Early progression and immune reconstitution inflammatory syndrome during treatment of mild-to-moderate Kaposi sarcoma in sub-Saharan Africa and South America: Incidence, long-term outcomes and effects of early chemotherapy

Abstract

Background

Early progression of AIDS-associated Kaposi sarcoma (KS-PD) and immune reconstitution inflammatory syndrome (KS-IRIS) sometimes occur after initiation of antiretroviral therapy (ART).

Methods

Early KS-PD and KS-IRIS were assessed in the A5264/AMC-067 trial in which participants with mild-to-moderate AIDS-KS were randomized to initiate ART with either immediate or as-needed oral etoposide. Early KS-PD was defined as tumor progression within 12 weeks of ART initiation. When investigators had concern that early KS-PD was KS-IRIS, additional evaluations were performed. Suspected KS-IRIS was defined as early KS-PD accompanied by a CD4+ count increase ≥50/mm³ or plasma HIV-1 RNA decrease ≥0.5 log₁₀ copies/mL. Clinical outcome was a composite endpoint categorized as Failure, Stable and Response at 48 and 96 weeks compared to baseline.

Results

Fifty of 190 participants had early KS-PD (27%): 28 had KS-IRIS, 22 were not evaluated for KS-IRIS. Early KS-PD and KS-IRIS incidences with immediate etoposide versus ART alone were 16% versus 39%, and 7% versus 21%, respectively. Week 48 clinical outcome was 45% Failure/18% stable/37% Response for no early KS-PD; 82% Failure/2% Stable/16% Response for early KS-PD; and 88% Failure/0% Stable/12% Response for KS-IRIS. Cumulative incidence of KS tumor response by week 96 was 64% for no early KS-PD, 22% with early KS-PD, and 18% with KS-IRIS.

Conclusion

Early KS-PD, including suspected KS-IRIS, was common after starting ART for AIDS-KS and was associated with worse long-term clinical outcomes. Starting ART concurrently with etoposide reduced the incidence of both early KS-PD and KS-IRIS compared to ART alone.

INTRODUCTION

While mild-to-moderate AIDS-related Kaposi sarcoma (AIDS-KS) often improves after starting antiretroviral therapy (ART), in some patients it worsens within a short period^1,2. Such early progression of KS (early KS-PD) may reflect the natural history of the disease, with failure of ART to influence its course. Alternatively early KS-PD may occur as a consequence of KS immune reconstitution inflammatory syndrome (KS-IRIS), which is characterized by increased inflammation of KS lesions and/or severe tumor-associated edema in the setting of suppressed HIV viremia and/or increased CD4+ count following ART^3,4. Although KS-IRIS has been associated with increased mortality in African retrospective cohorts², few studies have prospectively evaluated the outcomes of KS-IRIS and none have evaluated early KS-PD.

The randomized clinical trial, A5264/AMC067, found that initiating ART together with oral etoposide for mild-to-moderate KS provided early but non-durable clinical benefits, including reduced risk of suspected KS-IRIS⁵, in people living with HIV (PLWH) in low- and middle-income countries⁵. A5264/AMC067 included prospective definitions of KS-PD and suspected KS-IRIS. While prior studies have focused on KS-IRIS, we noted that KS-IRIS is a subset of all early KS-PD and clinical differentiation of KS-IRIS from other causes of early KS-PD is difficult. Therefore, we considered KS-IRIS events meeting our operational criteria to be “suspected”. Our analyses included assessments of both early KS-PD and suspected KS-IRIS. Our aims were to evaluate incidence, baseline risk factors, and long term clinical outcomes of early KS-PD and suspected KS-IRIS, and to compare participants with protocol-defined suspected KS-IRIS and participants with other early KS-PD that was not suspected to be KS-IRIS with respect to baseline characteristics and virologic and immunologic responses to ART.

METHODS

Study design, participants and outcome definitions

A5264/AMC067 was a randomized open-label clinical trial that compared two treatment strategies: ART with concurrent low-dose oral etoposide (ART+ET arm) versus ART with low-dose oral etoposide later, if needed (as–needed arm). In the as-needed arm, etoposide could be started between study weeks 8 and 80 after confirmed KS-PD occurred on ART alone. Eligible participants had confirmed HIV-1 infection, age ≥18 years, ≤14 days of ART use prior to enrolling to the study and biopsy-confirmed KS. Mild-to-moderate KS was defined as stage T0 (confined to skin and/or lymph nodes and/or minimal oral KS) and some presentations of stage T1, aligning with the WHO definition of mild-to-moderate KS⁶. Additional details of the study design and procedures were described previously⁵.

KS response criteria, including complete response, partial response, stable, and KS-PD were defined in the A5264/AMC067 protocol⁵ according to published criteria^7,8. KS-PD was defined as 25% or more increase in cutaneous lesion number, the number of skin lesions that were raised above the skin surface, or cutaneous lesion surface area; or new oral or visceral sites of involvement; or progression of existing oral or visceral disease; or the development of new or increasing tumor-associated edema or effusion that interfered with normal activities lasting for at least two consecutive evaluations. All KS-PD was confirmed by an independent endpoint review committee. Suspected KS-IRIS was defined in the protocol as KS-PD within 12 weeks of ART initiation accompanied by CD4+ increase of ≥50 cells/mm³ above the baseline value and/or plasma HIV RNA by ≥0.5 log₁₀ copies/mL decrease below the baseline value. The protocol instructed investigators to consider KS-IRIS for all worsening of KS lesions within the first 12 weeks of ART. However, evaluations for suspected KS-IRIS IRIS (CD4+ count and plasma HIV-1 RNA) were performed only if KS-PD occurring in the first 12 weeks was judged by the site investigator to potentially be KS-IRIS. Early KS-PD was defined post hoc for this analysis as any KS-PD within 12 weeks of ART initiation. Thus, KS-IRIS cases were a subset of all early KS-PD. Given that by these definitions, both early KS-PD and KS-IRIS occurred within the first 12 weeks of treatment and noting that etoposide was initiated in the as-needed arm only after KS-PD, the as-needed arm is considered “ART-alone” in analyses of early KS-PD and suspected KS-IRIS.

The primary study endpoint for ACTG-A5264/AMC-067 was a composite endpoint that compared week 48 outcomes between treatment arms, ordered from worst to best as follows. Failure: KS-PD compared to study entry, initiation of chemotherapy other than etoposide, death or loss to follow-up; Stable: without KS-PD, without partial response, and without complete response compared to study entry, and without initiation of new chemotherapy other than etoposide; Response: partial or complete response compared to study entry, without initiation of chemotherapy other than etoposide. Week 96 clinical outcome was a secondary endpoint defined as for week 48. Losses to follow-up were categorized as failures, assuming the worst outcome. Initial response to treatment was defined as the first instance of partial or complete response during study follow-up which persisted for at least 4 weeks and was also used as the clinical outcome for the analyses presented in this work.

The study provided co-formulated tenofovir disoproxil fumerate/emtricitabine/efavirenz (TDF/FTC/EFV) as single tablet antiretroviral regimen; 89% of participants were treated with this regimen, 11% received TDF/FTC+EFV as a two-tablet regimen, and <1% received a boosted protease inhibitor regimen.

Study oversight and conduct

The study was performed in accordance with the Declaration of Helsinki and approved by site-specific review boards or ethics committees. Each participant provided a written informed consent. The study was funded by the National Institute of Allergy and Infectious Disease (NIAID) through the AIDS Clinical Trials Group (ACTG) with additional funding from the National Cancer Institute through the AIDS Malignancy Consortium (AMC) and registered with Clinicaltrials.gov NCT01352117.

Statistical Analysis

For comparisons between groups, Wilcoxon rank sum tests were used for continuous and ordinal measures, and Fisher’s exact tests were used for binary or categorical measures. Confidence intervals for proportions were calculated using the Clopper-Pearson method.

Cumulative incidences of early KS-PD, suspected KS-IRIS and initial KS response were estimated using Kaplan-Meier methods. Gray’s approach was used to compare cumulative incidence distributions between groups to model the initiation of non-study KS treatment and death as competing risks⁹. Delayed etoposide initiation was also modeled as a competing risk for the initial KS response analysis as conducted for the primary analysis of the study⁵. Baseline factors associated with time to early KS-PD (and separately, suspected KS-IRIS) were assessed using sub-distribution hazards¹⁰ and cause-specific hazards models. Results were similar in both, and only the results from the cause-specific Cox-proportional hazard regression models are presented. Baseline factors for the analysis and all-subsets variable selection approach are as described elsewhere⁵.

Analyses of clinical outcome at weeks 48 and 96 were restricted to participants who had week 48 or week 96 data potential, respectively, by the date when early study closure was announced⁵. Clinical outcomes at 48 and 96 weeks were based on the clinical status at these time points, irrespective of progression or response at earlier time points; week 48 and 96 outcome did not preclude someone who had an early response from having failure at week 48 or 96. The van Elteren test with stratification by study arm was used to compare groups. All tests were conducted as two-sided at 5% significance using SAS software, version 9.4 (SAS Institute Inc., Cary, North Carolina).

RESULTS

Incidence, characteristics and outcomes of early KS-PD

Most participants were men (71%) and African (93%); median age was 34 years (IQR 29, 41) and median CD4 count was 184 cells/mm³ (IQR 78, 325). Early KS-PD occurred in 50 of 190 participants (12-week cumulative incidence 27.3%; 95% CI, 21.0–33.9%; Figure 1). Participants in ART-alone had higher incidence of early KS-PD (39.0%; 95% CI, 28.8–49.0%) than those in ART+ET (16.0%; 95% CI, 9.4–24.2%; P <0.001; Table 1).

Figure 1:

Outcomes of early Kaposi sarcoma progressive disease and Kaposi Sarcoma immune reconstitution inflammatory syndrome after initiation of antiretroviral therapy with immediate or as-needed chemotherapy. Abbreviations: ART, antiretroviral therapy; IRIS, immune reconstitution syndrome; KS, Kaposi Sarcoma.

TABLE 1.

Cumulative incidence of early progressive Kaposi’ sarcoma (KS) and Suspected Kaposi’ sarcoma associated immune reconstitution inflammatory syndrome (KS-IRIS) after 12 weeks of ART

		Early KS Progression			Suspected KS-IRIS
Study Arm	N	Events	Cumulative Incidence, % (95% CI)	P*	Events	Cumulative Incidence, % (95% CI)	P*
Overall	190	50	27.3 (21.0, 33.9)		28	15.1 (10.4, 20.7)
ART-alone	94	35	39.0 (28.8, 49.0)	< 0.001	21	23.1 (15.0, 32.3)	0.003
ART+ET	96	15	16.0 (9.4, 24.2)		7	7.4 (3.2, 13.9)

ART, antiretroviral therapy; ET, etoposide

^*Comparison of ART alone and ART with immediate ET arms by Gray’s test stratified by screening CD4+ lymphocyte count.

Participants who developed early KS-PD had lower BMI (21.4 kg/m² vs 22.0 kg/m²; P=0.029) and hemoglobin (11.3 g/dl versus 12.2 g/dl; P=0.016) than those who did not have early KS-PD (Supplemental Digital Content Table S1). In simple Cox regression models that adjusted only for treatment arm, BMI <20kg/m² (adjusted hazard ratio, aHR 2.5; 95% CI, 1.4–4.4), albumin ≤3.0 g/dL (aHR 2.0; 95% CI, 1.2–3.6), hemoglobin <10g/dl (aHR 2.6; 95% CI, 1.4–4.8), Karnofsky performance score <90 (aHR 1.9; 95% CI, 1.1–3.4) were associated with increased hazards of early KS-PD. In addition, age, sex, baseline CD4 and baseline chest X-ray abnormality were included in the full Cox regression model. In the final reduced model, ART alone, BMI <20 kg/m² and hemoglobin <10 g/dl remained associated with increased hazards of early KS-PD (Figure 2A); age ≥35 years, female sex and screening CD4+ cell count <200 cells/mm³ were associated with decreased hazards for early KS-PD (Figure 2A).

Figure 2:

Final multivariate Cox proportional hazards model for associations between baseline factors and development of (A) Early Kaposi Sarcoma (KS) progressive disease; (B) Suspected Kaposi sarcoma immune reconstitution syndrome (KS-IRIS).

Among participants with data potential at weeks 48 and 96 (N = 163 and 119, respectively; Table 2), participants with early KS-PD (N = 45 and 38, respectively) had higher rate of Failure at weeks 48 (82% versus 45%; P < 0.001) and 96 (71% versus 48%; P=0.04) compared to those without early KS-PD. There were similarly low rates of loss-to-follow-up in both groups, and the higher rates of the composite Failure endpoint among early KS-PD were driven by differences in the other components (progressive disease, initiation of alternative KS treatment and death). The cumulative incidence of KS complete or partial tumor response was 22% with early KS-PD compared to 64% without early KS-PD (Table 2; P<0.001).

TABLE 2.

Kaposi’ sarcoma clinical treatment outcomes by early progressive Kaposi’ sarcoma (KS) and suspected Kaposi’ sarcoma associated immune reconstitution inflammatory syndrome (KS-IRIS) status

		Early KS Progressive Disease			Suspected KS-IRIS
Clinical Outcome (%)		Yes (N=45)	No (N=118)	P	Yes (N=25)	No (N=138)	P
48 weeks	Failure**	37 (82)	53 (45)	< 0.001*	22 (88)	68 (49)	< 0.001*
	Progressive disease	14	14		11	17
	Initiation of alternative KS treatment	17	26		9	34
	Loss-to-follow-up	1	5		1	5
	Death	7	11		3	15
	Stable	1 (2)	21 (18)		0 (0)	22 (16)
	Response	7 (16)	44 (37)		3 (12)	48 (35)
		Yes (N=38)	No (N=81)		Yes (N=20)	No (N=99)
96 weeks	Failure**	27 (71)	39 (48)	0.040*	15 (75)	51 (52)	0.09*
	Progressive disease	3	9		3	9
	Initiation of alternative KS treatment	16	21		9	28
	Loss-to-follow-up	1	3		1	3
	Death	9	11		4	16
	Stable	0 (0)	4 (5)		0 (0)	4 (4)
	Response	11(29)	38 (47)		5 (25)	44 (44)
		Yes (N=50)	No (N=140)		Yes (N=28)	No (N=162)
Cumulative incidence of KS response by week 96, % (95% CI)		22 (11, 35)	64 (52, 74)	<0.001†	18 (5, 38)	59 (48, 68)	0.007†

^*Exact Wilcoxon test stratified by study arm for the ordinal, composite clinical endpoint.

^†Gray’s test stratified by study arm

^**The components of Failure do not add up to the Failure number, because participants could have initiated alternative KS treatment prior to loss to follow-up or death.

Of the 190 participants in the trial, 89 (47%) had grade 3 or 4 adverse events. Among participants with early KS-PD, the proportion of participants who experienced grade 3 or 4 sign/symptom events, grade 3 or 4 laboratory events, or new diagnoses was 52%, 46% and 70%, respectively, compared to 19%, 30% and 44% without early KS-PD (Supplemental Digital Content Table S2).

Incidence, characteristics and outcomes of KS-IRIS

Twenty-eight participants with early KS-PD were evaluated for KS-IRIS; all met criteria for suspected KS-IRIS (cumulative incidence 15.1%; 95% CI 10.4–20.7%). At the time of early KS-PD, all 28 had plasma HIV-1 RNA decrease ≥0.5 log₁₀ copies/mL and 18 had CD4+ increase ≥50 cells/mm³ above baseline values. Participants in ART-alone had a higher incidence of KS-IRIS (23.1%; 95% CI, 15.0–32.3%) than those in ART+ET (7.4%; 95% CI, 3.2–13.9%; P=0.003; Table 1).

Participants who developed suspected KS-IRIS were more likely to have edema interfering with physical function and raised cutaneous lesions at baseline (Supplemental Digital Content Table S1). There were no differences in age, sex, CD4 count, ethnicity and country of origin. In simple models using KS-IRIS as the outcome, baseline BMI < 20kg/m² (aHR 2.3; 95% CI, 1.1–4.9), albumin ≤ 3.0 g/dL (aHR 2.3; 95% CI, 1.1–4.9), edema interfering with physical function (aHR 2.9; 95% CI, 1.1–7.6) and raised cutaneous lesions (aHR 3.3; 95% CI, 1.0–11.0) were associated with increased hazards of KS-IRIS. In addition, sex and Karnofsky score were included in the full model. In the final reduced model, ART alone, albumin ≤3.0, Karnofsky score <90, and raised cutaneous lesions were associated with increased hazards of developing KS-IRIS (Figure 2B).

Among participants with week 48 data potential, participants who developed KS-IRIS N=25) had a higher rate of Failure at week 48 (Table 2; 88% versus 49%; P < 0.001). Among participants with week 96 data potential, 75% of participants with KS-IRIS (N=20), compared to 52% of participants without KS-IRIS (P=0.09), had Failure. There were similarly low rates of loss-to-follow-up in both groups at both weeks 48 and 96. The cumulative incidence of treatment response was 18% with KS-IRIS compared to 58.9% (48%−68%) without KS-IRIS (Table 2; P=0.007).

The proportion of participants who experienced grade 3 or 4 sign/symptom events, grade 3 or 4 laboratory events, or new diagnoses was 46%, 43% and 64%, respectively, among participants with KS-IRIS compared to 24%, 33% and 49% without KS-IRIS (Supplemental Digital Content Table S2).

Comparison of non-IRIS early KS-PD to KS-IRIS

The baseline characteristics of the 22 participants with early KS-PD who were not evaluated for the protocol-specified definition of KS-IRIS (Figure 1) were not significantly different from the characteristics of the 28 participants with suspected KS-IRIS (Table 3), except that those with suspected KS-IRIS were more likely to have raised cutaneous KS lesions (P = 0.042). Participants with KS-IRIS were also less likely to be hepatitis B surface antigen positive (0% versus 18%; P = 0.035) compared to those with non-IRIS KS-PD. The median CD4+ count change from baseline to week 12 was +82 (Q1, Q3: 23, 129) cells/mm³ in participants with KS-IRIS versus +47 (Q1, Q3: 13, 99) cells/mm³ in participants with early KS-PD who were not evaluated for KS-IRIS; 64% and 47% had CD4+ cell increases ≥50 cells/mm³, respectively. Median plasma HIV-1 RNA decreases at week 12 were 3.2 (Q1, Q3: 2.6, 3.6) log₁₀ copies/mL and 3.2 (Q1, Q3: 2.6, 3.8) log₁₀ copies/mL among participants with suspected KS-IRIS versus those with non-IRIS early KS-PD; all participants in each group had plasma HIV-1 RNA decrease ≥0.5 log₁₀ copies/mL. Of the 22 participants with early KS-PD who did not have protocol-specified KS-IRIS evaluations performed, for 12 participants early KS-PD occurred at the week 12 visit which coincided with scheduled CD4+ cell count and plasma HIV-1 RNA measurements. All 12 participants had CD4+ cell increase or plasma HIV-1 RNA decrease at week 12 that met criteria for suspected KS-IRIS.

TABLE 3.

Baseline characteristics by suspected Kaposi’ sarcoma (KS) associated immune reconstitution inflammatory syndrome (KS-IRIS) outcomes among participants with early progressive KS

	Suspected KS-IRIS
Variable*	Yes (N=28)	No (N=22)	P**
Age, years	33 (29, 37)	31 (28, 39)	0.91
Female sex	4 (14)	7 (32)	0.18
Black	26 (93)	21 (96)	1.00
Karnofsky Score			1.00
60	0 (0)	1 (5)
70	4 (14)	3 (14)
80	10 (36)	6 (27)
90	11 (39)	10 (46)
100	3 (11)	2 (9)
Body Mass Index, kg/m2	21.1 (19.4, 23.0)	21.4 (18.7, 22.9)	0.90
CD4+ lymphocytes/mm3	236 (80, 364)	251 (115, 340)	0.98
Plasma HIV-1 RNA, log10 copies/mL	5.19 (4.68, 5.57)	5.25 (4.53, 5.81)	0.98
Serum albumin, g/dL	3.5 (2.9, 3.9)	3.6 (3.1 3.9)	0.74
Neutrophils/mm3	2,235 (1,335, 2,989)	2,085 (1,570, 2,530)	0.95
Hemoglobin, g/dL	11.2 (9.8, 13.5)	11.4 (9.9, 12.7)	0.84
Abnormal chest X-ray	6 (21)	8 (38)a	0.22
Edema interfering with physical function	5 (18)	1 (5)	0.21
T1 ACTG stage	18 (64)	16 (73)	0.56
Oral KS present	14 (50)	16 (73)	0.15
>50 cutaneous KS lesions	12 (43)	12 (55)	0.57
Raised cutaneous lesions	25 (89)	14 (64)	0.04

^*Median with first and third quartile (Q1, Q3) for continuous variables. N (%) for categorical variables.

^**Fisher’s exact test for categorical variables; Exact Wilcoxon test for continuous and ordinal variables.

^aN=21

DISCUSSION

The reductions in early KS-PD and KS-IRIS in the ART+ET arm demonstrate the potential benefits of initiating chemotherapy and ART together for people with mild to moderate AIDS-KS. Both early KS-PD and suspected KS-IRIS were associated with higher frequencies of clinical and laboratory adverse events as well as higher proportions of clinical failure at week 48 and 96 and lower cumulative incidences of treatment response. To evaluate whether burden of disease at baseline was predictive of KS-PD or KS-IRIS, we investigated associations with T0/T1 stage, total cutaneous KS lesion count and the presence of tumor associated edema, raised cutaneous lesions and oral KS. While some of these measures had weak associations with early KS-PD and/or KS-IRIS in simple models, only the association of raised cutaneous lesions with KS-IRIS was retained in the final models. Not surprisingly, the final multivariate models also found that measures of poor overall health status (low BMI, anemia, hypoalbuminemia, and low Karnofsky performance scores) were associated with increased hazards of early KS-PD or KS-IRIS. In a prior ACTG study conducted at many of the same sites that participated in this study, younger age was independently associated with increased risk of antiretroviral treatment failure¹⁵. So it is plausible that age effects on antiretroviral efficacy explain the protective effect of older age on occurrence of early KS-PD.

Low baseline CD4 count was associated with decreased hazards of early KS-PD in this analysis and with decreased time to initial progression in our primary trial analysis⁵. Similarly, an association between higher CD4 count and increased KS-IRIS risk was detected in a prior study ³. The relationships between higher CD4+ count and early KS-PD and KS-IRIS suggest that the capacity to develop a robust immune response is an important factor for progression of KS soon after ART initiation. Since, in general, CD4+ lymphocyte count is inversely related to the duration of HIV infection, lower risk of early disease progression could mean that individuals with lower CD4 counts simply had longer-standing HIV infection with longer-standing and less active KS. The study did not collect information on the time from HIV diagnosis or KS diagnosis to study entry. However, at baseline 99% of participants were antiretroviral naïve, 73% had >25 cutaneous KS lesions, 71% had raised cutaneous lesions, 61% were ACTG Stage T1, 58% had tumor associated edema, and 55% had oral KS. The baseline characteristics of our study population do not suggest that many participants had long-standing indolent KS.

The proportion of men and women enrolled in our study is similar to what prior studies of AIDS-KS epidemiology in Africa have reported, with a roughly 2 to 1 ratio of men to women¹¹. In our prior analysis, women had decreased odds of week 48 clinical failure⁵. The few other studies that investigated the effects of sex on AIDS-KS treatment outcomes have suggested that women have worse treatment outcomes. Two retrospective cohorts found increased mortality or less clinical improvement among women^12,13 and in a prospective clinical trial, female sex was an independent risk factor for death¹⁴. Despite small numbers of women in our study, we detected a significantly decreased hazard for early KS-PD and a trend toward decreased hazards of KS-IRIS for women compared to men, a finding that differs from the prior studies mentioned above. Women have higher adherence to antiretroviral therapy¹⁵, so it is possible that there were also sex differences in adherence to oral chemotherapy. In general, women have less mortality from cancer than men^16,17 and a variety of mechanisms have been proposed to explain sex differences in cancer¹⁸. Women also have more vigorous immune responses to infections and vaccines, and are more likely to have systemic symptoms after certain vaccinations^19,20. Likewise, women with AIDS-KS are more likely to report systemic symptoms²¹. While the relationship of sex with early KS-PD and KS-IRIS appears paradoxical to the presumed immunologic basis of KS-IRIS, the definition of KS-IRIS used in our study did not require demonstration of inflammation and it is possible that early KS-PD that was classified as KS-IRIS may not have had an inflammatory basis. Measurement of immune responses in stored specimens is ongoing and will be reported in a future manuscript. We are hopeful that the results of these assays will inform our understanding of the biology of sex differences in early KS-PD and KS-IRIS.

Clinical confirmation of KS-IRIS is difficult in low-resource settings where there is limited access to specialized imaging and histopathology. A prior study proposed a revised general definition of IRIS based on agreement of two case definitions²². However, few cases of KS-IRIS were included in that analysis so it is unclear if the proposed revised IRIS definition is applicable for KS-IRIS. The definition of suspected KS-IRIS used in A5264/AMC067 was similar to definitions used in previous studies^2,23,24, and required suspicion for KS-IRIS by local clinicians who then initiated the CD4 count and plasma HIV RNA measurements only if KS-IRIS was suspected. Even though the protocol instructed investigators to consider KS-IRIS for all worsening of KS lesions within the first 12 weeks of ART, CD4+ count and HIV RNA were not performed at the time of progression in 44% of early KS-PD because the local clinicians did not suspect that tumor progression represented an IRIS event. Our finding that participants with suspected KS-IRIS were more likely to have raised cutaneous KS lesions compared to non-IRIS early KS-PD (89% and 64%, respectively) suggests that the appearance of cutaneous KS lesions was an important factor in the decision to initiate KS-IRIS evaluations. All participants with early KS-PD who were evaluated for KS-IRIS met either the CD4+ cell or HIV viral load criteria. Likewise, all participants with early KS-PD who were not evaluated for KS-IRIS met either the CD4+ cell or HIV viral load criteria by the scheduled laboratory tests at week 12. Thus, it is likely that KS-IRIS was underdiagnosed and the cumulative incidences of suspected KS-IRIS in our study, while within the 6% to 29% range reported by other studies^3,24,25, should be considered minimum estimates.

A5264/AMC067 is the first prospectively randomized clinical trial to show the ability of concomitant oral chemotherapy and ART to prevent early KS-PD and KS-IRIS compared to ART alone. A prior randomized trial found that the addition of chemotherapy with intravenous bleomycin/vincristine to ART produced higher 12-month KS response rates compared to ART alone, but bleomycin/vincristine did not affect rates of KS-IRIS¹⁴. Another randomized clinical trial, A5263/AMC066, which was conducted in parallel to A5264/AMC067 at many of the same sites included in our trial, and which used the same criteria for KS-IRIS, showed that suspected KS-IRIS occurred infrequently during treatment of advanced, symptomatic KS with ART in combination with each of three chemotherapy regimens tested (10% in the etoposide, 2% in the bleomycin/vincristine and 0% in the paclitaxel arms²⁶. A5263/AMC066 also found that paclitaxel was superior to both etoposide and bleomycin/vincristine for the endpoint of progression-free survival. Taken together, the results of A5264/AMC067 and A5263/AMC066 suggest that the frequency of KS-IRIS is inversely related to chemotherapy potency and that early initiation of potent chemotherapy together with ART effectively prevents early KS-PD and KS-IRIS.

A5264/AMC067 has several limitations to consider. Our cohort of HIV-infected persons with mild-to-moderate KS disease, excluded many people with T1 stage KS disease who could have higher risks of KS-IRIS. The open label design could have introduced bias in clinical assessment of KS-IRIS. As noted above, clinical confirmation of KS-IRIS is difficult in the low-resource settings where this study was conducted and a large number of early KS-PD cases were not assessed for IRIS by the local site investigators, so under-diagnosis of KS-IRIS could have influenced our results regarding KS-IRIS.

Decisions about whether to initiate chemotherapy together with ART for AIDS-KS must consider both the potential benefits and the added toxicities of chemotherapy plus ART over ART alone. Current United States Department of Health and Human Services guidelines recommend that chemotherapy, in combination with ART, should be administered to patients with KS visceral involvement²⁷. WHO guidelines recommend immediate ART initiation in combination with systemic chemotherapy for persons with severe or symptomatic KS⁶. Our finding that initiating chemotherapy and ART together effectively decreased early KS-PD, including KS-IRIS, provides additional support for these recommendations.