Incidence and correlates of HIV exclusion criteria in cancer clinical trials

In the era of modern antiretroviral therapy, exclusion of human immunodeficiency virus (HIV) seropositive patients from participation in cancer clinical trials has the potential to limit generalizability of trial results as well as curtail trial access equity.1-4 To better assess the incidence of exclusion criteria based on HIV status, we analyzed oncologic phase 3 randomized clinical trials (RCTs) and their eligibility requirements, with the goal of determining factors associated with HIV exclusion criteria (HEC) and analyzing HEC trends over time. Oncologic RCTs were identified through a search of ClinicalTrials.gov using the following parameters: Terms: “cancer”; Status: excluded “Not yet recruiting”; Phase: Phase 3; and Study Results: “With Results.” This yielded 1,239 trials, which were then screened for cancer-specific RCTs addressing a therapeutic intervention. Four hundred seventy-five trials were excluded, the majority of which were excluded for not being cancer-specific trials (302 of 475 excluded trials, 63.6%). For the remaining 764 included trials, information regarding enrollment criteria was collected from ClinicalTrials.gov, as well as the study protocol and primary publication of primary endpoint results, if available. Two individuals independently performed trial screening and data collection. Pearson's Chi-square tests and logistic regression analyses were performed using SPSS statistical software (Version 22.0). Seven hundred sixty-four trials met inclusion criteria, with a total combined enrollment of 462,449 patients. Of these 764 trials, 201 (26.3%) had overt exclusion criteria based on HIV seropositivity. Higher utilization of HEC was noted among industry-sponsored trials (30.7% vs. 13.7%, p < 0.001; Table 1), and lower HEC utilization was noted among cooperative-group-sponsored trials (18.1% vs. 29.8%, p = 0.001; Table 1). HIV exclusion was analyzed based on trial disease site; the highest incidence of HIV exclusion was observed among hematologic malignancy trials (47.8%), and lowest among breast cancer trials (13.7%, p < 0.001; Table 1). The analysis of HIV exclusion by disease site was repeated with disease site grouped based on the relationship to HIV or acquired immunodeficiency syndrome (AIDS). Trials addressing AIDS-defining malignancies (such as cancer of the uterine cervix) had the highest incidence of HEC (43.6%), followed by trials addressing non-AIDS-defining but HIV-elevated-risk cancers (such as Hodgkin lymphoma; 33.6%); these were both higher than the incidence of HEC for malignancies not associated with HIV/AIDS (23.9%, p = 0.002; Table 1).5 Most HIV-exclusionary trials were those addressing a systemic therapy (172 of 201 trials, 85.6%), and HEC incidence was higher among trials addressing a targeted therapy rather than cytotoxic chemotherapy (34.3% vs. 20.6%, p = 0.001; Table 1). The use of HIV exclusion criteria was not associated with either completion of trial accrual (p = 0.48; Table 1) or meeting the primary endpoint (p = 0.80; Table 1). Across all trials, utilization of HEC is increasing by an estimated 1.7% annually based on year of enrollment initiation (p < 0.001). Our study provides the first report of HIV exclusionary enrollment criteria incidence among oncologic RCTs; over one quarter of analyzed trials excluded patients on the basis of HIV seropositivity alone, and the use of HEC appears to be increasing over time. This trend is concerning since efforts to better define and refine HIV-specific enrollment criteria have been ongoing for over a decade.6 The association between industry sponsorship and HEC is particularly relevant given recent efforts to draft guidelines for industry as regards cancer clinical trial eligibility criteria and HIV status.7 Initiatives from the federal government in the United States, which at present are in draft form, encourage industry-sponsored trials to utilize more nuanced HIV-related eligibility criteria.5 These criteria evaluate patients’ immune function and antiretroviral therapy status to more thoughtfully define conditions and circumstances that preclude trial candidacy.1, 2, 7-9 Our data provide strong support for such initiatives in light of markedly higher incidence of exclusion by HIV seropositivity among industry-sponsored trials; future studies must continue to track the use of HEC within industry-supported trials to assess the impact of these initiatives after implementation. Increased utilization of HIV exclusion criteria among trials for HIV-associated or AIDS-defining malignancies is similarly noteworthy. We sought to better understand the basis for this observation, and correlate trials for these malignancies with other covariates. Industry sponsorship was more common among HIV/AIDS-associated cancer trials (82.1%) than other studies (74.2%, p = 0.03). Similarly, among systemic therapy trials, HIV/AIDS-associated studies were more likely to evaluate a targeted therapy than non-HIV/AIDS-associated studies (74.2% vs. 65.7%, p = 0.045). Given these correlations, we utilized a multiple binary logistic regression model to determine which trial factors predicted utilization of HEC. This model demonstrated preserved independent effects of industry funding (p = 0.04) and assessment of targeted systemic therapy (p = 0.005) on the use of HEC. No effect was seen for trials evaluating HIV/AIDS-associated malignancies (p = 0.12) or for cooperative-group-supported trials (p = 0.57). In a similar model including year of enrollment initiation, year of enrollment initiation (p = 0.03) and assessment of targeted therapy (p = 0.03) predicted HEC utilization, while industry sponsorship trended toward HEC utilization prediction (p = 0.068); as before, neither HIV/AIDS-associated disease site (p = 0.18) nor cooperative-group sponsorship (p = 0.28) predicted HEC use. Therefore, year of enrollment initiation, evaluation of targeted therapies and industry sponsorship, rather than HIV/AIDS-associated disease site, appear to independently predict HEC utilization. Our analysis is limited in that only strict HIV seropositivity exclusion criteria were examined. An additional 31 trials (of 764, 4.1%) were identified with enrollment criteria that could restrict participation of HIV-positive patients; these trials primarily included enrollment criteria excluding patients with “active infection.” These trials were not considered to have HEC for the purposes of our study. Exploratory analysis in which these 31 trials were included with the HEC cohort demonstrated unchanged conclusions from the above, including the effect of industry sponsorship on increased exclusionary enrollment criteria (34.0% vs. 19.8%, p < 0.001). Finally, we note that HIV seropositivity was required as part of eligibility criteria for one trial in our cohort (1/764, 0.1%). This trial (NCT01352117) evaluated the role of upfront cytotoxic chemotherapy with antiretroviral therapy for HIV-positive patients with AIDS-related Kaposi's sarcoma.10 Together, exclusionary enrollment criteria based on HIV seropositivity represent a pervasive challenge among oncologic trials, limiting generalizability of results. It is imperative that industry-sponsored studies and studies evaluating a targeted systemic therapy in particular utilize less restrictive eligibility criteria. The authors report no financial disclosures or conflicts of interest related to this work. Yours sincerely, Ethan B. Ludmir Andres F. Espinoza Amit Jethanandani Timothy A. Lin Walker Mainwaring Austin B.