The Role of Comorbidity in Long Term HIV Infection and Treatment Amy C. Justice, MD, PhD For the CHORUS and VACS Project Teams Support: National Institute on Alcoholism and Alcohol Abuse, National Institute on Aging, Robert Wood Johnson Foundation, National Institute for Mental Health, Veterans Affairs, GlaxoSmithKline Inc., Agouron, and Gilead Pharmaceuticals

Patient Outcomes in HIV in 2002 Aging Comorbid Medical and Psychiatric Disease HIV Disease HIV Treatment

HIV, Toxicity, and Comorbidity Easy to recognize event (anemia, hepatitis etc) Difficult to know cause HIV Toxicity from treatment for HIV “True Comorbidity” Comorbidity exacerbated by treatment Consider any combination of above

Functional Definition of Comorbidity Any condition not included in the CDC list of AIDS defining conditions

Motivation If comorbid conditions are a major determinant of survival or health related quality of life, they require careful clinical attention. If comorbid conditions are caused or exacerbated by HIV or its treatment, then clinical trials must account for these conditions as outcomes. Trials that fail to do so may underestimate either the benefits or the harms of treatment.

Hypotheses Comorbid conditions will be More prevalent than AIDS defining conditions More strongly associated with survival and health related quality of life than AIDS defining conditions Selected comorbid conditions will be associated with Advanced HIV disease (low CD4, high VL) HIV treatment (high CD4, low VL) Comorbid conditions may have important interactions

VACS 3 Subjects Prospective Consent Yes EMR Data & Surveys + Pharmacy Enrollment 99-00 Patients Enrolled 881 Deaths 71 Sites 3 VA Age (median years) 49 Predominant Race African American Predominant Risk IVDU, Heterosexual Women (n) 11 On ARV Treatment 89%

List of Comorbid Diagnoses Medical Diagnosis Charlson Dukes HIV Relevant Psychiatric Diagnosis Depression Severe Mental Illness Substance Use/Abuse Alcohol Drugs

Comorbid Diagnosis Validation No gold standard, measured several ways: Provider Report Patient Report Laboratory Data Diagnoses/Procedure Codes (ICD 9 Codes) Evaluated: Agreement, kappa Prevalence Strength of relationship with outcomes Survival HRQOL (SF 12 Summary Scores)

Example: Peripheral Neuropathy Patient report superior -Agreement: 72.9% Kappa: 0.22 -Prevalence higher -Strength of association stronger

Data Source for Comorbid Condition X designates sources of data, green background designates “best” source

Data Source for AIDS Condition X designates sources of data, green background designates “best” source

HIV and Comorbid Conditions (HIV in Red, Comorbidity in Blue)

Comorbidity and Outcomes After Adjustment for Age, CD4 Cell Count, and HIV-1 Viral Load

Association with Survival (HIV in Red, Comorbidity in Blue)

Association with Quality of Life (SF-12; HIV in Red, Comorbidity in Blue) 22.4

Comorbidity and Outcomes Among those on multidrug, multiclass ARV: Comorbid conditions are more prevalent than AIDS conditions Comorbid conditions are more strongly associated with survival and HRQOL than AIDS conditions

Comorbidity and HIV Laboratory Markers

Comorbidity and CD4 Cell Count *** * ** ** * ***p<0.01 **p<0.05 *p<0.1

AIDS Conditions and CD4 Count *** *** *** ** *** ** *** ** ** ** *** ***p<0.01 **p<0.05 *p<0.1

Comorbidity and HIV-1 Viral Load * * * *** ** * * ***p<0.01 **p<0.05 *p<0.1

AIDS Conditions and HIV-1 Viral Load * ***p<0.01 **p<0.05 *p<0.1

Comorbidity, CD4, and VL “Advanced HIV” (CD4<200 or VL>500) “Treatment Effects” (CD4>200 or VL<500) Transaminitis Hypertension Anemia Hyperlipidemia* Peripheral Neuropathy Coronary Artery Disease Renal Insufficiency Congestive Heart Failure Pancreatitis Cancer *While CAD and CHF were 38% correlated, hyperlipidemia was not highly correlated with either CAD or CHF <1% correlated.

Interactions Among Comorbid Conditions: Transaminitis and Anemia CHORUS and VACS analyses

Subjects CHORUS VACS 3 Prospective Consent Yes EMR Data & Surveys + Scripts + Pharmacy Enrollment 97-present 99-00 Patients Enrolled 5985 881 Deaths 400 71 Sites 4 3 Age (median years) 39 49 Predominant Race White African American Predominant Risk MSM IVDU, Heterosexual Women (n) 563 11 On ARV Treatment 88% 89% We used data from two diverse cohorts to answer these questions. Collaborations in HIV Research-US or CHORUS and the Veterans Aging 3 Site Study or VACS 3. This table summarizes these two cohorts. Of particular note, while CHORUS largely represents younger white men who have sex with men and a large number of women, VACS is predominated by older African American males with a history of intravenous drug use and heterosexual exposure.

Data Laboratory values closest to date of enrollment AST and ALT were standardized to laboratory normal >1 mean value above normal =1 are top normal <1 normal Laboratory values used for AST, ALT, CD4 cell count and HIV-1 RNA were those obtained closest to date of enrollment. Because normal values varied for ALT and AST, these were standardized to the highest value within normal range for the laboratory. To better approximate normal distributions, we used a square root transformation for CD4 cell count and log base 10 for HIV-1 RNA. Hemoglobin did not require transformation. Alcohol abuse was determined using diagnostic codes for alcohol abuse and dependence. Chronic hepatitis B was determined by a positive surface antigen, Hepatitis C by a positive qualitative laboratory test.

ALT, AST, and Survival Variable CHORUS VACS 3 HR* p AST/Top Normal 6.98 <0.001 2.29 0.02 ALT/Top Normal 3.15 0.002 0.18 0.9 AST/ALT 6.48 3.20 0.001 ALT and AST elevations were common in both cohorts. In CHORUS 32%, in VACS 42%, had an ALT or AST value greater than or equal to top normal. These were highly correlated. Because AST and ALT were correlated in both cohorts, we first used Cox proportional hazards modeling of survival to determine the indivudual hazard ratio for AST, ALT, and the AST/ALT ratio when each was separately entered into a model already adjusted for CD4 cell count, HIV-1 RNA, age, and hemoglobin. Variables listed in red were statistically significant in at least one cohort. Here and elsewhere we report standardized hazard ratios which represents the overall statistical importance of the variable in the model. Each of these measures of hepatocellular injury was significantly associated with survival in both cohorts except for ALT in VACS 3. AST and the AST/ALT ratio were more important than ALT in both cohorts. [[ Hemoglobin was less than 13 in 30.3% of VACS and16.7% of CHORUS ]] * ALT and AST were highly correlated in both cohorts. In CHORUS the correlation was 0.89, in VACS it was 0.79, p<0.0001. HR adjusted for CD4 and HIV-1 RNA. Proportional Hazards, HR standardized by dividing it by standard error, reflects statistical weight of variable in model.

Survival: Multivariable Model* CHORUS VACS 3 HR* p AST/Top Normal 6.40 <0.001 2.62 0.09 AST/ALT 3.08 0.002 1.50 0.1 Hemoglobin -6.40 -3.91 Hepatitis C 1.53 0.24 0.8 Chronic hepatitis B -1.35 0.2 -0.33 0.7 SQRT (CD4) -9.22 -1.46 Log (HIV-1 RNA) 5.82 3.71 Age 5.77 3.67 Here all variables were entered together in multivariable Cox proportional hazards models of survival. Variables listed in red were statistically significant in at least one cohort. AST independently predicts survival in CHORUS and demonstrated a strong trend at p=0.09 in VACS. Neither AST/ALT ratio nor ALT were independently predictive in either cohort after adjustment for AST. Neither hepatitis C nor chronic hepatitis B were associated with survival in either a univariate or a multivariate model in either cohort. * Proportional Hazards, HR standardized by dividing it by standard error, reflects statistical weight of variable in model. C Statistic for Models: 0.83, 0.78 respectively.

KM Survival Curves (Combined Cohorts) Hemoglobin AST/top nl p<0.001 p<0.001

Severity of AST Elevation and Anemia (Combined Cohorts) Hazard Ratio We next ran a Cox regression to determine the cumulative effect of AST elevation and anemia. Severity of AST elevation is listed on the x axis: Normal (<.5 top normal) on the left, Moderate (.5-2 fold top normal) in the middle, and Severe (>2 fold normal) on the right. Severity of anemia is listed along the z axis in differing colors, Normal (hemoglobin >13 mg/dl) in red, moderate (11-13 mg/dl) in peach, and severe (<11 mg/dl) in blue. We used 8 indicator variables for the combinations of AST and anemia. The comparison group (seen on your lower left) represents those with normal values for both. By definition the hazard ratio for this group is 1. All indicator variables demonstrated hazards ratios statistically greater than 1. Further, there is an increasingly poorer survival associated with each increment in severity for either condition. Thus, having severe AST elevation and severe anemia is substantially worse than having only severe AST elevation or severe anemia. And, having mild AST elevation with mild anemia is worse than either condition alone. Severity of Anemia Severity of AST Elevation AST values: Normal (<.5 top normal), moderate (.5-2 fold top normal), and severe (>2 fold normal) Hb values: Normal (>13 mg/dl); moderate (11-13 mg/dl), and severe (<11 mg/dl)

Limitations Exploratory (not definitive) analyses Prevalence only, incidence also needed Limited mortality events, other conditions may also be important after longer follow-up Only CHORUS has a number of women

Conclusions Comorbid conditions are a major determinant of Survival Health Related Quality of Life Given association with CD4 count and viral load, some comorbidity is likely caused/exacerbated by HIV HIV treatment Effects of comorbid conditions can be interactive Anemia and transaminitis are two of many comorbidities

Implications: Comorbid Conditions Require clinical attention and management —possibly with a team approach. Deserve study to determine whether the condition is caused or exacerbated by HIV and its treatment. Caused or exacerbated by HIV and its treatment must be considered an outcome in clinical trials --if optimal strategies are to be identified. These results have some important implications. CD4 and viral load alone do not describe enough of the variation in survival to be viable surrogate markers Transaminitis and anemia are major determinants of survival and require further investigation Comorbid medical disease, alcohol use, and drug toxicity are major determinants of transaminitis and anemia and may deserve increased clinical attention