Characteristics of Studies.

Characteristics of included studies [ordered by study ID]
Bower 2009
Methods	Prospective cohort study (single centre)
Participants	HIV patients diagnosed for the first time with histologically confirmed, biopsy-proven KS since the HAART era commenced (defined as 1 January 1996) 96% men Mean age at KS diagnosis: 39 years 79 T1 KS patients out of 254 patients 48 of 79 T1 KS patients were receiving HAART for at least 3 months at the time of KS diagnosis
Interventions	HAART alone (163 patients: 131 were antiretroviral-naive and 32 were on HAART at the time of KS diagnosis) (5 T1) HAART + chemotherapy (liposomal anthracycline): 73 patients (68 T1) HAART + radiotherapy: 15 patients (5 T1) Palliative care (1 T1) Surgery alone (2 patients)
Outcomes	Overall survival at 5 years KS IRIS
Notes	This study was conducted in the UK. Median duration of follow-up was 4 years and maximum 12 years. (Letang 2013 included Bower's UK cohort of 213 patients of which 129 had T1 KS; Bower 2014 contains updated data from the same Bower 2009 cohort which has already been included in the review)

Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Not a randomised controlled trial
Allocation concealment (selection bias)	High risk	Not a randomised controlled trial
Blinding (performance bias and detection bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Blinding of outcome assessment (detection bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Incomplete outcome data (attrition bias) All outcomes	Low risk	No differential loss to follow-up
Selective reporting (reporting bias)	Low risk	No evidence of selective reporting
Other bias	Unclear risk	All T1 KS patients were meant to receive chemotherapy as per protocol, patients who received HAART alone were therefore exceptions

Cianfrocca 2010
Methods	RCT (individual)
Participants	There were a total of 49 T1 KS patients with 24 in the paclitaxel group and 25 in the pegylated liposomal doxorubicin group Inclusion criteria: Serologic diagnosis of HIV infection Biopsy-proven, measurable KS with any of the following features: progressive cutaneous disease symptomatic oropharyngeal or conjunctival lesions visceral involvement tumour-related lymphoedema ulceration pain Patients aged 18 years or above Eastern Cooperative Oncology Group (ECOG) performance status of 0, 1 or 2 Exclusion criteria: Prior systemic cytotoxic chemotherapy for KS Radiotherapy could not have been delivered to marker lesions and had to be discontinued 7 days before randomisation Pregnant or nursing women History of cardiac insufficiency (New York Heart Association Functional Classification of II or higher) Active, untreated infection Prior or concomitant malignancy (other than curatively treated in situ cervical carcinoma or basal/squamous cell skin carcinoma) Sensitivity to Escherichia coli (E. coli)-derived proteins that would preclude the use of granulocyte colony-stimulating factor (G-CSF) Patients were required to be receiving a stable antiretroviral drug regimen for at least 14 days before study enrolment
Interventions	Paclitaxel (100 mg/m2), infused intravenously over 3 hours every 14 days or pegylated liposomal doxorubicin (20 mg/m2), infused intravenously over 30 to 60 minutes every 21 days. At baseline, 53 of the 73 patients were receiving a combination HAART regimen containing either a protease inhibitor (N = 20), a non-nucleoside reverse transcriptase inhibitor without a protease inhibitor (N = 21) or both (N = 12)
Outcomes	Tumour response graded as complete response (CR), partial response (PR), progressive disease (PD) and stable disease (SD) at baseline and every 3rd cycle Quality of life Adverse events evaluated and graded using the NCI Common Toxicity Criteria (version 2)
Notes	The trial was terminated prematurely because of slow accrual

Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding (performance bias and detection bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Blinding of outcome assessment (detection bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Incomplete outcome data (attrition bias) All outcomes	Low risk	No significant differential loss to follow-up. Intention-to-treat analysis described
Selective reporting (reporting bias)	Low risk	There is no evidence of selective outcome reporting
Other bias	Unclear risk	Trial terminated prematurely due to slow accrual

Cooley 2007
Methods	Double-blinded, multicentre RCT
Participants	79 patients with AIDS-related KS. 46 patients had T1 disease with 34 in the PLD group and 12 in the liposomal daunorubicin group Inclusion criteria: AIDS-related KS requiring systemic chemotherapy Life expectancy of at least 120 days At least 5 measurable mucocutaneous KS lesions 1 or more of the following symptoms: KS-associated oedema that impaired functional activity of the extremities, groin or face Symptomatic evaluable pulmonary KS or gastrointestinal KS that had been documented by bronchoscopy and endoscopy respectively within 3 months of entering the trial, and was definitely associated with KS and not with any other manifestation of HIV disease KS-associated pain reported by the patient to be moderate or severe despite analgesic use KS lesions that, according to the patient, were disfiguring and that impaired self image or daily activities Patients were also required to have a left ventricular ejection fraction (LVEF) > 50%; a Karnofsky performance status (KPS) > 40%; and a haemoglobin concentration > 9 g/dl, neutrophil count > 1200 cells/mm3, a platelet count > 75,000/mm3 and bilirubin and creatinine levels less than 2 times the upper limit of normal Exclusion criteria: Patients who received anti-KS therapy within 14 days of study entry or had received treatment with pegylated liposomal doxorubicin or liposomal daunorubicin at any time prior to study entry Presentation with clinically significant cardiac disease as defined by histopathologic evidence of anthracycline-induced cardiomyopathy, LVEF < 50% or abnormal wall motion Onset of or increased therapy for an opportunistic infection within 4 weeks of study entry Presence of significant non-KS-related pulmonary insufficiency (oxygen saturation < 90%) Presence of other active malignancies except basal or squamous cell carcinoma of the skin or in situ cervical or anal carcinoma Neuropsychiatric history or altered mental status that prevented informed consent or compliance with the protocol requirements Pregnancy or breastfeeding, or any women of child-bearing age not using a medically proven method of birth control
Interventions	Pegylated liposomal doxorubicin (20 mg/m2; n = 60) versus liposomal daunorubicin (40 mg/m2; n = 20) as a 60-minute intravenous infusion every 2 weeks for 6 cycles Patients were assessed at < 30 days before treatment, at each of the 6 treatment cycles and at the end of the study 76 patients in the study received HAART
Outcomes	Tumour response: complete response (CR), partial response (PR), progressive disease (PD), stable disease (SD); time to progression; survival; adverse events recorded and graded using the National Cancer Institute Common Toxicity Criteria
Notes	This study was conducted in the USA

Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding (performance bias and detection bias) All outcomes	Low risk	An independent AIDS expert without the knowledge of the patient evaluated the outcome
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Blinding of outcome assessment (detection bias) All outcomes	Low risk	An independent AIDS expert evaluated the outcome
Incomplete outcome data (attrition bias) All outcomes	Low risk	There was no differential loss to follow-up
Selective reporting (reporting bias)	Low risk	There is no evidence of selective outcome reporting
Other bias	Low risk	No significant baseline differences between arms

Gill 1996
Methods	RCT (individual)
Participants	Inclusion criteria: Serologic documentation of HIV infection and biopsy-proven KS Advanced KS defined as: the presence of > 25 mucocutaneous lesions; symptomatic visceral involvement; or the presence of tumour-associated lymphoedema Age > 18 years Entry Karnofsky performance status (KPS) score > 70% No prior systemic chemotherapy Prior radiation or other local therapies (cryotherapy or intralesional vinblastine) had to be discontinued > 14 days before study enrolment Patients were required to have a left ventricular ejection fraction (LVEF) of > 45% measured by multiple-gated acquisition (MUGA) scan or by echocardiogram Adequate bone marrow function (absolute neutrophil count (ANC) > 1.5 x 109/l, platelet count > 75 x 109/l, and haemoglobin level > 85 g/l), renal function (serum creatinine level < 2.0 mg/dl) and hepatic function (bilirubin level < 1.5 times the upper limit of normal and AST< 3.0 times upper limit of normal) Exclusion criteria: Acute intercurrent infection Active symptomatic AIDS-defining opportunistic infection Symptomatic peripheral neuropathy Concurrent therapy with ganciclovir or systemic corticosteroids Other primary malignant tumours (except basal cell skin carcinoma or carcinoma in situ of the cervix)
Interventions	Patients received liposomal daunorubicin at a dose of 40 mg/m2 infused over 30 to 60 minutes, or a regimen of doxorubicin 10 mg/m2, bleomycin 15 U and vincristine 1 mg (ABV) every 2 weeks intravenously on an outpatient basis. Cycles were repeated every 14 days provided absolute neutrophil count (ANC) was >0.75 x 109/l and the platelet count >75 x 109/l. Patients whose ANC decreased to less than 0.75 x 109/l had chemotherapy withheld and could receive G-CS F to enable resumption of chemotherapy once the ANC had returned to > 0.75 x 109/l During the course of the trial, 48 liposomal daunorubicin patients (41%) received concomitant zidovudine therapy, 38 (33%) were treated with didanosine and 24 (21%) received zalcitabine. Among the ABV patients, 47 (42%) received zidovudine, 29 (26%) received didanosine and 22 (20%) were treated with zalcitabine
Outcomes	Tumour responses were categorised based on modified AIDS Clinical Trials Group (ACTG) criteria as follows: complete response (CR), partial response (PR), progressive disease (PD), stable disease (SD) Quality of life Time to progression Adverse events
Notes	This study was conducted in the USA

Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Permuted-block randomisation
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding (performance bias and detection bias)	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
All outcomes
Blinding of participants and personnel (performance bias)	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
All outcomes
Blinding of outcome assessment (detection bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Incomplete outcome data (attrition bias) All outcomes	Low risk	No differential loss to follow-up
Selective reporting (reporting bias)	Low risk	There is no evidence of selective outcome reporting
Other bias	Low risk	No significant baseline differences between arms

Grünaug 1998
Methods	Retrospective, cohort, multicentre study
Participants	29 AIDS patient with bronchoscopy or histologically confirmed pulmonary KS patients. All participants had T1 KS
Interventions	Group 1: (n = 20) stealth liposomal doxorubicin (SL-DOX) 20 mg/m2 every 2nd week Group 2: (n = 9) no SL-DOX (conservative management) Of the 20 patients in group 1, 17 had ARV therapy; of the 9 patients in group 2, 4 had bleomycin and vincristine or vinblastine while the remaining 5 patients had no chemotherapy
Outcomes	Survival, clinical response
Notes	This study was conducted in Munich, Germany

Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Not a randomised controlled trial
Allocation concealment (selection bias)	High risk	Not a randomised controlled trial
Blinding (performance bias and detection bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Blinding of outcome assessment (detection bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Incomplete outcome data (attrition bias) All outcomes	Low risk	No differential loss to follow-up
Selective reporting (reporting bias)	Low risk	No evidence of selective outcome reporting
Other bias	Low risk	No other potential sources of bias identified

Hernandez 1997
Methods	Prospective non-randomised trial
Participants	Inclusion criteria: Homosexual or bisexual men Aged 21 to 45 years Positive ELISA test for HIV Confirmatory biopsy for KS Patients clinically staged according to AIDS Clinical Trial Group (ACTG) criteria into “good risk” and “poor risk”
Interventions	10 patients with limited KS received alpha-2 interferon (5 million units subcutaneously 3 times weekly) plus zidovudine (AZT 500 mg daily) 24 patients with advanced KS, 12 in each group, received either intramuscular bleomycin 5 mg daily every 2 weeks for 3 days or low-dose doxorubicin (20 mg/m2), bleomycin (10 mg/m2) and vincristine (1.4 mg/m2, 2 mg max) every 3 weeks 10 “poor risk” patients received no treatment due to financial constraint Patients were followed until death
Outcomes	Clinical response: complete remission, partial remission, stable disease, progression; mortality; survival; adverse events
Notes	This study was conducted in Venezuela

Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Not a randomised controlled trial
Allocation concealment (selection bias)	High risk	Not a randomised controlled trial
Blinding (performance bias and detection bias)	Unclear risk	Blinding not described
All outcomes
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding not described
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Blinding not described
Incomplete outcome data (attrition bias) All outcomes	Low risk	No differential loss to follow-up
Selective reporting (reporting bias)	Low risk	No evidence of selective outcome reporting
Other bias	Unclear risk	No other potential sources of bias identified

Martin-Carbonero 2004
Methods	Randomised, open-label, multicentre study
Participants	Inclusion criteria: Biopsy-confirmed HIV-KS-affected patients without HIV-RNA control (naive, without HAART or with failing treatment) At least 10 cutaneous lesions or mucosal or visceral involvement Exclusion criteria: Life-threatening KS
Interventions	25 treatment and non-treatment-naive HIV patients with moderate to advanced KS randomly assigned to receive: Intravenous PLD (group A) administered at doses of 20 mg/m2 every 3 weeks compared with HAART alone (group B) 5 of 13 patients in group A and 5 of 15 patients in group B were ACTG T1 KS patients
Outcomes	Response rate: complete response, partial response, disease progression
Notes	This study was conducted in Spain. Duration of follow-up was 48 weeks

Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer-generated sequence
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding (performance bias and detection bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Blinding of outcome assessment (detection bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention-to-treat analysis done. No differential loss to follow-up
Selective reporting (reporting bias)	Low risk	There is no evidence of selective outcome reporting
Other bias	High risk	The prevalence of previous opportunistic infections was higher in patients allocated to the pegylated liposomal doxorubicin group

Mosam 2012
Methods	Open-label, prospective, single-centre RCT
Participants	Inclusion criteria: Treatment-naive, proven HIV and histologically confirmed KS (no prior KS or HIV therapy) Exclusion criteria: KS requiring urgent chemotherapy (i.e. symptomatic visceral disease or fungating lesions) Peripheral neuropathy Clinical congestive heart disease or ejection fraction < 50% Neutrophil count of < 1000 units per litre Haemoglobin < 9.0 gm/dl, platelet count of < 75 x 109 per litre Serum creatinine > 114.4 μmol/l Direct serum bilirubin > 85 μmol/l, aspartate aminotransferase or alanine aminotransferase > 2.5 times the normal range Intensive phase of tuberculosis therapy
Interventions	HAART alone or HAART and chemotherapy HAART arm received fixed-dose combination of stavudine (40 mg), lamivudine (150 mg) and nevirapine 200 mg Chemotherapy consisted of doxorubicin (20 mg/m2 IV), bleomycin (10 U/m2 IV) and vincristine (1.4 mg/m2 IV, capped at 2 mg) (ABV) every 3 weeks, started within the first month of initiation of HAART, with the goal of continuing chemotherapy for 2 cycles beyond maximal response
Outcomes	Primary outcome: Overall KS response using AIDS Clinical Trial Group criteria 12 months after HAART initiation Secondary outcome: Time to response Progression-free survival Adverse events HIV control CD4 reconstitution Adherence Quality of life KS responses were graded as complete, partial, stable disease and progressive disease using ACTG criteria (for ACTG KS Response Criteria) Toxicities were graded using the Division of AIDS (DAIDS) toxicity scale Adherence was assessed by a 7-day recall questionnaire at week 2, week 4 and then monthly Previously validated quality of life (QOL) questionnaires (EORTC QOL-30) measured 6 functioning scales
Notes	Although the planned chemotherapy protocol was ABV, oral etoposide (50 to 100 mg for 1 to 21 days of a 28-day cycle) was used as an alternative therapy in the event of difficulties with the chemotherapy drug supply or intravenous administration during the protocol. Etoposide was started at 50 mg daily but could be escalated to 100 mg in patients in subsequent cycles with inadequate KS tumour regression and no limiting toxicities This study included a mix of T0 and T1 participants, but only T1 disease was included in our analysis (data provided by communication with the author) Results from Bihl 2007 were included in this study Study was conducted in South Africa. Duration of follow-up was 12 months

Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Use of computed-generated random numbers
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding (performance bias and detection bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Blinding of outcome assessment (detection bias) All outcomes	Low risk	No blinding. However, the reported outcomes are objective and not likely to be affected by lack of blinding
Incomplete outcome data (attrition bias) All outcomes	Low risk	The authors used an intention to treat analysis. There was no differential loss to follow-up
Selective reporting (reporting bias)	Low risk	There is no evidence of selective outcome reporting
Other bias	High risk	There was some baseline imbalance. Patients randomised to CXT but not receiving chemotherapy had low baseline CD4 counts (median, 77 cells/ml), compared with those receiving chemotherapy (median, 249 cells/ml). When ABV was not available, oral Etoposide was given. This alternative chemotherapy regimen was used in 31% of patients that received chemotherapy overall

Olweny 2005
Methods	4-arm, randomised, open-label RCT
Participants	Inclusion criteria Histologically confirmed KS HIV-positive adults ECOG performance status of ≤ 3 Systemic symptoms such as fever, weight loss and diaphoresis Other symptoms including diarrhoea, dyspnoea, as well as past histories of pneumonia, sexually transmitted diseases and opportunistic infections Cachexia, lymphadenopathy, oral and palatal lesions, skin lesions, chest signs and opportunistic fungal infections Patients with stage III and IV disease Patients were not on any antiretroviral therapy
Interventions	Group 1: supportive care only Group 2: supportive care + single-agent oral etoposide at a dose of 100 mg (two 50 mg capsules) daily for 5 days, repeated monthly Group 3: supportive care + ABV: combination of actinomycin-D (2 mg/m2, IV every 4 weeks), bleomycin (15 mg/m2, IV every 4 weeks) and vincristine (1.4 mg/m2, IV every 4 weeks) Group 4: radiotherapy + supportive care. They were treated on 60-cobalt single field or parallel opposed pair of fields, depending on site and volume of disease There were a total of 178 T1 KS patients in groups 2 and 3, with 90 in the oral etoposide group and 88 in the ABV group. No participant received antiretroviral therapy
Outcomes	Primary outcome: Quality of life (QOL) Secondary outcomes: Tumour response: complete response (CR), partial response (PR), progressive disease (PD), stable disease (SD) Survival Toxicity QOL was measured by the functional living index-cancer (FLI-C) and supplemented by the Kaposi's sarcoma module (KSM)
Notes	Study was conducted in a resource-poor setting: Harare, Zimbabwe. Patients were followed up until death

Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer-generated randomisation cards
Allocation concealment (selection bias)	Low risk	Sealed envelopes
Blinding (performance bias and detection bias) All outcomes	Unclear risk	No blinding, reported primary outcome (QOL) is subjective. However, the other reported outcomes are objective and not likely to be affected by lack of blinding
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	No blinding, reported primary outcome (QOL) is subjective. However, the other reported outcomes are objective and not likely to be affected by lack of blinding
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No blinding, reported primary outcome (QOL) is subjective. However, the other reported outcomes are objective and not likely to be affected by lack of blinding
Incomplete outcome data (attrition bias) All outcomes	Low risk	The authors imputed missing data; no differential loss to follow-up
Selective reporting (reporting bias)	Low risk	There is no evidence of selective outcome reporting
Other bias	Low risk	Differences between arms were adjusted for in the analysis

ABV: doxorubicin, bleomycin and vincristine

AIDS: acquired immunodeficiency syndrome

ANC: absolute neutrophil count

ART: antiretroviral therapy ARV: antiretroviral

ACTG: AIDS Clinical Trial Group

AZT: zidovudine CXT: chemotherapy

ECOG: Eastern Cooperative Oncology Group

ELISA: enzyme-linked immunosorbent assay

G-CSF: granulocyte colony-stimulating factor

HAART: highly active antiretroviral therapy

HIV: human immunodeficiency virus

IRIS: immune reconstitution inflammatory syndrome

IV: intravenous

KPS: Karnofsky performance status

KS: Kaposi's sarcoma

LVEF: left ventricular ejection fraction

PLD: pegylated liposomal doxorubicin RNA: ribonucleic acid

QOL: quality of life

RCT: randomised controlled trial

SL-DOX: stealth liposomal doxorubicin