Presented as a poster at the 46th Annual Meeting of the American Society of Clinical Oncology; Chicago, Illinois; June 4-8, 2010.
T.K.O., D.M.S., W.J.C., F.R.K., and S.S.R. are Distinguished Cancer Scholars of the Georgia Cancer Coalition.
Lung cancer is the leading cause of death among non-acquired immunodeficiency syndrome (AIDS)-defining malignancies. Because highly active antiretroviral therapy (HAART) has improved the survival of patients with human immunodeficiency virus (HIV), the authors evaluated lung cancer outcomes in the HAART era.
HIV-positive patients who were diagnosed with lung cancer at the authors' institution during the HAART era (1995-2008) were analyzed. Patient charts were reviewed for clinical and laboratory data. The CD4 count at diagnosis was treated as a continuous variable and subcategorized into distinct variables with 3 cutoff points (50 cells/mL, 200 cells/mL, and 500 cells/mL). Pearson correlation coefficients were estimated for each covariate studied. Survival was determined by using the Kaplan-Meier method.
Of 80 patients, 73 had nonsmall cell lung cancer. Baseline characteristics were as follows: median patient age, 52 years; male, 80%; African Americans, 84%; injection drug users, 25%; smokers, 100%; and previous exposure to antiretroviral agents, 55%. At the time of cancer diagnosis, the mean CD4 count was 304 cells/mL, and the mean viral load was 82,420 copies/mL. The latency between HIV diagnosis and lung cancer diagnosis was significantly shorter among women (4.1 years vs 7.7 years; P = .02), and 71% of patients received anticancer therapy. The 1-year and 3-year survival rates for stage IIIB/IV were 25% and 0%, respectively. Grade 3/4 toxicities occurred in 60% of patients who received chemoradiation versus 36% of patients who received chemotherapy. Cancer-related survival was better for patients with CD4 counts >200 cells/mL (P = .0298) and >500 cells/mL (P = .0076).
The availability of highly antiretroviral therapy (HAART) since 1996 has improved acquired immunodeficiency syndrome (AIDS)-related outcomes, transforming human immunodeficiency virus (HIV) from a rapidly lethal disease to a chronic illness.1 Although the incidence of HIV has reached a plateau, an estimated 1 million individuals in the United States were infected as of 2006, and the numbers continue to rise.2 Consequently, non-AIDS-related illnesses have become increasingly prevalent, contributing to the morbidity and mortality of HIV-infected patients.3 Thirty percent of HIV-infected patients will develop cancer by age 60 years in the post-HAART era.4 The marked decline in AIDS-defining cancers (ADCs) that has accompanied the use of HAART has resulted in an increase in the proportion of non-ADCs (NADCs). The largest study to examine cancer incidence revealed that lung cancer represented 20% of the 563 NADCs diagnosed from 1996 to 2002.5 Malignancies now account for approximately 33% of all HIV-related deaths, of which lung cancer is the leading cause of NADC mortality.6, 7
Patients with HIV are 2 to 4 times more likely to develop lung cancer than the general population.5, 8, 9 Although smoking remains an independent risk factor for lung cancer, it alone does not account for the increased incidence observed in this population.10, 11 In fact, organ transplantation recipients on immunosuppressive agents have a lung cancer incidence rate comparable to that of HIV-infected patients, thus implicating a role for immunosuppression.12 This is substantiated further by the association between the development of lung cancer and AIDS.5, 8 However, single data point estimation of CD4 counts does not appear to be directly correlated to lung cancer incidence, a finding that may be attributed to the inherent variability of CD4 counts.5, 10 Rather, the population range of CD4 levels, which correlates more accurately with the degree of immunosuppression, seems to have an association.13 A recent study suggested that malignancy rates approach those of the general population in patients with HIV who have CD4 counts >500 cells/mL.14 Additional evidence also suggests that declining CD4 counts are associated with higher lung cancer risk and advanced disease stage at diagnosis.13, 15 Although it has not been demonstrated that antiretroviral therapy (ARV) and its associated viral load suppression have a clear impact on lung cancer incidence, they may have an indirect effect by improving CD4 counts.9, 11, 13, 16
HIV-infected patients often present with advanced stage lung cancer at a younger age and have inferior overall survival than patients who have lung cancer without HIV.15 Although HAART has improved HIV-related outcomes, our understanding of its effect on the risk and clinical behavior of lung cancer is still evolving.15 The current literature in the modern HAART era is sparse with regard to information about lung cancer, optimal treatment, and outcomes in HIV-infected patients. Therefore, we conducted a retrospective review of lung cancer in HIV-positive patients who were treated at our institution to further characterize the clinical aspects of this population.
MATERIALS AND METHODS
All HIV-positive patients who were diagnosed with lung cancer from January 1995 to October 2008 were identified through International Classification of Diseases (9th revision) codes and tumor registries at the Atlanta Veterans Administration (VA) Medical Center and Grady Memorial Hospital. HIV and primary lung cancer histology were confirmed in 80 patients. Individual patient charts and laboratory values were reviewed for clinical data; specifically, age, sex, race, sexual orientation, intravenous drug use history, smoking exposure, the date of HIV and cancer diagnoses, antiretroviral (ARV) exposure, history of opportunistic infections, performance status, tumor pathology, American Joint Committee on Cancer AJCC Cancer Staging Manual (sixth edition) TNM stage at diagnosis, cancer treatment with associated complications, CD4 lymphocyte count, and HIV viral load at cancer diagnosis. Patient vital records were obtained from chart review and social security index. The protocol for this study was approved by the Institutional Review Board at Emory University.
Pearson correlation coefficients were estimated for each covariate studied, including age at HIV diagnosis, age at cancer diagnosis, latency (time from HIV diagnosis to cancer diagnosis), smoking, CD4 count at cancer diagnosis, cancer stage, and whether a patient received definitive treatment. CD4 count at diagnosis was treated as a continuous variable and also was subcategorized into discrete variables using 3 different cutoff points (50 cells/mL, 200 cells/mL, and 500 cells/mL). The Kaplan-Meier method was used for cancer-related survival analysis using either the date of death or the date of data censorship (December 29, 2009) if patients were still alive at the time of this analysis.
Eighty patients who had HIV and primary lung cancer were identified for analysis within the Atlanta VA Medical Center (n = 21) and Grady Memorial Hospital (n = 59). Table 1 shows the patient characteristics for the cohort. The majority of patients in this cohort were male (80%), African American (84%), and had a median age of 52 years (range, 28-73 years). Approximately 33% of patients (n = 23) were either homosexual or bisexual. Seventy-one percent of patients (n = 54) used recreational drugs, including marijuana, and 25% (n = 19) were intravenous drug users. Excessive alcohol consumption was noted in 61% (n = 49) of the study cohort. Twenty-one patients had detailed histories and consumed >4 alcohol-containing drinks per day. The remaining 28 patients had a diagnosis of alcohol abuse, but their detailed histories were not readily available. All 77 patients who had documented smoking histories were smokers with an average of 37 pack-years (range, 10-100 pack-years). Twenty patients were coinfected with hepatitis C, and 8 patients were coinfected with hepatitis B. In terms of other infectious comorbidities, past medical histories were significant for zoster (n = 16), latent tuberculosis (n = 15), disseminated Mycobacterium (n = 8), Clostridium difficile colitis (n = 5), and tuberculosis (n = 5). Twenty-seven patients were diagnosed clinically with recurrent bacterial pneumonias, and 21 were diagnosed with Pneumocystis jiroveci pneumonia. Patients were twice as likely to present with right-sided cancers as left-sided cancers, and >50% of primary tumors were located in the upper lobes. Seventy-three patients had nonsmall cell lung cancer (NSCLC), which included 38% adenocarcinoma (n = 30), 29% squamous cell cancers (n = 23), and 25% NSCLC unspecified (n = 20). Most patients presented with advanced stage disease (74% stage IIIB/IV vs 20% stage I/II); however, neither the CD4 count nor a previous history of AIDS was associated with advanced stage disease. Almost half of patients who had stage IV disease presented with multiple nodules. Of the 7 patients who had small cell lung cancer (SCLC), 5 presented with extensive stage disease.
Table 1. Patient Characteristics
Abbreviations: CA, cancer; HIV, human immunodeficiency virus; N, number of patients with available data; NOS, not otherwise specified; NSCLC, nonsmall cell lung cancer; SCLC, small cell lung cancer.
Age, N=80: Median [range], y
African Americans, N=80
Smoking, N=77: Mean pack-years [range]
77 (100): 37 [10-100]
Sexual orientation, N=69
Men who have sex with men
Intravenous drug use, N=76
Substance abuse—any, N=76
Alcohol abuse, N=80
Latency of HIV in to lung CA diagnosis, N=80: Mean [range], y
Disease stage at diagnosis
The mean latency from diagnosis of HIV to lung cancer was 7 years (range, 0-18 years) and was significantly shorter in women (4.1 years vs 7.7 years; P = .02). Eight patients were diagnosed with HIV at the time of their cancer diagnosis, whereas the rest of the cohort had a diagnosis of HIV that preceded their cancer diagnosis. Patients who were diagnosed with HIV at a younger age were more likely to have a longer latency (ρ = −0.47; P < .0001). Table 2 provides the HIV-related characteristics for patients in the cohort. At the time of cancer diagnosis, the mean CD4 count was 304 cells/mL (range, 3-1361 cells/mL), and the mean viral load was and 82,420 copies/mL (range, from <50 to >750,000 copies/mL). Smoking was correlated with a higher CD4 count (ρ = 0.27; P = .02). Fifty-nine patients (74%) had a previous diagnosis of AIDS, yet only 44 patients (55%) had previously been exposed to antiretroviral agents before being diagnosed with lung cancer. Twelve patients (15%) were started on ARVs after they were diagnosed with lung cancer. Exposure to antiretrovirals ranged from 1 to 12 regimens over a patient's lifetime and often involved modifications to 1 or 2 drugs within a regimen. Overall, patients received a median of 2 regimens with a maximum of 6 drugs at any given time. ARV therapy consisted mainly of a nucleoside reverse transcriptase inhibitors (NRTIs) alone or in combination with non-NRTIs (NNRTIs) or protease inhibitors (PIs). Only 2 patients received a combination that included either an integrase or a fusion inhibitor. The most common ARV regimens that a patient received included NRTI/NRTI/PI (n = 24), NRTI/PI/PI (n = 22), NRTI/NRTI (n = 20), NRTI (n = 15), NRTI/NRTI/NNRTI (n = 13), or NRTI/NNRTI/PI/PI (n = 7). Seventeen patients (27%) had an undetectable HIV viral load at the time of cancer diagnosis.
Table 2. Human Immunodeficiency Virus-Related Factors
Abbreviations: AIDS, acquired immunodeficiency syndrome; CA, cancer; N, number of patients with available data; VL, viral load.
CD4 at CA diagnosis, N=79: Mean [range], cells/mL
VL at CA diagnosis, N=61
Mean [range], copies/mL
82,420 [<50 to >750,000]
AIDS diagnosed before CA diagnosis, N=80
Antiretroviral use, N=80
Initiated before CA diagnosis
Initiated after CA diagnosis
Lung Cancer Treatment
Lung cancer treatment-related parameters for the cohort are summarized in Table 3. Thirty-one of 50 evaluable patients (62%) had an Eastern Cooperative Oncology Group performance status of 0 or 1. Fifty-three patients (71%) received some form of anticancer therapy. Among those who were treated, 11 patients (21%) underwent surgical resection, 15 patients (28%) received combined-modality therapy with chemotherapy and radiotherapy, 14 patients (26%) received chemotherapy alone, and 13 patients (25%) received palliative radiation only. Of those patients who underwent surgical resection, 1 had a wedge resection, 3 had a right-sided pneumonectomy, and 7 had a lobectomy. Poor performance status (n = 2) and acute infection (n = 2) prevented some patients from receiving treatment beyond palliative radiation alone, whereas others were not treated further either because they refused chemotherapy (n = 2) or for reasons that were not clearly documented (n = 7). Five patients declined all forms of anticancer therapy, and 17 patients (22%) were ineligible for treatment secondary to either poor performance status (n = 2), death before the initiation of treatment (n = 8), or reasons that were not clearly documented (n = 7). A greater proportion (76%) of treatment-ineligible patients had a CD4 count ≤200 cells/mL compared with patients who received definitive treatment (30%). Receiving cancer-directed therapy was associated significantly with the receipt of ARVs (P = .002) and a higher CD4 count (P = .02) irrespective of a previous history of AIDS or CD4 count nadir (Table 4). Treated patients had a mean CD4 count of 349 cells/mL versus a mean of 183 cells/mL in the untreated patients. Grade 3 or 4 toxicities occurred in 1 of 11 patients (9%) who underwent surgical resection, in 9 of 15 patients (60%) who received chemoradiation, and in 5 of 14 patients (36%) who received chemotherapy alone. One lobectomy was complicated by lung atelectasis, respiratory failure, and sepsis, resulting in end-stage renal disease. The most common complications included treatment delays secondary to neutropenia (n = 5), fever associated with neutropenia (n = 5), infections unrelated to neutropenia (n = 4), radiation-related esophagitis (n = 6), and acute renal failure (n = 4). One patient died of sepsis, acute renal failure, and respiratory failure after a dose of paclitaxel. Detailed treatment information is presented in Table 5. Patients who had a previous history of AIDS were more likely to experience treatment-related complications compared with patients who merely had a CD4 count ≤200 cells/mL. Five patients with NSCLC and 3 patients with SCLC received second-line chemotherapy for progression of disease. Only 1 patient received third-line chemotherapy for SCLC.
Table 3. Lung Cancer Treatment
Patients/Total No. (%)
Abbreviations: AIDS, acquired immunodeficiency syndrome; CA, cancer; ECOG, Eastern Cooperative Oncology Group; N, number of patients with available data; WBXRT, whole-brain radiation.
Fever with neutropenia, ARF, decline in PS, pneumonia
Cisplatin(50)/VP16(50) × 2 c + XRT(66Gy)
Cisplatin(75)/ Vp16(100) × 1 c->XRT(61 Gy)
Carboplatin/Paclitaxel × 4 c + XRT
Esophagitis requiring gastric tube
Carboplatin/Paclitaxel × 6 wkly c + XRT(66Gy)-> 3 c
Esophagitis, dehydration, pneumonia
Carboplatin(AUC 5)/Paclitaxel(135) × 1 c
Cisplatin × 4 wkly c + XRT
Carboplatin(AUC 5)/Paclitaxel(175) × 2 c + XRT(66 Gy)-> 2 c
Stage IV NSCLC
Carboplatin/Paclitaxel × 4 c
Cisplatin(30) wkly c +XRT(37.5 Gy)->Paclitaxel × 1 c
Esophagitis, ITP/hemoptysis, aspergillosis/VRE sepsis, ARF, respiratory failure and death
Carboplatin(AUC 6)/Paclitaxel(200) × 2 c
Carboplatin(AUC 2)/Paclitaxel(45) × 7 wkly c + XRT(57 Gy)
Prolonged neutropenia, esophagitis
Carboplatin/Paclitaxel × 4 c
Carboplatin/Paclitaxel × 1 c
Carboplatin/Paclitaxel × 5 c
Carboplatin(AUC 5)/Paclitaxel(175) × 4 c
Zoster, symptomatic anemia
Carboplatin/Paclitaxel × 4 c
Lung Cancer-Related Outcomes
The 1-year and 3-year survival rates were 31% and 4% overall, 25% and 0% for patients with stage IIIB/IV disease, and 53% and 8%, for all treated patients, respectively. Overall, the median survival for the entire cohort was 6.1 months. The median survival for patients with advanced stage (IIIB/IV) NSCLC was 4 months, which was significantly worse than the median survival for those who presented with earlier stage disease (P = .0013). Patients with stage I disease had an better survival rates compared with patients who had other stages of disease (Fig. 1). In a multivariate analysis that was adjusted for sex, race, age, latency, CD4 count, and smoking pack-years; early stage disease (P = .03) and definitive treatment (P = .0004) were associated with overall survival in patients with NSCLC (Table 6). CD4 count nadir, a history of AIDS, ARV use, and histology did not affect overall survival. Lung cancer survival was significantly better for patients who had CD4 counts >200 cells/mL (P = .0298) (Fig. 2) and >500 cells/mL (P = .0076) (Fig. 3), but not for patients stratified according to a CD4 count of 50 cells/mL (Fig. 4). When the CD4 count was separated in quartiles (<50 cells/mL, 51-200 cells/mL, 201-500 cells/mL, and >500 cells/mL), patients who had a CD4 count >500 cells/mL had an improved survival (Fig. 5). However, CD4 count in the multivariate analysis was not a significant predictor of survival when the analysis was adjusted for cancer stage. Patients who underwent surgery or who received chemoradiation had significantly better survival compared with patients who received palliative radiation or no treatment at all (P < .0001) (Fig. 6).
Table 6. Multivariate Analysis of Cancer-Related Overall Survival and Prognostic Factors
Hazard Ratio (95% CI)
Age at cancer diagnosis
0.993 (0.959, 1.029)
1.015 (0.997, 1.032)
0.998 (0.941, 1.058)
CD4 at cancer diagnosis
0.999 (0.998, 0.99999)
1.064 (0.521, 2.174)
1.054 (0.420, 2.643)
0.227 (0.079, 0.652)
1.436 (0.359, 5.745)
1.012 (0.492, 2.081)
No treatment or Palliative radiation
4.169 (1.893, 9.181)
In this large case series, we describe HIV patients with primary lung cancer in the post-HAART era. Consistent with other case series,15, 17-23 patients in this cohort were predominantly younger men who were smokers (median age, 52 years; 80% male; and all smokers), many of whom presented with adenocarcinoma (38%) and advanced stage disease (74% stage IIIB/IV). Although immunosuppressed patients have a higher risk of developing lung cancer,5, 8, 12, 13 HIV-infected patients who develop lung cancer often are characterized as having moderate immunosuppression. Although the majority of our patients (74%) had a previous diagnosis of AIDS, their mean CD4 count was 304 cells/mL at the time of cancer diagnosis. Several studies have failed to demonstrate a direct correlation between absolute CD4 count and lung cancer risk, which may be attributed in part to variability in the CD4 count.5, 10, 11, 15, 17, 24-28 In the current study, we demonstrated that smoking was associated with a higher CD4 count, as reported in other studies,29 supporting smoking as 1 of many factors that affect the absolute CD4 count.
Sex is another factor that may affect lung cancer risk. In the general population, women are more susceptible to the effects of smoking and to developing lung cancer. Emerging evidence suggests that estrogen may play a role.30-36 Women in our cohort had a significantly shorter latency from HIV to lung cancer diagnosis than men (4.1 years vs 7.7 years; P = .02), indicating that HIV-positive women also may be at greater risk for developing lung cancer. The Women's Interagency HIV Study supports the idea that women are at an increased risk for lung cancer irrespective of their HIV status.37 However, the retrospective nature of our study and its inherent biases may explain the difference observed in latency, especially if women either had a delay in HIV diagnosis or presented earlier with lung cancer.
Patients who were diagnosed with HIV at a younger age had a longer latency from HIV to lung cancer diagnosis. Although this may suggest that being HIV positive is a risk factor for lung cancer, the finding that 27% of the cohort had an undetectable viral load at cancer diagnosis suggests that this is likely a function of age rather than HIV infection itself. Ultimately, the duration of HIV positivity did not impact cancer-related survival. However, an analysis of the cohort revealed that cancer stage, CD4 count, and treatment all affected overall survival. Similar to the general population, advanced cancer stages are associated with a poorer prognosis. Although patients who had stage I disease clearly demonstrated improved survival compared with patients who had stage IIIB/IV disease, small patient numbers may explain the unexpectedly low survival of patients who had stage II lung cancer. Yet overall outcomes were poor: The 1-year survival rate was 25%, and the median survival was 4 months for patients with stage IIIB/IV disease. Although the median survival of 6.1 months for all patients is comparable to that in other reported case studies, it is considerably worse than the expected 10 to 12-month median survival of patients in the general population who have advanced stage cancer (IIIB/IV).38, 39
The CD4 count and cancer-directed treatment may account for some of the discrepancy in overall lung cancer survival observed between HIV patients in our cohort and the general population. A high CD4 count (>200/500 cells/mL) in this cohort was associated with improved lung cancer-related survival. However, when survival was analyzed in CD4 quartiles (<50 cells/mL, 51-200 cells/mL, 201-500 cells/mL, and >500 cells/mL), it appeared that a favorable outcome was observed only for patients who had CD4 counts >500 cells/mL, although the numbers of patients were small in each subset. Once the multivariate analysis was adjusted for stage, the CD4 count was no longer a significant predictor of survival, suggesting that cancer stage is a stronger predictor of outcome than CD4 count. Furthermore, patients who were treated had a significantly higher CD4 count (349 cells/mL) than those who were not treated (183 cells/mL). Brock et al demonstrated a similar correlation between the CD4 count and survival.15 Therefore, it appears that patients with a lower CD4 count are more likely to be deemed ineligible for definitive cancer-directed therapy because of the presence of either other comorbidities or a poor performance status. In our series, poor performance status or other comorbidities prevented 28 patients (37%) from being offered definitive therapy. Receiving definitive treatment had a greater impact on survival (P = .0004) in the multivariate analysis compared with receiving only palliative radiation or supportive care. Because systemic chemotherapy is associated with myelosuppression, it is conceivable that HIV-positive patients may be at greater risk for opportunistic infections during chemotherapy. This concern often has led to the use of suboptimal treatment regimens in this patient population. Furthermore, treatment data are limited to smaller case studies, and the optimal treatment of lung cancer in HIV-positive patients is yet to be determined. Of those who received chemotherapy or chemoradiation, approximately 33% (9 of 29 patients) developed an infection, and 1 resulted in death. However, our data suggest that eligible patients benefit in terms of improved survival from definitive treatment with manageable toxicities. A prior history of AIDS appeared to be more predictive of a treatment complication than merely having a CD4 count <200 cells/mL alone. However, patient numbers were small, which makes definitive conclusions difficult to assess.
Lung cancer is the most fatal non-AIDS-defining malignancy and frequently presents at an incurable stage.40 Although a CD4 count <500 cells/mL is associated with a worse prognosis, it may only reflect the reality that treatment options are limited by associated comorbid illnesses or a poor performance status. Future interventions need to focus on smoking prevention, early initiation of ARVs (for patients with CD4 counts <500 cells/mL), and a higher index of suspicion for lung cancer in patients who have suggestive symptoms. It also will be important to evaluate the prevalence of certain molecular abnormalities, such as epidermal growth factor receptor mutations, anaplastic lymphoma (ALK) translocation, K-ras, and p53 mutations. This will allow for the development of molecularly targeted therapeutic approaches for patients with HIV-related lung cancer. Prospective studies with commonly used therapeutic regimens are needed to further understand tolerability and efficacy outcomes. In summary, our study supports the use of stage-appropriate standard therapeutic options for lung cancer in HIV-positive patients who have a good performance status and who do not have significant comorbidities.
This work was supported by grant P01 CA116676 from the National Institutes of Health.