80
Clinical Endpoints for Randomized Clinical Trials: All AIDS-defining Conditions Are Not Created Equal
Amanda Mocroft and Antiretroviral Therapy Cohort Collaboration
Royal Free and Univ Coll London Med Sch, UK
Background: Clinical endpoints for studies of combination ART
(cART) include the development of an AIDS-defining condition (ADC). However,
the mortality associated with ADC in patients treated with cART may vary according
to the specific diagnosis.
Methods: Data from 15 HIV cohort studies in Europe and North America were combined. Eligible patients were aged
>15, and were ART-naïve without a prior ADC when they started cART. We used Cox
models to estimate mortality hazard ratios (HR) for each ADC
that occurred in >50 patients, compared with mortality in the absence of an
ADC. The adjusted mortality hazard ratios (stratified by cohort and adjusted
for gender, HIV transmission group,
number of drugs started, regimen, age, year of starting HAART, CD4, and viral
load at starting cART) were used to rank the severity of the ADC. Relapses of
ADC were excluded.
Results:
Of 30,343 patients,
975 died and 2469 were diagnosed with ADC during a median follow-up of 42
months (IQR 19 to 69). Patients were mostly male (n = 21,909; 72.2%); the median date of starting cART was June 2000
(IQR July 1998 to October 2001). CD4 at starting cART was 252/mm3
(IQR 145 to 380). The 5 most common ADC were: tuberculosis (n = 379; 15.4%), esophageal candidiasis (n = 304; 12.3%), Kaposi’s sarcoma (n = 266; 10.8%), Pneumocystis
carinii pneumonia (n = 233; 9.4%), and non-Hodgkin’s
lymphoma (NHL) (n = 196; 7.9%). The
figure shows the adjusted mortality HR for each ADC, ranked according to
severity. Only herpes simplex virus was not associated with increased mortality
(HR 0.97). The greatest hazard of death
was following a diagnosis of NHL (HR 19.31), followed by progressive multifocal
leukoencephalopathy (PML) (HR 9.56), cryptococcosis (HR 9.00), toxoplasmosis (HR
5.10), and Mycobacterium avium
complex (MAC) (HR 5.07).
Conclusions:
The ADC used to
define AIDS were historically identified to track an emerging epidemic and were not
intended for use as a clinical staging system, yet they are frequently used for
this purpose. In the cART era, mortality rates following an ADC depended
strongly on the ADC diagnosed, although it was not possible to determine whether
patients died from the ADC they contracted. The ranking of severity of ADC
would be useful in design of clinical endpoint trials and for patient
management.
|
