CROI 2019 Abstract eBook

Abstract eBook

Poster Abstracts

Conclusion: Chemotherapy/radiation resulted in loss of CD4 that was associated with increased risk of non-AIDS death, independent of demographic characteristics, viral load, cancer type, and stage. There was some indication that the loss of CD4 was also related to increased hazard of new ADI but results were impacted by increased risk of death (i.e., competing risk analysis). This study illustrates the importance of evaluating specific therapies in this population with respect to immunosuppression so that the most appropriate option can be recommended.

289 AIDS- AND NON–AIDS-DEFINING CANCER INCIDENCE, 2010-2015, IN THE DAT’AIDS COHORT Camelia Protopopescu 1 , Antoine Chéret 2 , Alain Makinson 3 , David Rey 4 , Claudine Duvivier 5 , Clotilde Allavena 6 , Pascal Puglièse 7 , Tristan Ferry 8 , Thomas Huleux 9 , Pierre Delobel 10 , André Cabié 11 , Isabelle Lamaury 12 , Patricia Carrieri 1 , isabelle Poizot-Martin 13 , for the Dat’AIDS study group 1 INSERM, Marseille, France, 2 Hôpital Bicêtre, Le Kremlin-Bicetre, France, 3 CHU de Montpellier, Montpellier, France, 4 Hôpitaux Universitaires de Strasbourg, Strasbourg, France, 5 Necker Hospital, Paris, France, 6 CHU Hôtel-Dieu, Nantes, France, 7 Nice University Hospital, Nice, France, 8 Hospices Civils de Lyon, Lyon, France, 9 Centre Hospitalier de Tourcoing, Tourcoing, France, 10 Toulouse University Hospital, Toulouse, France, 11 CHU Fort de France, Fort de France, Martinique, 12 CHU de la Guadeloupe, Pointe à Pitre, France, 13 Assistance Publique–Hopitaux Marseille, Marseille, France Background: Although antiretroviral therapy has reduced the risk of developing AIDS-defining cancers, people living with HIV (PLWH) still have a high risk for some cancers, in particular virus-related. Given the increased life expectancy of PLWH, incidence of age and behavioral related cancers are expected to increase. However, data concerning recent incidence trends are scarce. We analyzed the data of a large French multicenter cohort to estimate incidences of AIDS defining cancers (ADC) and non-ADC (N-ADC) between 2010 and 2015. Methods: We performed a retrospective study using longitudinal data from the DAT’AIDS cohort from 01/2010 to 12/2015. Cases were identified using ICD-10 codes. For incidence assessment, prevalent cases, occurring within 30 days after enrollment in the cohort, were excluded. If more than one cancer occurred in the same patient during the study period, only the first case was considered in the analysis. We performed a focus on some N-ADC (breast, colorectal, prostate, anal, liver, lung, Hodgkin lymphoma (HL), bladder, head and neck). Results: Among the 44 642 HIV-infected people followed-up in the DAT’AIDS cohort during the study period (median age 43 [36-50] years, 69.7%male), 1440 cancer cases were diagnosed, including 358 ADC of which 345 were first cases (non-Hodgkin lymphoma: n= 194, Kaposi sarcoma: n= 135, and cervical cancer: n= 16). Among the 1082 N-ADC, 989 were first cases (76 patients were diagnosed with two different N-ADC during the study period, 7 with 3 and one with 4). Prostate cancer (n=111) was the most frequent N-ADC followed by liver (n=96), lung (n=90) and HL (n=82). Of note, head and neck cancer (n=66) was more frequent than anal cancer (n=53). Breast, colorectal and bladder cancer accounted for 54, 38 and 23 cases, respectively. The cancer incidence [95% CI] among the 44 642 patients accounting for 180 216.4 person-years (PY) between 2010 and 2015 was 191.4 [172.3-212.7] per 105 PY for ADC and 548.8 [515.6-584.1] per 105 PY for N-ADC. Incidence rates by calendar year and sex are reported in the table. Conclusion: The incidence of N-ADC remained relatively stable over the 2010-2015 period overall and for both sexes, whereas ADC incidence decreased. This study highlights the growing importance of prostate, and head and neck cancers.

Poster Abstracts

288 VACS INDEX BETTER PREDICTOR OF MORTALITY AFTER CANCER IN HIV+ AND HIV- THAN CHARLSON Keith M. Sigel 1 , Lesley S. Park 2 , Kimberly Stone 1 , Cynthia L. Gibert 3 , Matthew B. Goetz 4 , Roger Bedimo 5 , Roxanne Wadia 6 , Maria Rodriguez-Barradas 7 , Fatma Shebl 8 , Janet Tate 6 , Amy C. Justice 6 , Sheldon T. Brown 9 , Kristina Crothers 10 1 Icahn School of Medicine at Mt Sinai, New York, NY, USA, 2 Stanford University, Stanford, CA, USA, 3 Washington DC VA Medical Center, Washington, DC, USA, 4 VA Greater Los Angeles Health Care System, Los Angeles, CA, USA, 5 VA North Texas Health Care Center, Dallas, TX, USA, 6 VA Connecticut Healthcare System, West Haven, CT, USA, 7 Baylor College of Medicine, Houston, TX, USA, 8 Massachusetts General Hospital, Boston, MA, USA, 9 James J. Peters VA Medical Center, Bronx, NY, USA, 10 University of Washington, Seattle, WA, USA Background: Cancer is a leading cause of morbidity and mortality for people living with HIV (PWH). Mortality risk indices, like Veterans Aging Cohort Study (VACS) Index 2.0, that incorporate routine laboratory and diagnostic data may be useful for supporting clinical decision-making and assessment of prognosis. We evaluated VACS Index 2.0, a well-validated index among PWH, as a predictor of long-term survival for cancer patients, both with and without HIV infection. Methods: We linked VACS data to Veterans Affairs Cancer Registry data, identifying 7,087 patients (1,855 PWH and 5,232 uninfected) with primary prostate, lung, colorectal, liver cancer or lymphoma. For all subjects we collected demographic data, tumor staging and VACS index 2.0 values and calculated Charlson comorbidity index (CCI) scores at time of cancer diagnosis. We fit multivariable survival models for the cohort of all, and for individual, cancers with VACS Index 2.0 alone (adjusting for tumor stage and demographics) and then fit alternate models including tumor stage and CCI to determine the relative predictive value of these indices. Results: Patients did not differ by HIV status in age (median 52 years), sex (>99%male). PWH had higher median VACS index score at cancer diagnosis (61 vs. 37; p<0.001). For the cohort combining all cancer types in PWH and uninfected, the VACS index predicted overall survival in adjusted models with significant hazard ratios (HRs) for mortality for each quartile (Figure 1; all p<0.001) of the index, whereas the CCI had more limited predictive value. Adjusted models including the VACS index also had the best discrimination (c=0.82 versus c=0.73 for CCI model) and in models including both risk scores the VACS index was a strong independent predictor (p<0.001) CCI was of borderline significance (p=0.05). In models stratified by tumor stage, VACS index discriminated risk of mortality more effectively for early stage (I-II) than advanced cancers; in stage IV cases it was not associated with survival (all p>0.2). For individual cancers, the VACS index also predicted survival for both PWH and uninfected persons. Conclusion: The VACS index 2.0, a prognostic index accurately predicted cancer survival after accounting for cancer stage, outperforming a traditional comorbidity index for both PWH and uninfected Veterans with cancer.

CROI 2019 105