Introduction
Non-Hodgkin lymphoma (NHL) has been a major source of morbidity and mortality among individuals with human immunodeficiency virus (HIV) since the first description of acquired immunodeficiency syndrome (AIDS) in the early 1980s [
1]. Regarding lymphoma patients without HIV, an increased focus on quality of life (QOL) has emerged [
2]. However, there are few studies of HIV-associated NHL that focus on QOL. We are aware of a single paper reporting QOL among HIV-infected patients who received oral combination chemotherapy with filgrastim (granulocyte colony-stimulating factor (G-CSF) [
3]. Indeed, a recent Cochrane review of treatment of AIDS-related NHL specifically requested more trials with QOL as an endpoint [
4]. Addressing this knowledge gap, we performed a case–control study comparing QOL as well as clinical characteristics between patients with HIV and NHL and NHL without HIV infection.
Since the introduction of highly active antiretroviral therapy (HAART) in the late 1990s, survival among patients with AIDS-related NHL has improved dramatically [
5]. While physicians 10 years ago were hesitant to treat patients with AIDS and NHL with chemotherapy, contemporary patients typically receive the same aggressive chemotherapy regimens that patients without AIDS receive [
6]. Despite these therapeutic improvements, we hypothesized that contemporary patients with HIV infection and NHL would have worse QOL than contemporary uninfected patients with NHL and historical HIV-infected patients.
HIV-infected NHL patients differ from NHL patients without HIV infection. They are typically younger and more likely to be male and racial/ethnic minorities [
7]. Because of these demographic differences, we matched by age, race and sex in comparing NHL patients with and without HIV-infection. Also, HIV-infected individuals (with or without concurrent NHL) are more likely to be poor, publicly insured, and illicit drug users than uninfected individuals [
8]. Because of the social stigma associated with HIV and its associated sexual and drug-related risk factors, the HIV-infected may feel rejected by their families and society as a whole. Compared with uninfected NHL patients, those with concurrent HIV and NHL have an additional co-morbidity which requires medication and may have caused pre-existing disability. NHL histology among the HIV-infected typically is more aggressive than among patients without HIV [
1]. Because of these social and biologic differences between NHL patients with and without HIV infection, we anticipated that we would find a lower QOL among the HIV-infected NHL patients in all categories of well-being measured: emotional, social, physical and functional.
Methods
The University of California Irvine Institutional Review Board approved the study prior to initiation (#2002-2428) in accordance with the 1964 Declaration of Helsinki. We recruited 100 residents of Orange or San Diego Counties diagnosed with systemic NHL from 2002–2006, 50 HIV-infected and 50 without HIV diagnoses. We identified subjects through the California cancer registry for Orange and San Diego Counties and provider referrals. We sought to contact patients within 1 year of their NHL diagnosis. Prior to contacting a patient, we sent a letter to the reporting physician to determine if there was any reason to not contact the patient. If there were no contraindications, we sent a packet with the consent form, medical record release form, medical history survey and QOL instrument to the potential subject along with a cover letter explaining the study. If the potential subject did not return the signed consent and surveys, a research associate telephoned the subject to answer questions, encourage participation and offer assistance completing the packet.
During 2002–2006, the California Cancer Registry for Orange and San Diego Counties contained 171 patients flagged as HIV-infected (potential cases) and 4,932 patients with systemic NHL flagged as HIV-seronegative or sero-unknown (potential controls). In previous research we have demonstrated this HIV flag to be 97% sensitive and 92% specific [
9]. Of the 171 potential cases, we recruited 51. Thirty potential cases died prior to contact and we were unable to make contact with 15; one was too ill to participate. The remaining 74 potential cases were reported to the Cancer Registry too late to be eligible for the study (more than 1 year after NHL diagnosis). We excluded one case from analysis because we did not have a control for him. We attempted to contact 130 controls with an initial goal of a 2:1 case:control ratio. Seven potential controls died prior to contact; we were unable to make contact with 37 and 16 declined participation. Although we recruited 70 controls, we selected the 50 sex and race-matched controls with complete data that matched the cases by age (±5 years) in order to achieve a 1:1 case:control ratio. Thus, the analysis included 50 cases and 50 controls.
The QOL instrument was the Functional Assessment of Human Immunodeficiency Virus Infection (FAHI) in patients with HIV and the Functional Assessment of Cancer Therapy-General (FACT G) in patients without HIV [
10,
11]. These QOL instruments have been validated in English and Spanish [
12,
13] and all patient forms and surveys were available in both languages. The FAHI contains the entire FACT G and additional items reflecting HIV/AIDS specific concerns. This allows cross-disease comparison of QOL through calculation of an HIV specific score and a FACT G score derived from the FAHI. A SAS (Cary, NC) program is available that calculates these scores [
13].
We used data from California Cancer Registry as of April, 2008 to obtain information on chemotherapy, histology, site and stage of NHL. We used the classification system of the International Classification of Diseases for Oncology (ICD-O Third Edition) [
14] to divide the histology into three categories: diffuse large B cell lymphoma (DLBCL, code 9680 and 9684) follicular (codes 9690, 9691, 9695, and 9698) and other. We abstracted CD4 cell counts and viral load results drawn within 3 months of NHL diagnosis (but prior to chemotherapy) from the medical record.
Statistical methods
We used SAS 9.2 for all statistical analyses. We performed the chi-square test to compare proportions between cases and controls. We used the
t-test to compare mean QOL scores between cases with or without a diagnosis of HIV prior to NHL diagnosis and time between NHL diagnosis and study entry between cases and controls. We used the General Linear Model (GLM) procedure to compare unadjusted quality of life scores between cases and controls. Although we matched cases and controls on age, race and sex, cases and controls differed socioeconomically and biologically. To account for correlations among the QOL subscales, we performed multivariate analysis of variance (MANOVA) adjusting for all socioeconomic and biologic variables that differed significantly between cases and controls (Table
1). We employed the Kaplan–Meier method to calculate survival. All tests are two-sided with
p < .05 considered statistically significant.
Table 1
Characteristics of Non-Hodgkin Lymphoma (NHL) patients with and without Human Immunodeficiency Virus (HIV) infection, residing in Orange and San Diego Counties, 2002–2006
Insurance
| | | .03 |
Government including Medicaid, Medicare, county, veteran, military, and Ryan White | 27 (54) | 14 (28) | |
Private | 14 (28) | 21 (42) | |
Self pay/declined to state/do not know | 9 (18) | 15 (30) | |
Income
| | | .001 |
Less than $20,000 annual | 23 (46) | 9 (18) | |
$20,000 or more annual | 18 (36) | 37 (74) | |
Declined to state | 9 (18) | 4 (8) | |
Histology
| | | .001 |
Diffuse large B cell (including immunoblastic) | 35 (70) | 19 (38) | |
Follicular | 0 (0) | 17 (34) | |
Other | 15 (30) | 14 (28) | |
Nodal status
| | | .02 |
Nodal | 34 (68) | 44 (88) | |
Extranodal | 16 (32) | 6 (12) | |
Stage
a
| | | .59 |
Stage I–III | 18 (37) | 21 (42) | |
Stage IV | 31 (63) | 29 (58) | |
Chemotherapy
| | | .78 |
Yes | 43 (86) | 42 (84) | |
No | 7 (14) | 8 (16) | |
Discussion
Despite the availability of antiretrovirals and chemotherapy, HIV-infected patients had worse overall QOL and survival than uninfected patients with NHL. We anticipated that we would find a lower quality of life among the HIV-infected NHL patients in all categories of well-being, physical and functional as well as social and emotional. However, we found that worse QOL among the HIV-infected was due to lower functional, physical and social well-being rather than emotional well-being (Table
2). HIV-infected patients were more likely to have diffuse large B cell and extranodal lymphoma (Table
1). These biologic differences between NHL patients with and without HIV infection combined with the debility and medication management associated with concurrent HIV infection may explain the lower physical QOL subscale scores among the HIV-infected. HIV-infected patients had less income and were less likely to have private insurance than uninfected patients (Table
1), possibly contributing to a lower functional QOL even prior to NHL diagnosis.
The low social well-being scores for the HIV-infected may reflect lack of familial or partner support. HIV-infected individuals may be estranged from their families because of relatives’ concerns regarding risk factors for HIV (e.g. homosexuality or injection drug use) or fear and prejudice. Lack of partner support may be due to the death or illness of an HIV-infected partner or lack of societal support for same sex unions.
There was no statistical difference in emotional well-being between cases and controls in the adjusted model. The issues reflected in the emotional well-being score (such as anxiety, hopelessness and fear of death) may be equally concerning to both cases and controls. Alternatively, some HIV-infected patients may have addressed these issues in the past at the time of HIV diagnosis and have become accustomed to living with them. We discuss below how contemporary patients with HIV and NHL have better emotional well-being subscale scores than historical norms for HIV patients, suggesting that the availability of HAART has influenced this subscale score in particular.
Although we initially hypothesized that contemporary patients with HIV-infection and NHL would have worse QOL than historical HIV-infected patients, we were surprised to find that the overall QOL of our patients with HIV-associated NHL was similar to that of historical HIV-infected patients without NHL as measured by the FAHI (Table
3) [
13]. These QOL norms for HIV-infected patients are from the pre-HAART era and reflect the poor prognosis and stigma associated with AIDS of that time. Contemporary patients with HIV-associated NHL have better emotional well-being than historical HIV-infected patients, perhaps due to increased optimism about prognosis due to the availability of HAART. The survival of patients with HIV-associated NHL was better in our study than historical figures [
1], also reflecting the benefit of HAART.
While there was no difference between contemporary and historical patients in the physical, social and cognitive subscales, the functional well-being subscale score was more than two points worse in contemporary HIV-infected patients with NHL than among historical HIV-infected patients but the difference was less than 0.5 standard deviation (SD). In the interpretation of differences in health-related quality of life, an effect size of 0.5 SD is a relatively stable important difference [
16]. The developers of the FACT-G and FAHI consider a five point difference in overall score and a two point difference in subscale score as clinically meaningful for cancer patients [
15]. A decrease in functional well being could be due to the effect of concurrent NHL and its treatment but might be mitigated by antiretroviral therapy.
Beyond the benefit of HAART, designing an intervention to improve the QOL of patients with HIV and NHL is not simple. Chemotherapy regimens of shorter duration with fewer side effects and oral administration would improve QOL. In-home support, both physical and social, to bridge the period of decreased function would improve QOL. While home health care for hygiene requires in-person assistance, perhaps social support could be provided by telephone. Given that AIDS-related NHL is relatively uncommon (390 cases per 100,000 person-years among people with AIDS in 1996–2002) [
17], there would be few people able to participate in support groups for this illness. Interventions in physician offices, infusion centers or hospitals during routine care would be convenient for patients. Since most physicians and nurses lack the training and time to intervene, social workers or psychologists might play a role. Given that many patients with HIV-related NHL are poor and publicly insured, it is unclear how to cover the costs of any intervention.
This manuscript is unique in that it is the only study, to our knowledge, to use information from the cancer registry to identify and contact patients with a HIV-related malignancy. There was a population-based case–control study of NHL using rapid case ascertainment in the San Francisco Bay Area between 1988 and 1995. In that study, all subjects received a questionnaire inquiring about their HIV status and men had the option to participate in a laboratory portion of the study that included HIV serologic testing [
18]. The standardized and validated QOL instruments employed in the study enhance the value of our findings [
10‐
13,
15].
Our study is limited by the moderate number of subjects. The high morbidity and mortality associated with AIDS and lymphoma as well as the social and economic issues discussed above made it challenging to contact and follow-up cases. We also had difficulty finding controls matched by age and race/ethnicity. Because there is no routine system for rapid case ascertainment and we were afraid of patients dying prior to study participation, we did not standardize the time to interview. Since medical records were scattered between HIV specialist and oncologist offices and various hospitals, our information regarding prior AIDS diagnoses and current CD4 cell counts is incomplete. Some patients declined to answer survey questions regarding their income and insurance, also resulting in incomplete data.
Since some potential subjects died prior to contact, there is bias toward improved survival in our HIV-infected study participants compared with all HIV-infected patients with lymphoma. The HIV-infected patients who participated were healthy enough to give consent and complete the surveys and thus were not representative of all HIV-infected patients with NHL. Another source of bias is the different NHL histology between cases and controls. However, given the inherent imbalance in histologic pattern between the HIV-infected and uninfected, it was not feasible to match on histology as well as age, sex, and race/ethnicity. Although we used the population-based cancer registry to recruit, we matched our controls to cases; the controls are not representative of all patients with systemic NHL who would typically be older and include more women than our controls. HIV infection in the United States is more common in men than women, so our study included few women [
7].
Quality of the life among patients with lymphoma is an important yet understudied topic. We conclude that even among the HIV-infected lymphoma patients most likely to do well, QOL and survival lag behind that of uninfected patients. However, emotional QOL has improved compared to historical HIV-infected patient norms. Survival among HIV-infected patients with lymphoma also is better than in the past. Despite the widespread availability of HIV-testing, undiagnosed individuals present with advanced disease [
19]. Similarly, not all HIV-infected patients access antiretroviral therapy. The low CD4 cell counts and elevated viral loads of our patients suggest uncontrolled HIV infection and issues regarding antiretroviral access, adherence, tolerance and/or resistance. Therefore, NHL will continue to be a source of morbidity and mortality among the HIV-infected. Our study provides a foundation for future research.
Acknowledgments
We greatly appreciate the efforts of the patients who participated and physicians who supported our research, especially Dr. Stephen Baird. We are grateful for technical support from Archana Kulkarni, Maureen Gildea-Schroeder, Ed Barber and Georgiana Margarit. We also thank our exceptional research assistants Jeanette Overton and Katherine Miller. The work was funded by National Institutes of Health K07 CA96480 and University AIDS Research Program/California HIV Research Program California Collaborative Treatment Group grants CC02-SD-003, CH05-SD-607, CH07-SD-607.