KS status is a composite, categorical outcome, ordered from worst to best as E1 (Failure: KS progression (PD), initiation of an alternate KS treatment, or no follow-up at Week 48 including death and missed visit), E2 (Stable: in follow-up at Week 48 with no KS PD nor response and without initiation of an alternate KS treatment) and E3 (Response: in follow-up at Week 48, with KS partial or complete response (PR or CR) and without initiation of an alternate KS treatment). Alternate KS treatment was defined as chemotherapy agent other than ET or other treatment triggered by worsening KS. KS outcome status (PR, stable, PR, CR) compared to study entry was evaluated at Week 48 based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002). Data on initiation of alternate KS treatment, loss to follow-up and dea

All eligible participants who initiated study treatment and had Study Week 48 data potential (i.e. participants enrolled at least 45 weeks prior to the date when the Data and Safety Monitoring Board's [DSMB] recommendation to close the study became public).

KS progressive disease (PD) compared to study entry based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002).

All eligible participants who initiated study treatment and had Study Week 48 data potential (i.e. participants enrolled at least 45 weeks prior to the date when the DSMB's recommendation to close the study became public) who had Week 48 KS exam performed and had not initiated alternate KS treatment before Week 48.

KS partial response (PR) compared to study entry based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002).

All eligible participants who initiated study treatment and had Study Week 48 data potential (i.e. participants enrolled at least 45 weeks prior to the date when the DSMB's recommendation to close the study became public) who had Week 48 KS exam performed and had not initiated alternate KS treatment before Week 48.

KS complete response (CR) compared to study entry based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002).

All eligible participants who initiated study treatment and had Study Week 48 data potential (i.e. participants enrolled at least 45 weeks prior to the date when the DSMB's recommendation to close the study became public) who had Week 48 KS exam performed and had not initiated alternate KS treatment before Week 48.

KS partial response (PR) or complete response (CR) compared to study entry based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002).

All eligible participants who initiated study treatment and had Study Week 48 data potential (i.e. participants enrolled at least 45 weeks prior to the date when the DSMB's recommendation to close the study became public) who had Week 48 KS exam performed and had not initiated alternate KS treatment before Week 48.

Premature study discontinuation by Week 48 due to any reason, including death.

All eligible participants who initiated study treatment and had Study Week 48 data potential (i.e. participants enrolled at least 45 weeks prior to the date when the DSMB's recommendation to close the study became public).

KS status is a composite, categorical outcome, ordered from worst to best as E1 (Failure: KS PD, initiation of an alternate KS treatment, or no follow-up at Week 96 including death and missed visit), E2 (Stable: in follow-up at Week 96 with no KS progression nor response and without initiation of an alternate KS treatment) and E3 (Response: in follow-up at Week 96, with KS PR or CR and without initiation of an alternate KS treatment). Alternate KS treatment was defined as chemotherapy agent other than ET or other treatment triggered by worsening KS. KS outcome status (PD, stable, PR or CR) compared to study entry was evaluated at Week 96 based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002). Data on initiation of alternate KS treatment, loss to follow-up and deaths are from entry through Week 96.

All eligible participants who initiated study treatment and had Study Week 96 data potential (i.e. participants enrolled at least 93 weeks prior to the date when the DSMB's recommendation to close the study became public).

KS progressive disease (PD) compared to study entry based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002).

All eligible participants who initiated study treatment and had Study Week 96 data potential (i.e. participants enrolled at least 93 weeks prior to the date when the DSMB's recommendation to close the study became public) who had Week 96 KS exam performed and had not initiated alternate KS treatment before Week 96.

KS partial response (PR) compared to study entry based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002).

All eligible participants who initiated study treatment and had Study Week 96 data potential (i.e. participants enrolled at least 93 weeks prior to the date when the DSMB's recommendation to close the study became public) who had Week 96 KS exam performed and had not initiated alternate KS treatment before Week 96.

KS complete response (CR) compared to study entry based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002).

All eligible participants who initiated study treatment and had Study Week 96 data potential (i.e. participants enrolled at least 93 weeks prior to the date when the DSMB's recommendation to close the study became public) who had Week 96 KS exam performed and had not initiated alternate KS treatment before Week 96.

KS partial response (PR) or complete response (CR) compared to study entry based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002).

All eligible participants who initiated study treatment and had Study Week 96 data potential (i.e. participants enrolled at least 93 weeks prior to the date when the DSMB's recommendation to close the study became public) who had Week 96 KS exam performed and had not initiated alternate KS treatment before Week 96.

Premature study discontinuation by Week 96 due to any reason, including death.

All eligible participants who initiated study treatment and had Study Week 96 data potential (i.e. participants enrolled at least 93 weeks prior to the date when the DSMB's recommendation to close the study became public).

KS progressive disease (PD) compared to study entry or best response based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002). Cumulative incidence was estimated with death and initiation of alternate KS treatment as competing risks. Time at risk was censored at the end of Step 1 (Week 96 or premature study discontinuation) or on the date when the DSMB's recommendation to close the study became public, whichever occurred earlier.

All eligible participants who initiated study treatment.

KS partial response (PR) or complete response (CR) compared to study entry based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002). Cumulative incidence was estimated with death and initiation of delayed KS treatment (alternate KS treatment or delayed ET in Arm A) as competing risks. Time at risk was censored at the end of Step 1 (Week 96 or premature study discontinuation) or on the date when the DSMB's recommendation to close the study became public, whichever occurred earlier.

All eligible participants who initiated study treatment.

KS partial response (PR) compared to study entry based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002). This was initially part of the outcome specified in the study protocol as "PR and CR combined and separately" to address the secondary objective on the KS response. Due to the extremely limited number of participants with KS complete response, the Statistical Analysis Plan was updated, and the Final Analysis was conducted only on the combined KS response. The outcome on PR separately was withdrawn.

Due to the extremely limited number of participants with KS CR, KS partial response (PR) and CR were combined. The analyses of CR and PR separately which were initially specified in the study protocol were withdrawn.

KS complete response (CR) compared to study entry based on clinical evaluation of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002). This was initially part of the outcome specified in the study protocol as "PR and CR combined and separately" to address the secondary objective on the KS response. Due to the extremely limited number of participants with KS CR, the Statistical Analysis Plan was updated, and the Final Analysis was conducted only on the combined KS response. The outcome on CR separately was withdrawn.

Due to the extremely limited number of participants with KS CR, KS partial response (PR) and CR were combined. The analyses of CR and PR separately which were initially specified in the study protocol were withdrawn.

Number of participants who experienced an AE (sign/symptom or laboratory abnormality) of Grade 3 or higher. The AEs were graded by the clinicians according to the Division of AIDS (DAIDS) AE Grading Table (see reference in the Protocol Section) as follows: Grade 1=Mild, Grade 2=Moderate, Grade 3=Severe, Grade 4=Potentially Life-Threatening.

All eligible participants who initiated study treatment.

KS-IRIS was defined as KS progressive disease that occurs within 12 weeks of initiation of ART that is associated with an increase in peripheral blood CD4+ lymphocyte cell count of at least 50 cells/mm^3 above the study screening value and/or a decrease in the HIV RNA level by at least 0.5 log10 below the study entry value prior to, or at the time of, documented KS progressive disease. Cumulative incidence was estimated with death and initiation of alternate KS treatment as competing risks. Time at risk was censored (1) when lost to follow-up, (2) at the study visit following Week 12 or (3) on the date when the DSMB's recommendation to close the study became public, whichever occurred earlier. Time of KS-IRIS was defined as the time of the initial KS progressive disease.

All eligible participants who initiated study treatment.

Etoposide (ET) was administered for a maximum of 8 cycles (16 weeks) from study entry (Arm B) or from Step 2 entry (Arm A). Dose modifications were reported as temporarily held, resumed at a different dose, deferred, prematurely discontinued and underdosed. The percentage of participants who experienced each dose modification is provided in the data table below. The categories are not mutually exclusive. A participant may have experienced multiple dose modifications and may be counted in more than one category. Each participant is counted at most once within category.

All eligible participants who initiated study treatment. Arm A participants are limited to those who entered Step 2 to initiate delayed ET.

HIV-1 RNA suppression was defined as plasma HIV-1 RNA <400 copies/mL. Only Arm A participants could enter Step 2 to initiate delayed ET.

All eligible participants who initiated study treatment and had HIV-1 RNA results available at the specific visit. HIV-1 RNA testing was done locally using Abbott RealTime HIV-1 Test or Roche AmpliPrep/Taqman HIV-1 Test. Only Arm A participants could enter Step 2 to initiate delayed ET.

ARV dose modifications were reported as temporarily held, prematurely discontinued and increased. The percentage of participants who experienced each dose modification is provided in the data table below. The categories are not mutually exclusive. A participant may have experienced multiple dose modifications and may be counted in more than one category. Each participant is counted at most once within category.

All eligible participants who initiated study treatment.

Absolute change in log10 HIV-1 RNA from entry at study visits calculated as value at a given time point minus value at entry. This outcome was initially specified in the study protocol. However, at the first post-entry visit, most participants had unquantifiable HIV-1 RNA levels. It would be misleading to calculate change from entry to HIV RNA-1 levels that could not be quantified. Therefore, the data collected did not support the outcome and the analytic method initially proposed in the study protocol, and this outcome measure could not be analyzed. The SAP updated the HIV-1 RNA outcome measure to HIV-1 RNA suppression at study visits (please refer to the secondary outcome measure #20).

At the first post-entry visit, most participants had unquantifiable HIV-1 RNA levels. Therefore, this outcome measure could not be analyzed.

Absolute change in CD4+ cell count was calculated as value at a given visit minus the value at study screening in Step 1, and as value at a given visit minus Step 2 entry in Step 2. Only participants in Arm A could enter Step 2 to initiate delayed ET.

All eligible participants who initiated study treatment and had CD4 cell count results available at screening and the respective Step 1 visit or at Step 2 entry and the respective Step 2 visit.

KS progressive disease (PD) compared to Step 2 entry (prior to initiation of delayed ET) or Step 2 best response based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002). Cumulative incidence was estimated with death and initiation of alternate KS treatment as competing risks. Time at risk was censored at the end of Step 2 (at up to 84 weeks or premature study discontinuation) or on the date when the DSMB's recommendation to close the study became public, whichever occurred earlier.

All Arm A participants who experienced KS progression and entered Step 2 to initiate delayed ET.

KS response, partial or complete (PR or CR) compared to Step 2 entry (prior to initiation of delayed ET) based on clinical assessment of KS cutaneous lesions (count, character and marker lesion area), oral KS, visceral KS and tumor-associated edema and as described in the publications (Krown et al 1989, Cianfrocca et al 2002). Cumulative incidence was estimated with death and initiation of alternate KS treatment as competing risks. Time at risk was censored at the end of Step 2 (at up to 84 weeks or premature study discontinuation) or on the date when the DSMB's recommendation to close the study became public, whichever occurred earlier.

All Arm A participants who experienced KS progression and entered Step 2 to initiate delayed ET.