
| I | 3,853 (29.1%) | 3,257 (29.2%) | 596 (28.5%) | 0.20 |
| II | 2,158 (16.3%) | 1,842 (16.5%) | 316 (15.1%) | |
| III | 1,864 (14.1%) | 1,572 (14.1%) | 292 (13.9%) | |
| IV | 4,505 (34.0%) | 3,748 (33.6%) | 757 (36.2%) | |
| Unstaged | 864 (6.5%) | 731 (6.6%) | 133 (6.4%) | |
| Extranodal Disease | ||||
| Yes | 5,093 (38.5%) | 4,276 (38.4%) | 817 (39.0%) | 0.57 |
| No | 8,151 (61.5%) | 6,874 (61.7%) | 1,277 (61.0%) | |
| Year of Diagnosis | ||||
| 2001–2003 | 2,812 (21.2%) | 2,439 (21.9%) | 373 (17.8%) | <0.0001 |
| 2004–2007 | 4,229 (31.9%) | 3,642 (32.7%) | 587 (28.0%) | |
| 2008–2011 | 4,024 (30.4%) | 3,359 (30.1%) | 665 (31.8%) | |
| 2012–2013 | 2,179 (16.5%) | 1,710 (15.3%) | 469 (22.4%) | |
| B Symptoms *** | ||||
| Absent | 4,078 (30.8%) | 3,360 (30.1%) | 718 (34.3%) | <0.0001 |
| Present | 1,590 (12.0%) | 1,313 (11.8%) | 277 (13.2%) | |
| Unknown | 4,283 (14.1%) | 3,696 (33.2%) | 587 (28.0%) | |
| NA | 3,293 (24.9%) | 2,781 (24.9%) | 512 (24.5%) |
Nonwhite race includes 49 patients with unknown race
Education based on census-level; if census tract missing, based on zip code
Diagnoses prior to 2004 coded as unknown
FPL: Federal Poverty Level
In our cohort, 2,094 patients (15.8%) met our claims-based definition of pre-existing depression or anxiety at the time of their DLBCL diagnosis. 1,176 patients (8.9%) met diagnostic criteria for depression, 608 patients (4.6%) had anxiety, and 310 patients (2.3%) had both depression and anxiety. Patients who met our claims-based criteria for pre-existing depression and/or anxiety were more likely to be female (66.1% vs 50.3%, p <0.0001), non-married (48% vs 38.6%, p <0.0001), and more likely to have more comorbidities (40.4% vs 25.2% Charlson Comorbidity score of 2+) when compared to those without pre-existing anxiety and/or depression. The median follow-up for the total cohort was 2.0 years (IQR 0.4 – 6.9 years) and for those alive at the end of the study, the median follow-up was 8.5 years (IQR 6.5 – 11.5 years).
The unadjusted Kaplan-Meier curves for OS are shown in Figure 2. The 5-year OS in patients with depression and/or anxiety was 27.0% (95% CI, 25.1% to 28.9%) compared with 37.4% (95% CI 36.5%−38.3%) in patients without any pre-existing mental disorder. When stratified by different types of mental health disorders, individuals with pre-existing depression had the worst overall survival (5-year OS 23.4%, 95% CI 21.0%−25.8%). Results from the multivariable regression analysis are listed in Table 2. Each category of pre-existing mental health disorders (depression, anxiety, concurrent depression and anxiety) was associated with significantly inferior OS compared to those with no pre-existing mental health disorder. The strongest effect was seen in those with pre-existing depression (Hazard Ratio [HR] 1.37, 95% CI 1.28 – 1.47). Results disaggregated by sex were similar across sexes (appendix p 4). In our moderation analysis, we found that income significantly moderated the effect of mental health disorders on OS, with an interaction p value of 0.02 (appendix p 6).
Figure 2a.
Overall Survival: Any mental health disorder versus no mental health disorder
Table 2:
Cox Proportional Regression Results of Factors Associated with 5-Year Overall Survival
| Cox Proportional Hazard | ||
|---|---|---|
| HR (95% CI) | p-value | |
| Mental Health Disorder | ||
| No mental health disorder | Ref | Ref |
| Depression Only | 1.37 (1.28 – 1.47) | <0.0001 |
| Anxiety Only | 1.17 (1.06–1.29) | 0.003 |
| Depression and Anxiety | 1.23 (1.08–1.41) | 0.002 |
| Age at Diagnosis, yr | ||
| 67–69 | Ref | Ref |
| 70–74 | 1.35 (1.22–1.49) | <0.0001 |
| 75–79 | 1.62 (1.48–1.79) | <0.0001 |
| 80+ | 2.55 (2.33–2.79) | <0.0001 |
| Sex | ||
| Male | Ref | Ref |
| Female | 0.81 (0.77–0.85) | <0.0001 |
| Race | ||
| White | Ref | Ref |
| Non-White | 0.94 (0.85–1.04) | 0.20 |
| Ethnicity | ||
| Non-Hispanic | Ref | Ref |
| Hispanic | 1.07 (0.95–1.21) | 0.27 |
| Income (census-level) | ||
| <5% below FPL | Ref | Ref |
| 5% to <10% below FPL | 1.07 (1.004–1.13) | 0.03 |
| 10% to <20% below FPL | 1.09 (1.02–1.16) | 0007 |
| 20% to 100% below FPL | 1.07 (0.98–1.17) | 0.11 |
| Unknown | 1.05 (0.95–1.17) | 0.37 |
| Education | ||
| <25% with <12 years of education | Ref | Ref |
| ≥25% with <12 years of education | 1.14 (1.06–1.23) | 0.0005 |
| Marital Status | ||
| Married | Ref | Ref |
| Non-Married/Other | 1.19 (1.14–1.25) | <0.0001 |
| Unknown | 0.88 (0.79–0.98) | 0.02 |
| SEER Registry Region | ||
| Northeast | Ref | Ref |
| Midwest | 1.11 (1.03–1.19) | 0.007 |
| South | 1.10 (1.02–1.19) | 0.02 |
| West | 0.95 (0.90–1.01) | 0.10 |
| Urbanicity | ||
| Metro | Ref | Ref |
| Urban | 1.07 (1.00–1.14) | 0.0504 |
| Rural | 0.90 (0.78–1.05) | 0.18 |
| Charlson Comorbidity Score | ||
| 0 | Ref | Ref |
| 1 | 1.25 (1.19–1.32) | <0.0001 |
| 2+ | 1.80 (1.71–1.89) | <0.0001 |
| Stage | ||
| I | Ref | Ref |
| II | 1.32 (1.23–1.41) | <0.0001 |
| III | 1.49 (1.38–1.61) | <0.0001 |
| IV | 1.66 (1.56–1.78) | <0.0001 |
| Unstaged | 1.27 (1.15–1.40) | <0.0001 |
| Extranodal Involvement | ||
| No | Ref | Ref |
| Yes | 0.99 (0.94–1.04) | 0.69 |
| Year of Diagnosis | ||
| 2001–2003 | Ref | Ref |
| 2004–2007 | 0.78 (0.72–0.85) | <0.0001 |
| 2008–2011 | 0.65 (0.59–0.71) | <0.0001 |
| 2012–2013 | 0.61 (0.56–0.68) | <0.0001 |
| B Symptoms | ||
| Absent | Ref | Ref |
| Present | 1.33 (1.24–1.43) | <0.0001 |
| Unknown | 0.94 (0.87–1.02) | 0.12 |
| NA | 1.03 (0.95–1.11) | 0.48 |
HR: Hazard Ratio (values higher than 1 indicate greater mortality)
FPL: Federal Poverty Level
Figure 3 depicts the Kaplan-Meier curves for LSS. The 5-year LSS in patients with depression and/or anxiety was 42.0% (95% CI, 39.8% to 44.3%) compared with 51.3% (95% CI 50.3%−52.2%) in patients without any pre-existing mental disorder. Patients with pre-existing depression had the worst LSS (5-year LSS 42.3%, 95% CI 36.3%−48.1%). In our Cox proportional regression analyses, we observed a significantly worse LSS for each category of pre-existing mental health disorder compared to no mental health disorder, with the strongest effect observed in those with pre-existing depression (HR 1.37, 95% CI 1.26 – 1.49; appendix p 7). Results disaggregated by sex were consistent across sexes (appendix p 9). Moderation analysis did not reveal significant effect moderation of mental health disorder on LSS (p=0.10; appendix p 11).
Figure 3a.
Lymphoma-Specific Survival: Any mental health disorder versus no mental health disorder.
DISCUSSION
In this large cohort of older patients with DLBCL, we found a substantial prevalence of pre-existing mental health disorders, with nearly 1 in 6 individuals experiencing depression and/or anxiety prior to their DLBCL diagnosis. Patients with pre-existing mental health disorders had significantly lower OS and LSS, even after adjustment for known factors related to survival (lymphoma stage, age, comorbidity). Of the disorders studied, pre-existing depression was associated with the worst survival, with a 37% increased hazard for both all-cause and lymphoma-specific death compared with those without any mental health disorder. Our results suggest that mental health disorders are predictive of survival for patients with DLBCL and underscore the urgent need for systematic mental assessment and interventions for patients diagnosed with DLBCL.
Our finding that DLBCL with pre-existing mental health disorders had worse OS is consistent with prior studies of patients with solid malignancies.5–7 For example, in a study of older women diagnosed with stage I-IIIa breast cancer, individuals with pre-existing depression had a 39% increased hazard for all-cause mortality.5 A key distinction in our study is that pre-existing anxiety—and not only depression—also significantly impacted all-cause and disease-specific mortality. This suggests that assessing for and effectively managing anxiety is also important for patients with DLBCL. The relative dearth of literature examining the survival impact of pre-existing mental health disorders among patients with hematologic malignancies limits our ability to compare our findings to this population; however, existing studies demonstrate that patients who develop mental health disorders after diagnosis of Hodgkin lymphoma or leukemia have worse overall survival.12,13
Our study results raise the question of what mechanism might account for the relationship between pre-existing mental health disorders and decreased OS and LSS. One hypothesis is that patients with depression and/or anxiety may experience delays in lymphoma-directed treatment, thus resulting in worse survival outcomes. A new cancer diagnosis is destabilizing, and it may be too overwhelming for patients with underlying depression and anxiety to coordinate complicated treatment plans in a timely manner if they are not provided with adequate mental and practical resources. One study of patients with breast cancer demonstrated associations between severe mental illness and treatment delays.24
Another possible mechanism contributing to the survival impact we demonstrated is that patients with mental health disorders may experience more barriers—financial constraints, limited social support—that limit adherence to their treatment plan.25,26 For example, a large meta-analysis found that individuals with depression had three times greater odds of not adhering to medical treatment recommendations compared to individuals without depression.26 Lack of adequate social support—even when treatment is initiated in a timely fashion—may make it more difficult to adhere to treatment schedules or seek out appropriate management of treatment adverse effects. Stigma towards mental health disorders may be another factor contributing to decreased survival. Prior research suggests that cancer patients with mental health disorders are still stigmatized by their care teams, and physicians who stigmatize patients with underlying mental health disorders are more dubious of their patients’ ability to adhere to treatment.27 Future research that investigates the potential mechanisms by which pre-existing mental health disorders may impact survival among DLBCL patients is needed.
Identifying and treating pre-existing mental health disorders as a prognostic factor for patients with DLBCL is clinically important, as it is potentially modifiable. While many prognostic factors, such as age and disease stage, have been identified for DLBCL, few of them are modifiable. In contrast, depression and anxiety are treatable conditions in many cases. Psychological interventions, such as cognitive-behavioral, supportive-expressive, and psychodynamic therapies, and pharmacologic treatments that improve mental illness may positively affect survival.28 Indeed, in one study, individuals with metastatic breast cancer who were randomized into weekly group therapy sessions lived on average 18 months longer than those randomized to a control group.29 Interventions targeting potential mechanisms by which mental health disorders may influence outcomes (e.g., provision of additional social support and practical resources to maintain treatment schedule and intensity) may also attenuate the effect of these disorders on survival. Moreover, these interventions can improve quality of life for patients undergoing treatment for DLBCL irrespective of their impact on survival. In addition to intervening on existing mental health disorders, innovative risk prediction models that identify patients at high risk of developing depression can be used to target primary prevention interventions to these individuals.30
We acknowledge limitations to our study. First, a claims-based approach to measuring depression and anxiety likely underestimates the true rate of mental health disorders. Physicians may not file a claim if their patients have only mild psychological symptoms, thereby falsely lowering the apparent rate of mental health disorders. Additionally, physicians may not prioritize billing claims for psychological conditions for complex patients that have many active medical conditions. Second, our cohort only included patients who were 67 years or older, which potentially minimizes generalizability to younger populations. Nonetheless, DLBCL most commonly affects an older population, with an average age of 66 at diagnosis, thus still ensuring relevance of our results.1 Third, our study could not uncover the mechanism for decreased survival among patients with pre-existing health disorders. We are thus unable to determine whether the relationship between pre-existing mental health disorders and survival is causative or merely correlative. Future studies (e.g., mediation analyses) that examine potential factors such as treatment delays that may mediate the relationship between mental health disorders and survival are warranted. Fourth, our cohort was limited to the U.S. population, and it is possible that the prevalence and impact of mental health disorders for patients with DLBCL differs in other international populations. Finally, although we adjusted for several covariates in our analyses, there were other factors such as smoking and physical symptoms (e.g., pain) that we did not have access to, which may be sources of residual confounding.
In conclusion, we found that a substantial proportion of patients with DLBCL have pre-existing mental health disorders. These disorders are highly prognostic, as affected patients have decreased OS and LSS, with pre-existing depression conferring the greatest impact on survival. These data suggest that comprehensive psychosocial assessment should be an essential component of evaluation at time of DLBCL diagnosis. Knowledge of depression and/or anxiety would lead to more accurate prognostication and would help oncologists appropriately deploy necessary resources to affected patients. Moreover, attention to and improvement of these conditions have the potential benefit of improving survival. As the field of lymphoma progresses with state-of-the-art diagnostics and targeted therapeutics, our analysis argues for increased attention to developing and disseminating high-quality psychological interventions for patients with DLBCL.
Supplementary Material
Figure 2b.
Overall Survival: Specific mental health disorder versus no mental health disorder.
Figure 3b.
Lymphoma-Specific Survival: Specific mental health disorder versus no mental health disorder.
Research in Context.
Evidence before this study
We searched PubMed using the following terms without language restriction “hematologic malignancy” AND “depression” or “anxiety” or “mental health” AND “mortality” or “survival.” We found five studies dating between January 1, 1990 and November 1, 2022 examining the associations between mental health disorders and mortality among patients with hematologic malignancies. All studies were relatively small (ranging from 92 to 795 patients). Two studies included patients with specific hematologic malignancies (Hodgkin lymphoma or acute myeloid leukemia), while the other three examined patients with a variety of different blood cancers. The majority of studies (4/5) found depression correlated with decreased survival and underscored the importance of improving psychosocial care. The one study that did not show a survival impact was from 1990 and had only 92 patients.
Added value of this study
To our knowledge, this is the largest retrospective cohort analysis assessing the association between mental health disorders and mortality in patients with a blood cancer and the first specifically for patients with DLBCL. We demonstrated that pre-existing mental health disorders in patients with DLBCL was associated with decreased overall and lymphoma-specific survival, suggesting an innovative approach to improve prognosis.
Implications of all the available evidence
We found that almost one in six patients with newly diagnosed DLBCL suffers from pre-existing mental health disorders, and that their presence is associated with worse mortality. Given underlying depression and/or anxiety is potentially modifiable, interventions aimed at treating mental illness and improving mortality for this population are critically needed.
Acknowledgements
T.M. Kuczmarski received support from the American Society of Hematology (HONORS Award), O.O. Odejide received support from the American Society of Hematology (Scholar Award), the Alan J. Hirschfield Lymphoma Research Award, and the National Cancer Institute [NCI] (K08-CA218295), and L. Roemer and O.O. Odejide received support from the NCI (U54 DF/HCC-UMB-CA156732; CA156734)
The collection of cancer incidence data used in this study was supported by the California Department of Public Health pursuant to California Health and Safety Code Section 103885; Centers for Disease Control and Prevention’s (CDC) National Program of Cancer Registries, under cooperative agreement 1NU58DP007156; the National Cancer Institute’s Surveillance, Epidemiology and End Results Program under contract HHSN261201800032I awarded to the University of California, San Francisco, contract HHSN261201800015I awarded to the University of Southern California, and contract HHSN261201800009I awarded to the Public Health Institute. The ideas and opinions expressed herein are those of the author(s) and do not necessarily reflect the opinions of 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.
Funding
American Society of Hematology, National Cancer Institute, Alan J. Hirschfield Award.
Footnotes
Declaration of interests
AL reports royalties or licenses from UpToDate; payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing or educational events from Research to Practice; participation on a Data Safety Monitoring Board or Advisory Board from Seagen and Kite Pharma. All other authors declare no competing interests.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Data Sharing
Data for this manuscript were made available through application to the SEER-Medicare database. SEER-Medicare data use agreement, SEER-Medicare data cannot be publicly shared. Researchers may obtain approval from the NCI by submitting an Application Form, Data User Agreement, and documentation of IRB approval. Requests for access to the SEER-Medicare dataset can be made through NCI at https://healthcaredelivery.cancer.gov/seermedicare/obtain/requests.html or through the SEER-Medicare contact, Elaine Yanisko at YaniskoE@imsweb.com. The authors confirm they did not have any special access to this data that other researchers would not have.
REFERENCES
- 1.Teras LR, DeSantis CE, Cerhan JR, Morton LM, Jemal A, Flowers CR. 2016 US lymphoid malignancy statistics by World Health Organization subtypes. CA Cancer J Clin. 2016;66(6):443–59. [DOI] [PubMed] [Google Scholar]
- 2.Ziepert M, Hasenclever D, Kuhnt E, Glass B, Schmitz N, Pfreundschuh M, et al. Standard International Prognostic Index Remains a Valid Predictor of Outcome for Patients With Aggressive CD20+ B-Cell Lymphoma in the Rituximab Era. J Clin Oncol. 2010. May 10;28(14):2373–80. [DOI] [PubMed] [Google Scholar]
- 3.Plana-Ripoll O, Pedersen CB, Agerbo E, Holtz Y, Erlangsen A, Canudas-Romo V, et al. A comprehensive analysis of mortality-related health metrics associated with mental disorders: a nationwide, register-based cohort study. Lancet Lond Engl. 2019. Nov 16;394(10211):1827–35. [DOI] [PubMed] [Google Scholar]
- 4.Polański J, Chabowski M, Chudiak A, Uchmanowicz B, Janczak D, Rosińczuk J, et al. Intensity of Anxiety and Depression in Patients with Lung Cancer in Relation to Quality of Life. Adv Exp Med Biol. 2018;1023:29–36. [DOI] [PubMed] [Google Scholar]
- 5.Iglay K, Santorelli ML, Hirshfield KM, Williams JM, Rhoads GG, Lin Y, et al. Impact of Preexisting Mental Illness on All-Cause and Breast Cancer-Specific Mortality in Elderly Patients With Breast Cancer. J Clin Oncol Off J Am Soc Clin Oncol. 2017. Dec 20;35(36):4012–8. [DOI] [PubMed] [Google Scholar]
- 6.Paredes AZ, Hyer JM, Tsilimigras DI, Palmer E, Lustberg MB, Dillhoff ME, et al. Association of pre-existing mental illness with all-cause and cancer-specific mortality among Medicare beneficiaries with pancreatic cancer. HPB. 2021. Mar;23(3):451–8. [DOI] [PubMed] [Google Scholar]
- 7.Sathianathen NJ, Fan Y, Jarosek SL, Konety I, Weight CJ, Vinogradov S, et al. Disparities in Bladder Cancer Treatment and Survival Amongst Elderly Patients with a Pre-existing Mental Illness. Eur Urol Focus. 2020. Nov 15;6(6):1180–7. [DOI] [PubMed] [Google Scholar]
- 8.Giese-Davis J, Collie K, Rancourt KMS, Neri E, Kraemer HC, Spiegel D. Decrease in depression symptoms is associated with longer survival in patients with metastatic breast cancer: a secondary analysis. J Clin Oncol Off J Am Soc Clin Oncol. 2011. Feb 1;29(4):413–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Øvlisen AK, Jakobsen LH, Kragholm KH, Nielsen RE, de Nully Brown P, Dahl-Sørensen RB, et al. Mental health among patients with non-Hodgkin lymphoma: A Danish nationwide study of psychotropic drug use in 8750 patients and 43 750 matched comparators. Am J Hematol. 2022;97(6):749–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kuczmarski TM, Jaung T, Mancuso CE, Mozessohn L, Roemer L, Abel GA, et al. Precancer and cancer-associated depression and anxiety among older adults with blood cancers in the United States. Blood Adv. 2022. Feb 11;6(4):1126–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Oberoi DV, White VM, Seymour JF, Prince HM, Harrison S, Jefford M, et al. Distress and unmet needs during treatment and quality of life in early cancer survivorship: A longitudinal study of haematological cancer patients. Eur J Haematol. 2017;99(5):423–30. [DOI] [PubMed] [Google Scholar]
- 12.Tao R, Chen Y, Kim S, Ocier K, Lloyd S, Poppe MM, et al. Mental health disorders are more common in patients with Hodgkin lymphoma and may negatively impact overall survival. Cancer. 2022;128(19):3564–72. [DOI] [PubMed] [Google Scholar]
- 13.Ding T, Wang X, Fu A, Xu L, Lin J. Anxiety and depression predict unfavorable survival in acute myeloid leukemia patients. Medicine (Baltimore). 2019. Oct;98(43):e17314. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Prieto JM, Atala J, Blanch J, Carreras E, Rovira M, Cirera E, et al. Role of depression as a predictor of mortality among cancer patients after stem-cell transplantation. J Clin Oncol Off J Am Soc Clin Oncol. 2005. Sep 1;23(25):6063–71. [DOI] [PubMed] [Google Scholar]
- 15.Hasegawa T, Okuyama T, Uchida M, Aiki S, Imai F, Nishioka M, et al. Depressive symptoms during the first month of chemotherapy and survival in patients with hematological malignancies: A prospective cohort study. Psychooncology. 2019. Aug;28(8):1687–94. [DOI] [PubMed] [Google Scholar]
- 16.Richardson JL, Zarnegar Z, Bisno B, Levine A. Psychosocial status at initiation of cancer treatment and survival. J Psychosom Res. 1990. Jan 1;34(2):189–201. [DOI] [PubMed] [Google Scholar]
- 17.Gong I, Cheung M, Chan K, Arya S, Faught N, Liu N, et al. Incidence of Mental Health Events and Its Association with Survival Among Patients with Diffuse Large B-Cell Lymphoma: A Population-Based Cohort Study. Blood. 2021. Nov 23;138:1969. [Google Scholar]
- 18.About the SEER Program [Internet]. SEER. [cited 2022 May 26]. Available from: https://seer.cancer.gov/about/overview.html [Google Scholar]
- 19.Enewold L, Parsons H, Zhao L, Bott D, Rivera DR, Barrett MJ, et al. Updated Overview of the SEER-Medicare Data: Enhanced Content and Applications. J Natl Cancer Inst Monogr. 2020. May;2020(55):3–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.McDermott C, Bansal A, Ramsey SD, Lyman GH, Sullivan SD. Depression and healthcare utilization at end of life among older adults with advanced non-small cell lung cancer. J Pain Symptom Manage. 2018. Nov;56(5):699–708.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.NCI Comorbidity Index Overview [Internet]. [cited 2023 Mar 26]. Available from: https://healthcaredelivery.cancer.gov/seermedicare/considerations/comorbidity.html [Google Scholar]
- 22.Prasad SM, Eggener SE, Lipsitz SR, Irwin MR, Ganz PA, Hu JC. Effect of depression on diagnosis, treatment, and mortality of men with clinically localized prostate cancer. J Clin Oncol Off J Am Soc Clin Oncol. 2014. Aug 10;32(23):2471–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Shreders AJ, Niazi SK, Hodge DO, Chimato NT, Kureti M, Kirla N, et al. Correlation of sociodemographic and clinical parameters with depression and distress in patients with hematologic malignancies. Ann Hematol. 2018. Mar;97(3):519–28. [DOI] [PubMed] [Google Scholar]
- 24.Iglay K, Santorelli ML, Hirshfield KM, Williams JM, Rhoads GG, Lin Y, et al. Diagnosis and treatment delays among elderly breast cancer patients with pre-existing mental illness. Breast Cancer Res Treat. 2017. Nov 1;166(1):267–75. [DOI] [PubMed] [Google Scholar]
- 25.Zamanian H, Amini-Tehrani M, Jalali Z, Daryaafzoon M, Ala S, Tabrizian S, et al. Perceived social support, coping strategies, anxiety and depression among women with breast cancer: Evaluation of a mediation model. Eur J Oncol Nurs Off J Eur Oncol Nurs Soc. 2021. Feb;50:101892. [DOI] [PubMed] [Google Scholar]
- 26.DiMatteo MR, Lepper HS, Croghan TW. Depression Is a Risk Factor for Noncompliance With Medical Treatment: Meta-analysis of the Effects of Anxiety and Depression on Patient Adherence. Arch Intern Med. 2000. Jul 24;160(14):2101–7. [DOI] [PubMed] [Google Scholar]
- 27.D’Alton P, O’Meara R, Langford S, McDonnell Z, Nuzum A, Murthy VE, et al. Barriers to cancer care for people with significant mental health difficulties: What healthcare staff say? Psychooncology. 2021;30(12):2032–8. [DOI] [PubMed] [Google Scholar]
- 28.Grassi L, Spiegel D, Riba M. Advancing psychosocial care in cancer patients. F1000Research. 2017. Dec 4;6:2083. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Spiegel D, Bloom JR, Kraemer HC, Gottheil E. Effect of psychosocial treatment on survival of patients with metastatic breast cancer. Lancet Lond Engl. 1989. Oct 14;2(8668):888–91. [DOI] [PubMed] [Google Scholar]
- 30.Belvederi Murri M, Cattelani L, Chesani F, Palumbo P, Triolo F, Alexopoulos GS. Risk Prediction Models for Depression in Community-Dwelling Older Adults. Am J Geriatr Psychiatry. 2022. Sep 1;30(9):949–60. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data for this manuscript were made available through application to the SEER-Medicare database. SEER-Medicare data use agreement, SEER-Medicare data cannot be publicly shared. Researchers may obtain approval from the NCI by submitting an Application Form, Data User Agreement, and documentation of IRB approval. Requests for access to the SEER-Medicare dataset can be made through NCI at https://healthcaredelivery.cancer.gov/seermedicare/obtain/requests.html or through the SEER-Medicare contact, Elaine Yanisko at YaniskoE@imsweb.com. The authors confirm they did not have any special access to this data that other researchers would not have.